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 6 Greenhouse Gas Emissions and Broad-Based Tax Expenditures INTRODUCTION Chapters 3 through 5 analyzed the effects targeted tax provisions have on greenhouse gases (GHGs). The current chapter examines broad-based tax ex- penditures. While the narrow provisions may cause significant impacts because they affect the energy sector, which has relatively high GHG emissions per unit of output, the broad-based provisions may be important because of their size and importance for the overall economy. Broad-based provisions of the tax code are likely to have two kinds of im- pacts on GHG emissions. First, they may change the composition of national output from high (low) GHG-intensive sectors to low (high) GHG-intensive sectors. Second, they may increase or reduce the overall size of the economy and therefore change emissions simply because the economy is larger or smaller. Although there are many large tax expenditures, here as in other areas the committee found it necessary to limit the scope of its inquiry. After examining the largest tax expenditures, the committee decided to examine three broad types of provisions: (1) tax incentives that affect investment generally, (2) tax incen- tives for owner-occupied housing, and (3) tax incentives for the provision of health insurance. Among tax incentives for investment, the one that results in the largest revenue loss is the provision allowing firms to claim accelerated depreciation for investment in machinery and equipment under the Modified Accelerated Cost Recovery System (MACRS).1 Tax incentives for housing include the mortgage 1 Accelerated depreciation for machinery and equipment was the largest business tax expenditure in the lists compiled by the Office of Management and Budget (OMB) and Joint Committee on Taxation (JCT) in 2012. In 2013, however, JCT shows the 5-year revenue loss from the provision as negative. This reflects the fact that special temporary bonus depreciation in rules in effect in 2011 and 2012 expired, so that companies will be 113
114 Effects of U.S. Tax Policy on Greenhouse Gas Emissions interest deduction, the second largest income tax expenditure in the tax code, as well as the deduction for property taxes on owner-occupied homes and the par- tial exemption of capital gains on the sale of a principal residence. The single largest tax expenditure, according to both Joint Committee on Taxation (JCT) and Treasury Department estimates, is excluding employer-provided health in- surance benefits from being taxed as income. Health insurance benefits are also excluded from wages used to determine payroll tax liability. We discuss each of these provisions in detail below.2 FINDINGS FROM PRIOR LITERATURE More than the other provisions considered by the committee, the literature provided very little in the way of prior research on the impact of the tax code on GHG emissions. For example, a few studies consider the effects of the tax codeâs subsidies on housing, while other studies consider the impact of the capi- tal gains exclusion on the rate of home sales. None of the studies the committee or consultants found consider the impacts of housing-related tax provisions on GHG emissions (Shan, 2011; Cunningham, 2008). Many studies estimate the effects of the subsidy provided by the mortgage interest deduction on federal receipts, the distribution of the tax burden, and housing prices, and its differen- tial effects across states and among metropolitan areas, but none address its po- tential effects on GHG emissions. Important studies of these provisions include Poterba 1992, 2011; Toder, 2010; Gale, 2007; Glaeser, 2003; Sinai, 2001; Mann, 2000; Brinner, 1995; Capozza, 1996; Voith, 1999; and Bruce, 2001. Research on the provisions for accelerated depreciation considers the im- pact on the relationship between capital and energy but makes no attempt to assess greenhouse gas impacts (Solow, 1987; Field, 1980; Metcalf, 2008). While there is a voluminous literature on the economic impacts of the tax expenditures for health care, the committeeâs review of the literature found no studies ad- dressing energy or greenhouse gas impacts of the tax subsidies to health care. receiving much smaller deductions over the next few years from investments they made in 2011 and 2012 than would normally be the case. The tax expenditure figures show cash flow revenue effects, not the incentive effects of provisions. 2 Some economists believe that the ideal tax system is a consumption tax, which ex- empts the return to saving, instead of an income tax. Under a consumption tax, the mort- gage interest deduction and the exclusion of employer-provided health insurance and health care would still be tax expenditures, but accelerated depreciation would not be. Furthermore, under a consumption tax, the normal rule would be to allow capital costs to be deducted in the year incurred, so that any depreciation system that spreads deductions over time would be viewed as a tax penalty. Such a system, however, would also have many other features that differ from the current law, including disallowing deductions for interest expenses. Given that the United States in fact has an income tax, the committee viewed the tax expenditure definitions used by the Treasury and the JCT as the appropri- ate ones to use for this study.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 115 MODELING APPROACH The committee undertook a literature review to determine whether the ex- isting literature provides estimates of the impact of broad-based tax expenditures on GHG emissions. The answer was negative. While there is an extensive litera- ture on the economic impacts of tax expenditures (see, for example, Poterba, 2011), no complete study has addressed the questions posed to the committee. Given the lack of existing work, the committee decided here, as in the oth- er areas, to undertake a study using existing models. After a review of possibili- ties, the committee concluded that the most promising model to simulate the effects of these broad-based tax policies on GHG emissions was the Intertemporal General Equilibrium Model (IGEM), maintained by Dale Jorgen- son Associates (DJA). IGEM is an integrated energy-economic model of the U.S. economy that includes the following features that are useful for this investigation: (1) a de- tailed representation of the U.S. tax structure, (2) a detailed breakdown by in- dustry sector of output in the United States economy, including representation of industries involved in the production and consumption of energy, and (3) a dis- aggregated representation of the household sector based on data for individual households. Because of its structure, IGEM can provide many useful insights about the effects of these policies on GHG emissions. Nonetheless, IGEM, as discussed below, also has certain limitations for analyzing some of the broader tax policies discussed in this chapter. A more complete description of the IGEM model and alternatives is con- tained in Chapter 2 and in Appendix A. One