TPC’s Methodology for “Off-Model” Revenue Estimates
The Urban-Brookings Tax Policy Center (TPC) uses its large-scale microsimulation model to produce “on-model” revenue estimates of nearly all proposals affecting the individual income taxes, the Social Security and Medicare payroll taxes, certain Affordable Care Act (ACA) excises, and the estate tax. TPC makes “off-model” revenue estimates for proposals affecting taxes not on the model (corporate income, most excises, broad-based consumption taxes), as well as certain proposals affecting individual income, payroll, and estate taxes that cannot be estimated on-model. Although TPC’s on- and off-model revenue estimating methodologies differ in specific detail, they share several key characteristics:
- Both methodologies follow the estimating conventions of the official government estimating agencies, such as the Joint Committee on Taxation (JCT).
- Both are based on the Congressional Budget Office (CBO) forecast of economic and demographic variables (GDP, inflation, interest rates, population, etc.).
- Both use the same definition of “current law” used by the JCT and the CBO, and the estimates are benchmarked to CBO’s forecast of revenues under current law.
- In both, we take into account (to the extent possible) behavioral responses of taxpayers.
These shared characteristics are intended to make both the on- and off-model revenue estimates produced by TPC match as closely as possible the estimates that would be produced by the JCT, the official scorekeeper of any legislation.
TPC’s microsimulation model and the methodology for making on-model revenue estimates are described elsewhere.1 TPC’s methodology for estimating the revenue effects of broad-based consumption taxes is also described elsewhere.2 This paper describes how TPC produces most other off-model revenue estimates. The next section describes the CBO forecasts of macroeconomic variables and the revenue baseline that TPC uses in its estimates. The following sections describe TPC’s methodology for estimating the revenue effects of certain proposals affecting individual income and payroll taxes that cannot be estimated on-model, and various proposals affecting the corporate income tax and most excise taxes which are all estimated off-model.
Macroeconomic Variables and Baseline Revenues
The macroeconomic variables underlying TPC’s revenue estimates are from the January 2015 CBO forecast3 for FY2015-FY2025, and from the June 2015 CBO long-term forecast4 for FY2026 and later years. These variables include nominal GDP, real GDP, the percentage change in the CPI, the nominal 10-year Treasury bond yield, and the population. Revenues (under current law) by major source for FY2015-FY2025 are also from the January 2015 CBO forecast. The major revenue sources that CBO projects are: individual income taxes, social insurance taxes (with detail for Social Security, Medicare, UI, Railroad Retirement, and other retirement), corporate income taxes, excise taxes (with detail for taxes allocated to the highway trust fund, and taxes on tobacco, aviation, alcoholic beverages, health insurance providers and other goods and services), estate and gift taxes, Federal Reserve receipts, customs duties, and other miscellaneous receipts (with detail for Universal Service Fund and other fees and fines). In addition, CBO provides some information on the excises, fines and fees enacted as part of the Patient Protection and Affordable Care Act (PPACA).
The CBO baseline does not project specific revenue sources (except, indirectly, individual income taxes) after FY2025. For FY2026 and later years, TPC projects revenues from each source in the following way:
- Individual income taxes. The ratios of individual income tax receipts to GDP from the July 2014 CBO long-term forecast5 are multiplied by the GDP forecast to generate the forecast of individual income tax revenues.
- Social insurance taxes (with detail for Social Security, Medicare, UI, Railroad Retirement, and other retirement). The projections for each source starting in FY2026 are based on the trend in the log of the ratio of the source to GDP over the FY2019-FY2025 period.
- Corporate income taxes. We use the CBO assumption that these revenues are fixed as a percentage of GDP at the level projected for FY2025.
- Excise taxes (with detail for highway trust fund taxes and taxes on tobacco, aviation, alcohol, health insurers and other goods and services). All excises are projected by extrapolating the trends in receipts for 2014-2025. Information provided on the ACA excises in the January 2015 CBO forecast is used to separate the “other” excises into ACA-related and a residual (“non-ACA other”) for all years.
- Estate and gift taxes. The projection for FY2026 and later years is based on the trend in the ratio of estate and gift taxes to GDP over the FY2015-FY2025 period.
- Other revenues.6 The projections for FY2026 and later years are based on the trends in the ratio of these revenues to GDP over earlier periods.
Estimates Made Off-Model
We discuss below TPC’s methodology for producing off-model estimates for each tax source: individual income, payroll, corporate income, and excise taxes.