important feature leads to a major difference between IGEM and other models that have a detailed representation of the energy sector, such as the National Energy Modeling System (NEMS model). IGEM does not have an explicit representation of different types of cap- ital in the energy sector, or of the details of energy supply. In conjunction with the putty-putty nature of capital (see further discussion later in this chapter and in Appendix A), this may lead to quite different responses in the energy sector to price changes. For example, a tax policy might lead to an increase in the demand for electricity. In the NEMS model, this might be met at the outset from existing capacity, up to a point. But beyond that point, the NEMS model would recog- nize the long lead times of nuclear power, the tightening environmental re- strictions on coal, and the current advantageous position for these reasons of natural gas-fired generation. The IGEM assumptions, by contrast, would simply switch all generation to the cheapest power source almost immediately. Because fossil electrical generation has a much higher GHG intensity than other fuels and renewable sources, IGEM is likely to (and in some cases we review below, probably does) project some tax policies as having greater GHG impacts than other models would suggest. The ground rules for the modeling of broad-based tax expenditures were similar to those described in Chapter 2. The model calculated the impact of re- moving tax provisions on the economy and on GHG emissions. Because IGEM
116 Effects of U.S. Tax Policy on Greenhouse Gas Emissions is a general equilibrium model and explicitly represents federal government rev- enues and expenditures, it was necessary to specify the fiscal adjustments. The simulations were performed keeping the budget deficit unchanged on a year-by- year basis. This was done by returning any revenues using two alternative as- sumptions about how the revenue from eliminating tax expenditure provisions was recycled: first, as lump-sum payments to households, and second, as an equal proportional reduction in all marginal tax rates applied to individual and corporate income. The background data on GHG intensities will be useful in understanding the current structure of the economy. Table 6-1 shows IGEMâs estimated carbon dioxide (CO2) emission intensities of different sectors for the year 2010; these coefficients are based on input-output use tables. We show the five most CO2- intensive sectors and the five least CO2-intensive sectors. Additionally, the sec- tors considered in this study are shown in bold. An important consideration in thinking about tax policies is how very intensive a few sectors are compared with the rest. For example, electric utilities are approximately 20 times more CO2-intensive that the economy on average. This implies that small shifts in the composition of output can have large impacts on overall GHG intensity because of changes in the composition of output within the economy. TABLE 6-1 Energy Intensities of Different Sectors Intensities in CO2-e per unit real output Five most intensive Emissions intensities Electric utilities (services) 5.740 Petroleum refining 4.412 Gas utilities (services) 2.746 Primary metals 2.293 Coal mining 1.887 Personal and business services 0.108 Five least intensive Apparel and other textile products 0.055 Finance, insurance and real estate 0.034 Non-electrical machinery 0.027 Communications 0.018 Electrical machinery 0.015 Consumption 0.168 Investment 0.030 Government 0.106 Exports 0.335 Total 0.247 Note: Emissions in millions of metric tonnes CO2 equivalent (MMTCO2-e). Domestic industry output and GDP quantity indices in $1996 billions. Source: IGEM model docu- mentation. Source: IGEM model documentation.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 117 Tables 6-2 and 6-3 present the major results of the IGEM analyses. Table 6-2 shows the basic results of the IGEM calculations. The striking result is that the major impact, if there is one, comes through the output effect. For all the broad-based tax expenditure scenarios, GHG intensities change by less than 0.4 percent from the base case when the tax expenditure provision is removed. But in all cases except for accelerated depreciation, removing the tax preference and substituting lower tax rates raises total emissions, because the lower tax rates on labor and capital income raise national output. In the case of removal of the ac- celerated depreciation preference, national output also increases, because the capital stock, although smaller, is deployed more efficiently and lower tax rates raise saving and labor supply. However, the decline in total emissions per unit of output from lower capital investment reduces GHG emissions by more than the higher output increases them. In the following sections, we describe the results of the modeling in more detail along with an interpretation of the results. INVESTMENT INCENTIVES Investment incentives encourage capital formation and tilt the playing field towards business investment in tax-favored assets. Both the shift towards more capital-intensive production methods and the reallocation of output among industries can affect the level of GHG emissions, even for incentives that do not directly target energy production or conservation investments. This section dis- cusses how accelerated depreciation for machinery and equipmentâa long- standing preference in the federal income taxâmay affect both the total level of GHG emissions and emissions per dollar of national output.3 3 This report uses the term ânational outputâ as an indicator of overall income and out- put. In principle, it would include improvements in the allocation of resources that are difficult to capture in the national economic accounts. The committee prefers the net national product (NNP) as its measure of national output and income to gross domestic product (GDP). NNP is a better measure of a nationâs output than the standard measure. NNP differs from GDP in two respects. First, it subtracts depreciation of capital from the measures of investment and total national product. By so doing, it measures the net addi- tion to the nationâs capital stock (and therefore future living standards) instead of total spending on new investment (gross investment), including investment that simply replac- es capital that has worn out. Second, it counts as investment amounts invested by U.S. residents overseas as well as abroad, whereas GDP measures only domestic investment. This second correction measures net additions to wealth owned by U.S. residents instead of additions to capital in the United States, some of which will provide future income to foreign residents. IGEM assumes that the U.S. current and financial account balances of the balance of payments are unaffected by policy changes, so for IGEM the only differ- ence between changes in NNP and changes in GDP is changes in depreciation of capital. By using the NNP measure, we recognize that higher depreciation does not contribute to living standards.
118 Effects of U.S. Tax Policy on Greenhouse Gas Emissions TABLE 6-2 Effects of Different Revenue Recycling Options (% change from reference scenario, average over 2010-2025) EPHI: Mortgage Property Capital Tax on interest tax gains on Accel. Petroleum Services deduction deduction homes Depreciation Tax Tax rate cut case GDP 1.03 2.25 0.62 0.26 0.20 -0.16 Total GHG 0.65 2.35 0.68 0.22 -0.17 0.99 CO2 0.65 2.28 0.66 0.21 -0.11 1.28 GHG intensity -0.38 0.10 0.06 -0.04 -0.33 1.22 (per unit GDP) Lump sum rebate case GDP 0.45 -0.04 -0.09 0.21 -1.87 0.11 Total GHG -0.74 0.23 0.02 0.17 -2.06 1.24 CO2 -0.71 0.20 0.01 0.17 -1.98 1.53 GHG intensity -0.29 0.28 0.11 -0.04 -0.17 1.20 (per unit GDP) Source: Dale Jorgenson Associates, âEffects of Provisions in the Internal Revenue Code on Greenhouse Gas Emissions,â Report to the Board on Science, Technology and Eco- nomic Policy of the National Academies, June 6, 2012. TABLE 6-3 Estimated Effects Relative to Base Case of Eliminating Accelerated Depreciation Preference on Key Economic Variables, 2010-2035 With Cut-in Economic Variable With Lump-sum Rebate (%) Marginal Tax Rates (%) National Output (NNP) -1.65 0.38 Capital Stock -2.54 -0.50 Labor Input -0.81 1.01 Coal Mining -2.93 -0.42 Electric Utilities -2.66 -0.64 Total GHG Emissions -2.06 -0.17 CO2 Emissions -1.98 -0.11 GHG Emission Intensity -0.34 -0.49 Source: Dale Jorgenson Associates, âEffects of Provisions in the Internal Revenue Code on Greenhouse Gas Emissions,â Report to the Board on Science, Technology and Eco- nomic Policy of the National Academies, June 6, 2012. Description of Provisions The federal income tax includes a number of tax expenditures to encour- age private investment. The largest of these is the provision to allow accelerated depreciation of machinery and equipment. Under current law, the costs of in-
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 119 vestment in machinery and equipment can be recovered using asset lives and depreciation methods specified under the Modified Accelerated Cost Recovery System, a system that allows companies to deduct the cost of assets at rates fast- er than the rates at which their values actually decline. In addition, in certain years, companies have been allowed to claim bonus depreciation allowances, which allow them to deduct a portion of the cost in the year the investment is made, with the remaining amount then recovered using MACRS. In 2011, for example, companies could recover 100 percent of the cost for machinery and equipment immediately (this being expensing), while in 2012 the bonus depreci- ation rate was 50 percent. The Office of Management and Budget (OMB) estimated that accelerated depreciation of machinery and equipment will reduce federal revenues by $354.1 billion between fiscal years 2011 and 2015 and by $374.6 billion be- tween fiscal years 2013 and 2017.4 The Joint Committee on Taxation (JCT) es- timated the cost at $109.0 billion between fiscal years 2011 and 2015. OMB describes its estimate of the tax expenditure as the difference between revenue under existing tax provisions and a hypothetical alternative that allows business- es to claim depreciation over time in accordance with the decline of the econom- ic value of assets due to wear, tear, and obsolescence. JCT describes the tax ex- penditure as âdepreciation of equipment in excess of the alternative depreciation system.â Because the two agencies do not necessarily measure the tax expendi- ture against the same baseline law and may make different assumptions about levels and timing of investment in qualified equipment, they arrive at very dif- ferent estimates of the cost of the tax preference. Economic Effects Accelerated depreciation reduces the cost of investing in machinery and equipment and thereby raises the level of investment in the short run and the capital intensity of the economy in the long run. Higher investment raises the growth of real output in the short run and raises output per capita and real wages in the long run by increasing capital per worker. But the net economic benefit of accelerated depreciation depends on how it is financed and what other changes take place. Higher fiscal deficits could raise interest rates and crowd out investment, while paying for the tax benefit through raising other tax rates or cutting spending could have other potentially harmful effects, depending on how taxes are raised or what types of spending are reduced. There are a number of channels through which accelerated depreciation could affect greenhouse gas emissions: 4 The Office of Tax Analysis at the U.S. Treasury Department performs the estimates for OMB for inclusion in each yearâs federal budget documents.
120 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Positive effects on economic growth could raise GHG emissions, simp- ly because a larger economy, with all other things unchanged, will re- quire greater use of fossil fuels. A more capital-intensive economy could raise or lower GHG emissions per unit of output, depending on whether capital and energy are substi- tutes or complements in production. Shifts in output among industries (from those that use less to those that use more machinery and equipment) could increase or decrease fossil fuel use, thereby raising or lowering GHG emissions per unit of output. Accelerated replacement of old capital could reduce GHG emissions per unit of output if new capital is more energy efficient than the capital it replaces. The combination of these effects means that eliminating the accelerated depreciation preference could either increase or decrease GHG emissions. Addi- tionally, policy makers will also be interested in knowing the extent to which changes in GHG emissions reflect changes in the level of national output (all things the same, higher output will be associated with more GHG emissions); the extent to which they reflect changes in the composition of output; or the pos- sibility that changes affect the emissions intensity of different sectors. Assessment of Modeling Results IGEM represents the accelerated provisions in current law as based on the MACRS provisions that are permanently in the tax code, assuming that the tem- porary bonus depreciation provisions will expire as scheduled. The effects of eliminating MACRS are estimated by assuming that it would be replaced by capital cost recovery rules consistent with economic depreciation.5 This would raise the cost of capital in different industries by amounts that vary with the composition of assets (machinery and equipment, structures, and inventories) used in the industry and the degree of preference currently given to the equip- ment specific to the industry by MACRS. IGEM is well designed to capture these effects. Some features of IGEM may overstate the effects of reducing the cost of capital on national output and growth. One important feature is the assumed flexibility of the capital stock (this is called the putty-putty assumption). The 5 The report submitted by Dale Jorgenson Associates (2012) provides the formula they used for computing the cost of capital under economic depreciation, but does not specify how they derived the economic depreciation rates for the various industries. One source for calculating these parameters is the asset life categories used by the Bureau of Eco- nomic Analysis (BEA). The BEA asset lives can be combined with assumptions about the pattern of decline in asset values to compute estimates of the annual percentage deprecia- tion rate.