Individual Income Tax
Tax expenditures. Some tax expenditures in the individual income tax cannot be estimated on-model, generally because no information on the provisions is included in the model’s database. Off-model estimates for the effects of eliminating such provisions are based on the latest tax expenditure estimates published by JCT,7 except for a few provisions that Treasury, but not JCT, list as tax expenditures. Estimates of these latter provisions are based on Treasury’s tax expenditure estimates reported in the Budget.8 In general, the trend in the growth rate of each tax expenditure relative to the growth rate in GDP over the FY2014-FY2018 period (for JCT tax expenditures) or over the FY2018-FY2024 period (for Treasury tax expenditures) is assumed to continue in later years, and to be the same on both a fiscal year and calendar year basis. The following fiscal year splits are assumed to apply: all noncorporate business provisions (45/45/10), non-business individual income tax credits (10/90), wage-related tax expenditures (75/25), and all other individual tax expenditures (50/50). Special fiscal splits are used for tax-exempt bond provisions (these bonds are assumed to pay interest evenly over 20 years, with the corresponding corporate or individual split), the low-income housing credit (which is generally claimed over 10 years, with the corporate or individual credit split), and accelerated depreciation on rental housing (which applies over 39 years, with the corporate 45/45/10 split). Calendar year estimates for 2015 and later years are then made using these fiscal splits and the CBO forecast for calendar year GDP.9
If the proposed repeal of a tax expenditure is part of a broader proposal that changes statutory tax rates or otherwise changes effective tax rates (e.g., by repealing or imposing phaseins or phaseouts), the calendar year estimates for repeal of tax expenditures (other than credits) are adjusted. The adjustment factor is the ratio of the effective marginal tax rate (EMTR) for the proposal to the current law EMTR on the source of income affected (corporations, noncorporate business, wages, capital gains, interest, or life insurance interest).10 Credits are generally assumed to be usable to the same extent as under current law, and therefore are not adjusted. All calendar year estimates are then converted to fiscal year estimates using the corresponding fiscal year splits for each provision.
Expiring and Budget provisions. Off-model individual income tax estimates are also made for the proposed extension of certain provisions that have expired or will expire over the next few years (“extenders”), and for enactment of provisions included in the President’s Budget. Off-model estimates for such proposals are based on CBO or JCT estimates through FY2024 or FY2025.11 Unlike tax expenditures, for which JCT and Treasury estimates reflect fiscal year effects of the provision on all prior calendar year liabilities, estimates for extenders and budget provisions only reflect the change in calendar year liabilities beginning in the year the provision becomes effective.12 The initial fiscal year estimate for an extender or budget provision therefore reflects only the change in calendar year liability in the first year and the first year fiscal split, the second fiscal year estimate reflects only the change in calendar year liability in the first year and the second year fiscal split plus the change in calendar year liability in the second year and the first year fiscal split, etc. Assuming that each provision grows by a multiple, k of the growth rate in GDP (which could be more or less than 1), it is therefore possible to derive the fiscal year splits (which must sum to 1) and initial calendar year change in liability. The value of k is derived through iteration, solving for the value that makes the sum of the fiscal year splits closest to 1. Calendar year estimates for 2015 and later years are then made using k, the fiscal splits, the forecast for calendar year GDP, and the initial calendar year change in liability. Each provision is then categorized as a credit or a provision that changes taxable income. If the latter, its revenue effects depends on the marginal tax rate that applies to the income source: corporate, noncorporate business, wages, capital gains, interest, life insurance interest, payroll tax, or excise. Most business provisions affect both corporate and noncorporate businesses, so the corporate and the noncorporate business marginal tax rates are used to compute the revenue effects of the change in the corresponding amounts of taxable income. Consequently, the available estimates for such provisions had to be split into corporate and noncorporate portions. Generally, these splits are based on the shares of related tax expenditures that are attributed to corporate and individual (i.e., noncorporate business) taxpayers. As with tax expenditures, the final step in making the estimates for each such provision is to adjust these calendar year estimates (except credit provisions) by the corresponding ratio of EMTRs for the complete proposal to current law EMTRs, and convert these into fiscal year estimates using the corresponding fiscal year splits for the provision.