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 121 structure of capital is assumed to respond immediately to changes in prices. This is particularly questionable for long-lived assets in the energy sector, such as nuclear power plants. A more realistic assumption, which is followed, for exam- ple, in the NEMS model, is that once capital is in place, its characteristics and particularly its input mix are fixed (this is called the putty-clay assumption). While the reality is somewhere in between these polar cases, it is clearly unreal- istic to assume that a coal-fired plant can overnight turn into a wind plant if coal prices rise sharply or if the regulatory environment changes quickly. IGEM assumes that agents (households and firms) have an infinite time horizon. This can be interpreted as assuming the current generation acts as if the well-being of their offspring is perfectly substitutable for their own well-being. Some other general equilibrium models, in contrast, are based on the life-cycle hypothesis that households allocate consumption over their lifetime to optimize their own economic well-being, given their lifetime resources (initial wealth and present value of lifetime earnings). The effect of the infinite horizon assumption is generally to increase the effect of investment tax preferences on current in- vestment. Taken with other assumptions in IGEM, the result is that after-tax interest rates are unaffected by tax and other policy changes, so that increased demand for tax-favored investments will not drive up interest rates and crowd out other investment. IGEM is a model with fixed current account and financial account flows in the balance of payments. This implies that none of the adjustments to saving and investment occur outside the United States. This will overestimate the impact of changes on domestic consumption and saving, which is assumed to respond completely to changes in investment demand. In a broader and more accurate model that allows for changes in international capital flows, changes in the de- mand for investment could also result in changes in international capital flows, as saving from abroad finances some of the new investment. Table 6-3 shows the estimated effects of replacing the current accelerated depreciation preference (MACRS) with economic depreciation. In the case where the increased revenues are used to provide lump-sum rebates to house- holds, national output would decline an average of 1.65 percent per year over the 26-year period 2010â2035 (Table 6-3). Because lump-sum rebates leave the tax structure unchanged, this simulation is the closest to isolating the effects of the tax preference alone. It indicates that accelerated depreciation provides a significant boost to national output, largely due to an increased capital stock. The capital stock will eventually be 2.5 percent smaller than in the base case scenario if the preference were removed. Labor input would also decline, as the reduced capital per worker lowers real wages (compared with the baseline) and therefore causes households to substitute leisure for labor (i.e., work less). Total GHG emissions and CO2 emissions both decline about 2 percent rel- ative to the baseline, reflecting to a large degree reduced investment and output in the coal-mining and electric-utility sectors. The decline in emissions is larger than the decline in national output, so the emissions intensity of production falls by about 0.34 percent. Put differently, if accelerated depreciation provisions are
122 Effects of U.S. Tax Policy on Greenhouse Gas Emissions removed, GDP declines 1.65 percent, emissions intensity falls by 0.34 percent, so emissions decline by 2.06 percent. The second column of Table 6-3 shows the impact when revenues are raised through tax-rate changes. When the tax revenue from eliminating acceler- ated depreciation is used to lower all marginal tax rates on individuals and cor- porations, national output increases by about 0.4 percent, relative to the baseline projection. In other words, the impact on national output is close to neutral if the capital tax-expenditure provisions are replaced by tax-rate reductions. The capi- tal stock still falls (by 0.50 percent), but the reduction in marginal tax rates in- duces a 1.01 percent increase in labor supply. The reduction in preferential treatment of those selected industries that benefit most from accelerated depreci- ation leads to a more efficient allocation of the capital stock. The impact of removing accelerated depreciation on total emissions of GHGs is essentially zero in the case when revenues are recycled through lower tax rates. While national output is higher by about 0.38 percent, the emissions intensity of the economy declines by 0.49 percent. The net effect on GHG emis- sions of -0.17 percent is essentially zero and probably not within the resolution of the model. Conclusions: Accelerated Depreciation Accelerated depreciation is one of the largest tax expenditures in the fed- eral income tax code (although, as indicated above, the cost of the preference is imprecisely estimated). According to the estimates prepared for the committee, its overall impact on national output is uncertain and depends upon the method by which the revenues are recycled. Eliminating accelerated depreciation would in both recycling cases reduce the capital stock. However, the effect on capital stock would be partially offset and labor supply and national output would in- crease if the additional revenue were used to finance cuts in individual and cor- porate marginal tax rates. The impact of removing accelerated depreciation on overall GHG emis- sions is probably negative, but the amount depends upon the fate of the reve- nues. If the revenues are returned by lowering tax rates, then the overall impact on GHGs is essentially zero. In contrast, if they are returned through lump-sum rebates, then GHGs are probably lower because the lower emissions intensity is combined with lower economic growth, and overall emissions are calculated to fall by about 2 percent. TAX INCENTIVES FOR OWNER-OCCUPIED HOUSING Federal tax incentives for owner-occupied housing are usually justified as encouraging more people to own homes. Homeownership is often held to have positive economic spillovers, such as better property maintenance, and higher levels of community involvement and voting participation. But the incentives
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 123 also encourage people to own larger houses, affect patterns of urban develop- ment, impact the rate of housing stock turnover, and reallocate capital from the business to the household sector. Each of these changes could affect greenhouse gas emissions, and possibly in opposing directions. Description of Provisions The federal income tax provides significant tax incentives for investments in owner-occupied housing. First, although investment income actually received is generally taxable under the U.S. federal income tax, implicit investment re- turns are generally not taxable. In the present case, the implicit income home- owners receive from housing services they provide to themselves is exempt from taxation. This implicit return is called âimputed rent on owner-occupied housingâ and represents the net rental income that owners would receive, after deducting costs of operation, maintenance, depreciation, and interest, if they had to pay rent to themselves as tenants. JCT and the Treasury Department original- ly did not include the exemption of net imputed rent as a tax expenditure provi- sion on the grounds that it would be impractical to require homeowners to place an implicit value on the rent they effectively pay to themselves and report that rent as income.6 In recent years, Treasury, but not JCT, has listed imputed rent as a tax expenditure line item. Treasury estimates that the exclusion of net rental income will cost the government $337.4 billion in lost revenues between fiscal years 2013 and 2017. The committee chose not to include the tax preference to imputed rent in our simulations, primarily because none of the major tax-reform proposals include this on the list and it is in practice difficult to define and en- force. Capital gains from the sale of owner-occupied homes also receive favora- ble tax treatment, compared with how the tax law generally treats capital gains. Homeowners may exclude from gross income up to $250,000 ($500,000 for a married couple filing a joint return) of the gain from the sale of a principal resi- dence if the taxpayer has owned and used the property for at least 2 of the 5 years preceding the date of sale. OMB scores the cost of the capital gains exclu- sion as $171.1 billion during fiscal years 2013 and 2017 and $121.1 billion dur- ing 2011 and 2015. JCT estimates the cost at $123.2 billion during fiscal years 2011 and 2015. In general, business firms and investors can deduct from their taxable in- come costs of investment, such as interest paid and state and local property tax- es. But because owner-occupied homes do not generate taxable income (due to the exclusion of imputed rent), deductions for housing costs are regarded as tax 6 Although not regarded as part of their income by most people, imputed rent is count- ed as income in the U.S. National Income and Product Accounts on the grounds that the measured capital return generated by the nationâs housing stock should not change when people switch from being tenants to owners.
124 Effects of U.S. Tax Policy on Greenhouse Gas Emissions preferences.7 OMB estimates that the deduction of mortgage interest on owner- occupied housing will cost $606.4 billion during fiscal years 2013 and 2017, making this deduction the second largest tax expenditure in the federal income tax. Between fiscal years 2011 and 2015, OMB scores the cost of this provision at $491.1 billion, while JCT scores its cost at $464.1 billion. OMB estimates that the deduction of state and local property taxes on owner-occupied homes will cost $140.6 billion between fiscal years 2013 and 2017 and $118.6 billion be- tween 2011 and 2015. In comparison, JCT estimates the cost at $117.1 billion between 2011 and 2015. Economic Effects Most taxpayers receive little or no benefit from the tax preferences for owner-occupied housing other than the exclusion of imputed rent. Taxpayers who do not itemize deductions on their tax returns receive no benefit from the deductibility of mortgage interest or state and local property taxes.8 Taxpayers with modest incomes who are in the 15 percent rate bracket receive a relatively small subsidy, compared with the benefit received by higher-income taxpayers in the 28 percent or 35 percent rate bracket. JCT reports that the 18 percent of tax returns with incomes of $100,000 and over received 78 percent of the benefits from the mortgage interest deduc- tion and 73 percent of the benefits from the real estate tax deduction. Among beneficiaries of the mortgage interest, 55 percent had incomes of $100,000 or more, while 52 percent of those claiming deduction for residential property taxes had incomes of $100,000 or more. Because the mortgage interest and property tax deductions largely benefit those with high incomes who are likely to have very high home ownership rates without a tax preference and does little or nothing for lower income families, it probably does little to increase the rate of homeownership.9 It does, however, substantially reduce the costs of housing capital. This could lead either to the construction of larger and more expensive homes, or to the bidding up of prices in areas, such as densely populated urban centers, where there is little space for housing expansion. The JCT does not report the distributional effects of the tax preference for housing capital gains, but it is probably also highly concentrated at the top end of the income distribution, even though it is capped for taxpayers with very large 7 Interest incurred to finance the acquisition of other non-income-producing household assets, such as cars, TV sets, and furniture, is not deductible. 8 JCT reports that, in 2010, only 29 percent of taxpayers claimed itemized deductions. But 72 percent of returns with incomes between $100,000 and $200,000 and 94 percent of returns with incomes of $200,000 and over were itemizers. 9 See, for example, William G. Gale, Jonathan Gruber, and Seth Stephens-Davidowitz, âEncouraging Homeownership Through the Tax Code,â Tax Notes, June 16, 2007.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 125 housing gains. Taxpayers in the 15 percent bracket or belowâabout 75 percent of all returnsâcurrently face a zero tax rate on capital gains, so the additional exclusion for housing gains does not benefit them. Instead, it benefits higher- bracket taxpayers who would otherwise face rates ranging from 15 to 23.8 per- cent on the first $250,000 ($500,000 if married) of gains from the sale of their house. The overall consequence is that all the housing incentives probably have little effect on the decision whether to rent or to own a home, but they very like- ly do affect the allocation of capital between housing and other assets. By so doing, they probably reduce the productivity of the capital stock and lower na- tional output relative to a more neutral tax system that treats different capital assets more similarly. Effects on Greenhouse Gas Emissions The mortgage interest deduction encourages a shift towards additional in- vestment in owner-occupied housing at the expense of business-sector invest- ment. This reallocation of investment reduces the overall efficiency of the capi- tal stock, thereby lowering national output and reducing GHG emissions associated with higher production. Beyond this, for any given level of national output, the reallocation of cap- ital from business to housing capital has several effects on GHG emissions, in- cluding: The shift from other assets to housing could raise or lower GHG emis- sions if other industries generate less or more emissions per unit of out- put than housing. In particular, to the extent a shift from business assets is associated with a reduction in production of fossil fuels and in the generation of electric power, it could lower GHG emissions per unit of national output. The increase in the housing stock could raise GHG emissions per unit of national output if larger houses are associated with higher consump- tion of fossil fuels than the output that is replaced. The tax incentives for housing could also raise GHG emissions per unit of national output, if they contribute to reduced density of residences, raising commuting times and distances and leading to increased auto- mobile use. Thus, while the tax incentives for housing do not affect fossil fuel consump- tion and GHG emissions directly, they could have substantial indirect effects through changes in the composition of output between industries and changes in the composition and geographic dispersion of the housing stock.