Outlay effects. The federal budget classifies the portion of refundable credits that exceeds a taxpayer’s tax liability as an outlay. However, the JCT generally includes the outlay portion of tax proposals in its revenue tables (although it provides separate estimates of any outlay effects of provisions in footnotes to its revenue tables).13 TPC generally follows the JCT convention of including outlay effects in its revenue tables for both on- and off-model estimates, but has the capability of separating the estimated receipts and outlay effects of proposals. Estimates for any outlay effects related to off-model estimates of tax expenditures, extenders, and Budget proposals are made using the same methodology used for the corresponding revenue estimate.
Off-model estimates for provisions that affect payroll taxes are prepared in the same way as the corresponding estimates for individual income tax effects. JCT does not report the effects of tax expenditures on payroll tax receipts, so to estimate the revenue effect of repealing payroll tax expenditures we first have to estimate their revenue effect under current law. We make these estimates by multiplying the individual income tax expenditure under current law by the ratio of the current law payroll tax EMTR to the current law individual income tax EMTR on wages. We then adjust for any change in the payroll tax EMTR that is part of the proposal.14 We make this adjustment by multiplying the estimates of payroll tax expenditures under current law by the ratio of the payroll tax EMTR under the proposal to the current law payroll tax EMTR.15 The estimates are then converted to a fiscal year basis (generally assuming a 75/25 split).
Corporate Income Tax
All corporate income tax estimates are made off-model. Some of the methods described below (such as for changes in cost recovery provisions) are used in making estimates for the individual income tax effects for noncorporate businesses as well as corporate income tax estimates.
Corporate AMT. To estimate the effect of proposals to repeal the corporate alternative minimum tax (AMT), we use the JCT estimates through FY2023 for that provision in Representative Camp’s “Tax Reform Act of 2014.”16 We assume that in future years revenues from the corporate AMT will represent the same percentage of GDP as in 2023.
Rate change. The first step in estimating the effect a corporate rate change is to estimate corporate taxable income under current law. We do this by adding projected corporate income tax receipts under current law (excluding receipts from the AMT), the amount of credits included in corporate tax expenditures, and TPC’s estimate of the amount of foreign tax credit claimed by corporations with an excess foreign tax credit limitation.17 This gives us total corporate income taxes before credits. We divide this amount by the average corporate tax rate (which we estimate to be 34.7 percent)18 to derive corporate taxable income. The effect of a proposed rate change is estimated by multiplying the change in rates by the estimate of corporate taxable income, and adjusting the result for behavioral responses (including changes in dividend payments, compliance levels, and shifts between corporate and other forms of business organization).19 We generally assume that credit use remains unchanged, except for proposals that reduce the corporate rate below the effective foreign tax rate for corporations currently in an excess credit position, which would reduce their foreign tax credit. These calendar year estimates are then converted to a fiscal year basis using a 45/45/10 calendar year liability/fiscal year receipts split.
Tax expenditure, expiring, and Budget provisions. Estimates of the revenue gain from repeal of corporate income tax expenditures are based on the latest JCT tax expenditures and estimated in the same manner as repeal of individual income tax expenditures. Estimates for corporate extenders and Budget proposals are also estimated in the same manner as for individual income tax provisions. All of these estimates are adjusted if these provisions are part of a broader proposal that changes the corporate income tax rate. Generally, the adjustment factor is the ratio of the proposed to the current law top corporate rate. These adjusted calendar year estimates are then converted to a fiscal year basis.
Depreciation, expensing, and other cost recovery methods. Various proposals would either lengthen depreciation lives, require at least partial capitalization of expenses (such as advertising and research and experimentation) that are currently expensed with amortization over some specified period, or substitute expensing for depreciation (and other current cost recovery methods). These proposals generally apply to investments made after the effective date and leave depreciation methods on existing assets unchanged. However, proposals that replace the corporate income tax with a value-added tax or other broad-based consumption tax might include special transition rules for the recovery of the undepreciated basis of business investments.
To estimate the effects of proposals that affect currently depreciable assets, we first estimate total investment in depreciable assets by businesses (corporate and noncorporate) in each year as 8.2 percent of GDP, the average ratio we calculate for 2003 through 2013 from NIPA data.20 We then allocate that investment by MACRS21 class and business ownership type using the average percentage distribution of investment for 2005-2007 from Table 5 in Mackie and Kitchen (2013).22 The proposed change in the depreciation allowances for each MACRS class is then applied, assuming that investments are all made mid-year. For example, for a proposal that would lengthen depreciation lives by requiring use of the ADS system, the difference between the mid-year MACRS and ADS allowance each year in each MACRS class would be applied to investment in each year, separately for corporate and noncorporate business owners. These estimates are then adjusted for any change in effective tax rates on corporations and/or noncorporate businesses, and converted to a fiscal year basis.