126 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Assessment of Modeling Results IGEM is able to capture the effects on national output from provisions that favor residential capital over business-sector capital. It also captures the effects on national output and its composition from offsetting cuts in marginal tax rates that may accompany any reduction in housing tax preference as well as the ef- fects of shifts in the composition of output that result from differences in energy intensity among industries (including the household sector). Unfortunately, IGEM omits two potentially important effects of tax incen- tives for housing on GHG emissions: (1) the effects of changes in housing capi- tal and spending on household energy use, particularly on the energy used in heating and cooling; and (2) the effects of changes in housing capital and spend- ing on transportation that results from changing the pattern of dwelling density and location. Because of these omissions, the committee believes that the IGEM results do not capture completely the impact of the housing provisions on GHG emissions. IGEM does not capture the effects of changes in the size and composition of housing units because IGEM represents housing as capital that uses energy in construction (including replacement investment for the capital that depreciates) and maintenance, but not in its operation. This means that the portion of annual output that is used to maintain and improve the stock of housing capital gener- ates GHG emissions, but the portion of annual output that represents the service flow from the existing asset does not. To capture the full effects of increases in the housing stock on energy use, a model would have to represent the flow of annual consumption from the housing stock as a service that is produced with inputs from other sectors (electric power for air conditioning and lighting as well as oil, gas, or electricity for space heating). That is, a model would need to cap- ture the complementarity between housing space and fuel consumption in the form of electricity use (for cooling and other uses) and direct fuel consumption (for heating), instead of simply representing housing space as a final consump- tion good that substitutes for other goods that use electric power generation and fossil fuels as inputs in production. IGEM also does not capture any effects of the tax incentives on the spatial allocation of the housing stock. Therefore, it cannot capture any possible linkag- es between changes in the size composition of housing and changes in commut- ing patterns, the transportation of goods, and the like. A final concern is that IGEM has low resolution of the housing sector. Housing is included in a broad sector, âFinance, insurance and real estate.â Moreover, there is no distinction between owner-occupied and rental housing. The low resolution makes interpretation of the results particularly difficult.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 127 Modeling Results Table 6-4 shows the estimated impacts of changes in housing tax prefer- ences according to IGEM. The simulation finds that eliminating the mortgage interest and property tax deductions would have almost no effect on national output if the tax revenue is recycled through lump-sum transfers to households, for example, through tax rebates or credits. The efficiency loss from capital mis- allocation totally offsets any output increase from lower taxes on capital. Na- tional output would increase slightly in spite of declines in both labor input and investment. By contrast, using the revenues from eliminating the deductions to lower marginal tax rates on individual and corporate income would lead to substantial increases in the capital stock, labor input, and national output. The efficiency gain from an improved allocation of the capital stock causes national output to increase by about twice the increase in productive inputs. The simulations also show that eliminating the tax subsidy on mortgage interest would raise GHG emissions per unit of output. For both recycling cases, removing the subsidy increases the GHG intensity of national output. This re- flects compositional changes in the economy, as elimination of the subsidy in- creases the output of heavy energy-using sectors (coal mining and electric utili- ties) at the expense of sectors not using energy (services from housing capital). Overall GHG emissions are estimated to increase in both recycling cases, pri- marily because of the increase in national output. TABLE 6-4 Estimated Effects of Eliminating the Home Mortgage Interest Deduction Relative to Reference Scenario, 2010-2035 With Lump-sum With Cut-in Marginal Economic Variable Rebate (%) Tax Rates (%) National Output (NNP) +0.17 +2.42 Capital Stock -1.12 +1.13 Labor Input -0.77 +1.22 Coal Mining -0.11 +2.70 Electric Utilities -0.55 +2.81 Total GHG Emissions +0.23 +2.35 CO2 Emissions +0.20 +2.28 GHG Emission Intensity +0.28 +0.10 Source: Dale Jorgenson Associates, âEffects of Provisions in the Internal Revenue Code on Greenhouse Gas Emissions,â Report to the Board on Science, Technology and Eco- nomic Policy of the National Academies, June 6, 2012.
128 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Overall, it appears that because of omission of major complements to housing, the modeling results do not fully capture the impact of changing hous- ing subsidies on GHG emissions. Conclusions: Housing Subsidies According to IGEM estimates, eliminating the tax subsidies for owner- occupied housing and using the revenue to lower marginal tax rates would im- prove the efficiency of allocation of the capital stock and increase national out- put. According to the modeling results, the impact on overall GHG emissions would be determined primarily by the overall economic reaction: If the provi- sion increases GDP, then GHG emissions would change at about the same rate. However, the simulation does not accurately capture the full effects of the housing subsidy on emissions intensity. The simulation fails to capture the ef- fects of housing size on household consumption of electricity, natural gas, and home heating oil and the potential effects of changing residential patterns on automobile use and gasoline consumption. Moreover, the resolution of the hous- ing industry is low, with no distinction between rental and owner-occupied dwellings. On the whole, the committee finds that the simulation results do not present a complete picture. The committee therefore concludes that the existing literature, as well as the results commissioned for the present study, is inconclu- sive regarding the impacts of eliminating housing tax incentives on GHG emis- sions. Understanding the full impacts remains an important topic for future re- search. TAX SUBSIDIES FOR HEALTH INSURANCE AND HEALTH CARE As is the case for other broad-based provisions, tax subsidies for health care do not directly affect energy use. But removing them could lower or raise GHG emissions, depending upon the impact on national output and on whether the composition of the economy shifts to more or less GHG-intensive activities. Description of 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. OMB estimates that exclusion of EPHI benefits will re- duce income tax revenue by $1,012 billion between fiscal years 2013 and 2017 and by $904.6 billion between fiscal years 2011 and 2015. JCT estimates the
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 129 exclusion will reduce income tax receipts by $725 billion during fiscal years 2011 and 2015.10 The EPHI exclusion is treated as a tax-expenditure provision because a comprehensive definition of income holds that all compensation, whether in the form of cash wages or fringe benefits, is counted as taxable income. Generally, fringe benefits (such as the annual rental value of housing or automobiles sup- plied by employers) are taxable to employees, but there are a number of statuto- ry exemptions. The most important of these exemptions is for health insurance and health benefits. There are other tax preferences for health that IGEM could not simulate. These include the deduction of health insurance premiums for the self- employed, the itemized deduction for medical expenses in excess of 7.5 percent of adjusted gross income (10 percent under the alternative minimum tax, and, beginning in 2013 under the regular income tax as a provision under the Afford- able Care Act), the deductibility of contributions to Medical Savings Accounts, and the exemption of investment income accrued within those accounts. Economic Effects The exemption of EPHI premiums provides an incentive for more em- ployers to provide health insurance coverage to their employees, and for those employers who do provide coverage to provide more generous plans, with lower co-payments and deductibles and more types of health services covered. More generous insurance coverage, in turn, encourages employees to spend more on health care services. The exclusion may also reduce the marginal effective tax rate (METR) on earnings.11 Including previously exempt fringe benefits in taxable income, without any compensating reduction in other taxes, raises the average effective tax rate on earnings. If the base broadening is accompanied, however, by a revenue- 10 EPHI benefits are also excluded from the wage base for payroll taxes used to fund Social Security retirement and disability benefits and Medicare Hospital Insurance bene- fits. OMB estimates that the exclusion will reduce payroll tax receipts by $619.2 billion during fiscal years 2013 and 2017, over 60 percent of the loss in income tax receipts. Some of the increase in the federal deficit is offset, however, by reduced Social Security retirement and disability benefits accrued by workers who will be credited with lower earnings used for benefit determination. These implications of EPHI benefits in the social insurance programs are excluded from the calculations for this report. 11 The effect here is complicated and can be seen as follows. The average tax-effective tax rate on earnings can be expressed by the equation, te = (tsW + tsaF)/(W+F), where te = the effective tax rate on earnings, ts = the average statutory tax rate, W = cash wages, F = the cash value of fringe benefits and a = the share of fringe benefits included in the tax base. If fringe benefits are excluded from the tax base (a = 0), the effective tax rate is simply equal to te = tsW/(W+F). If fringe benefits are included in taxable compensation (a = 1), then the effective tax rate is equal to the statutory rate ts.
130 Effects of U.S. Tax Policy on Greenhouse Gas Emissions neutral reduction in the statutory tax rate on earnings, then the effective tax rate on earnings will be unchanged. Ifâas with the present analysisâbroadening the tax base is accompanied instead by a reduction in statutory marginal tax rates that is applied to both labor and capital income, the effective tax rate on earnings will rise and the effective tax rate on capital income will decline.12 The operation of the EPHI is particularly complicated, because the effect on tax rates differs for different workers. The average tax rate on earnings can affect the decision whether or not to work for pay (the extensive margin). But for those who are working, the decision to work additional hours (the intensive margin) depends on the marginal effective tax rate, that is, the additional tax paid per dollar of additional compensation. Calculating how taxing EPHI benefits affects the METR on labor compen- sation requires an assumption about how EPHI benefits vary with additional hours of work. For many workers, the EPHI benefits are fixed or lump-sum â that is, their value is independent of wages and salaries. If EPHI benefits in any job are fixed as hours of work change, then taxing EPHI benefits raises the aver- age effective tax rate on compensation, but leaves the METR unaffected. If in contrast, the value of EPHI benefits rises in proportion to the increase in wages, then eliminating the tax provisions that exclude EPHI from income tax increases the METR by the same amount as an equal revenue increase in the marginal tax rate on wages. Generally, therefore, there is no simple way to characterize how taxing EPHI benefits changes the METR on labor compensation and the consequent incentive to work. At the extensive margin (deciding whether to work), it would reduce the incentive to work in the marketplace instead of working at home. Similarly, it may affect the decision to work full-time instead of part-time if, as is often the case, the availability of health benefits is restricted to employees who exceed some threshold level of hours worked. For workers considering small changes in hours worked (such as manufacturing workers taking on over- time), taxing EPHI benefits may have no effect on the incentive to work. These considerations imply that the taxation of EPHI benefits would potentially be important primarily for people who are on the workâno-work margin, and less important for prime-age workers with strong labor-force attachments. This sug- gests that the work incentives of EPHI provisions will be particularly important for secondary earners, for people near retirement age, and for workers who are potentially eligible for disability coverage. 12 Strictly speaking, the effective tax rate on earnings may also be affected by the tax rate on capital income, which affects how much future consumption a worker can pur- chase with her current earnings. The assumption in this discussion is that IGEM does not incorporate this direct effect of capital income taxation on the incentive to work, but in- stead looks only at the effective tax rate on wages.
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 131 Assessment of Modeling Results The committee asked for two separate pairs of simulations of the health care exclusion. For the first pair of simulations, DJA modeled the exclusion as a partial exemption from taxation of labor compensation (we call this âEPHI ex- clusionâ). This set of simulations did not capture the effect of the exemption on the price of health care services and therefore did not capture any of the effects of shifting the composition of production between health care and other sec- tors.13 Consequently, in order to include sectoral shifts, the committee requested a second pair of simulations in which the elimination of the health care exclu- sion was modeled as if it were equivalent to imposing a new excise tax on the consumption of health services (we call this âEPHI servicesâ). For each pair of simulations, as with the simulations of other incentives, the additional revenue from removal of the tax benefit was recycled in two different waysâas a lump- sum transfer to all households and as reductions in marginal tax rates propor- tionately on individual and corporate income. We requested that IGEM simulate the elimination of the exclusion of EPHI from income tax as if it were a new excise tax because the reference sce- nario contains no tax preference for EPHI services. Imposing an excise tax is equivalent to removing a tax-expenditure subsidy from the sector. We note that the health sector, like the housing sector, is poorly resolved in IGEM. IGEM does not include health services as a separate sector but as part of a larger sector called âpersonal and business services.â Consequently, the excise tax was im- posed on the personal- and business-services sector in which health is contained. Health is a substantial part of this sector, and its GHG intensity is similar, so we might expect that this treatment is a reasonable approximation to a more detailed treatment. Table 6-5 shows the results of the calculation. The change in tax treatment reduces national output (relative to the reference scenario) by 0.5 percent over the 26-year period (2010-2035) when the revenues are used to provide a lump- sum subsidy to households. In this case there are no efficiency gains to offset the efficiency costs of the new tax on EPHI services.14 13 The EHPI exclusion has no effect on the marginal effective tax rate on labor com- pensation in IGEM and therefore no effect on the incentive to work. As a result, when EPHI benefits are made taxable and the revenue is used to provide lump-sum transfers to households, there is virtually no effect on the key variables that characterize the econo- my. But when the revenue from taxing EPHI benefits is used to lower marginal tax rates on individual and corporate income, output, capital stock, and labor supply all increase by more than 2 percent relative to baseline projections. 14 This is likely to be an anomalous result because IGEM does not start from the point where the service sector is subsidized (due to health care exclusion) so that the new ex-
132 Effects of U.S. Tax Policy on Greenhouse Gas Emissions By contrast, when the revenues from the excise tax are used to reduce marginal tax rates on individuals and corporations, the capital stock rises by slightly over 1 percent, national output rises by about 1 percent, and labor sup- ply rises by slightly under 1 percent. One of the surprising results of the simulations is that they show that GHG intensity (GHG emissions per unit of national output) declines in both simula- tions. With the lump-sum substitution, GHG emissions decline by more than the drop in national output, while with the offsetting cut in marginal tax rates, GHG emissions increase by less than the increase in output. This result on GHG intensity appears anomalous because the business- and personal-services sector is less GHG intensive than the economy as a whole. As shown in Table 6-1, the GHG intensity of the sector is about 40 percent of the economy-wide GHG intensity.15 We would therefore assume that reducing the size of the health sector would increase the size of the rest of the economy and thereby raise overall GHG intensity. This is not the case with the IGEM simulations. TABLE 6-5 Estimated Effects Relative to Reference Scenario of Eliminating the Exclusion for Employer-supplied Health Insurance, 2010-2035 With Lump-sum With Cut-in Economic Variable Rebate (%) Marginal Tax Rates (%) National Output (NNP) -0.49 +0.98 Capital Stock -0.26 +1.14 Labor Input -0.51 +0.84 Coal Mining -0.78 +1.03 Electric Utilities -0.84 +0.61 Total GHG Emissions -0.74 +0.65 CO2 Emissions -0.71 +0.65 GHG Emission Intensity -0.36 -0.29 Source: Dale Jorgenson Associates, âEffects of Provisions in the Internal Revenue Code on Greenhouse Gas Emissions,â Report to the Board on Science, Technology and Eco- nomic Policy of the National Academies, June 6, 2012. cise tax may reduce a distortion instead of producing a new one. But it is perfectly con- sistent with the effects of imposing a new excise tax on an unsubsidized sector. 15 In addition to the data shown in Table 6-1, the same results are found in the Depart- ment of Commerceâs estimates of CO2 intensity looking at the direct and indirect impacts using input-output techniques. See Department of Commerce (2009).
Greenhouse Gas Emissions and Broad-Based Tax Expenditures 133 The committee discussed the result with the IGEM modelers and exam- ined the underlying results for individual industries in IGEM to determine the source of the decline in GHG intensity. It appears that the decline in the output of services is accompanied by a decline in the production of the petroleum and utilities sectors. In other words, there is a decline in the share of the most energy intensive sectors with the tax on services. Moreover, because the GHG intensity of energy-intensive sectors is so high, even a small decrease in their share can produce a decrease in overall GHG intensity. The committee was unable to de- termine what economic mechanism was behind this shift, however, and the shift is not completely persuasive from an economic point of view. The committee concluded that this is likely to be a statistical anomaly and one that is not within the ability of IGEM to resolve accurately. Conclusions: Health Subsidies Simulations for the committee found that eliminating the EPHI exemption would raise national output if offset by cuts in marginal tax rates. The estimates probably underestimate the distortionary effect of the health subsidies, so the impact on national output would probably be larger if the model could provide a more detailed representation. The impact in IGEM of removing the tax preferences on health on GHG intensity is anomalous. Economic intuition would suggest that eliminating health care subsidies would raise GHG emissions per unit of output because the health care sector is less energy-intensive than the rest of the economy (see Table 6-1). The modeling results from IGEM show the opposite effect, however, with a small decrease in GHG intensity. An examination of the simulations indicates that the results are driven by sectoral shifts that are not entirely plausible. The committee concludes that the model cannot resolve the impact of the health sub- sidies with sufficient precision are the impact on GHG intensity of removing the tax benefit for employer-sponsored health insurance is still an open question. OVERALL CONCLUSIONS This chapter examines the impact of broad-based tax expenditures on GHG emissions. The provisions examined were tax incentives that affect in- vestment, tax incentives for owner-occupied housing, and tax incentives for pro- vision of health insurance. The committee investigated the impacts of removing these provisions through runs commissions by a single model, IGEM. The major finding for all three sets of provisions is that the impact on GHG emissions is primarily driven by the impact on the growth of national out- put. In most cases, the change in GHG emissions is close to the change in na- tional output growth induced by the change in the tax provision, and there is little change in the emissions intensity (the emissions-output ratio). The primary result is therefore intuitively sensible. If the emissions-output ratio does not
134 Effects of U.S. Tax Policy on Greenhouse Gas Emissions change markedly, then the change in emissions will be determined by the impact on national output. A second finding is that the effects on national output and GHG emissions can be significantly influenced by the way the revenues are recycled. The com- mittee examined two revenue recycling mechanisms, lump-sum rebates and tax rate reductions. The simulations find that the impact of each of the broad-based tax changes on output and GHG emissions could be up to 2 percent if the reve- nues are recycled through tax rate cuts that increase efficiency. They would be small (i.e., close to zero) if the revenues are recycled in a lump-sum fashion. The third and parallel finding is that the impact of the broad-based provi- sions on emissions intensities is generally small. The impacts range from a -0.3 percent to a +0.2 percent change in emissions intensities over the period 2010- 2035. The committee could not undertake a model comparison to test these find- ings, and no statistical tests of significance for the results were made with IGEM. Because of these limitations, the committee finds that no reliable esti- mates of the impacts of the provisions on emissions intensities can be deter- mined on the basis of existing evidence, but the effect on emissions intensities is likely to be small. The summary result, therefore, is that changes in broad-based tax provisions are likely to have a small impact on overall GHG emissions out- side of their impact on overall economic growth. Finally, the impact of the broad-based provisions is extremely small rela- tive to emissions growth over the period of investigation. IGEM projects a 43 percent change in emissions over the period in the reference scenario simulation. Even the tax preference with the largest impact would change the growth of emissions over the analysis period by at most a small fraction of the total emis- sions growth.