NIPA investment now includes spending on research and experimentation by type of business owner, which provides a basis for estimating the effect of proposals that would require capitalization and amortization of this spending. Estimates from the JCT or Treasury for proposals that would require capitalization and amortization of currently expensed items are also drawn on to produce TPC estimates of such proposals.
Interest deductions. Some proposals would disallow some or all of the deduction for interest paid by businesses (corporate and noncorporate), and remove some or all of the taxability of interest paid by businesses. Generally these changes do not apply to financial institutions, which either retain current-law treatment of interest paid and received, or are subject to some alternative method of taxation. These changes are also generally prospective (i.e., do not apply to interest paid or received on existing loans and debt instruments with fixed terms that are in place when the proposal becomes effective).
To estimate the revenue effect of such proposals, we rely on information on nonfinancial businesses’ levels of debt and loans reported in the Financial Accounts of the United States.23 Because proposals are generally prospective, we need to develop estimates of annual originations and retirements of debt and loans. We use the following method to derive such estimates from the data in the Financial Accounts. In year t, debt and loan balances (Bt) are the prior-year balance (Bt-1) plus originations (Ot) and less retirements (Rt):
Assuming originations grow at the rate of GDP,
Ot=Ot-1 GDPt /GDPt-1
Retirements in year t of debt or loans that originated in year i are assumed to be a fraction (ai) that changes over the n-year term of the debt or loan. So total retirements in year t are:
Rt=Σ aiOi = Σ aiOt GDPi /GDPt
Using the first and third equation above,
Ot=(Bt-Bt-1)/(1-Σ ai GDPi / GDPt)
Using this last equation and assumptions about ai and n for each type of debt or loan,24 we derived estimates of Ot/GDPt for 2006 through 2014 (the latest full year for which Financial Accounts data is currently available). Because of the financial crisis, the pattern over time for certain types of debt and loans was highly variable, so we decided to use the ratio for 2014 for all types except corporate bonds, for which we used the average ratio over the 2006-2014 period. These ratios were assumed to be fixed for 2015 and later years, and using these ratios and the first equation above allows us to estimate (for any starting date of a proposal that applies prospectively) the level of debt by type for which interest would no longer be deductible (i.e., debt that originated after the proposal’s start date) and the level of loans by type for which interest would no longer be taxable, separately for corporate and noncorporate businesses.
The applicable interest rate for each type of debt or loan in each year is estimated to be the 10-year Treasury rate plus a percentage point differential.25 Starting with the first year of the proposal, for each type of debt and loan we apply these interest rates to the remaining amount of debt or loan that originated in each year since the proposal went into effect and sum over all origination years. The change in the corporate (or noncorporate) tax base in each year is then the sum of the interest over each type of debt, less the sum of the interest over each type of loan.
The final step in the estimating process is to apply the proposed corporate rate or noncorporate EMTR, and convert these calendar year results to a fiscal year basis.
Other corporate proposals. For a wide range of proposals the JCT has produced estimates for the same, or a similar, proposal. For example, the JCT’s estimates for Representative Camp’s “Tax Reform Act of 2014” include separate estimates for such items as the deemed repatriation of the accumulated profits of foreign subsidiaries. Starting from these existing estimates, TPC makes adjustments for differences between the proposal and the one estimated by JCT, differences in related provisions in the proposal, differences in effective dates, timing differences (e.g., phaseins), and tax rate differentials.
All excise tax estimates are made off-model, with the exception of three ACA-related excises which are estimated on-model.26 For most excise tax proposals, we first estimate total calendar year receipts from the proposal. Then, following the procedure used by JCT, we reduce these amounts for the associated reduction (“offset”) in income (or consumption) and payroll tax revenues per dollar of additional excise tax revenue.27 The excise tax offset percentages differ among proposals, reflecting differences in marginal income (or consumption) and payroll tax rates applied to earnings and marginal tax rates applied to capital income and corporate profits. We convert these calendar year estimates to a fiscal year basis using a 75/25 split. Excise tax proposals that were proposed in the President’s Budget or that are extenders are estimated in the same manner as income and payroll Budget and extender provisions, but the EMTR adjustment is the ratio of the current law excise tax offset to the excise tax offset under the full proposal (as estimated on TPC’s microsimulation model).
Estimating sources and methods for particular excises are as follows.
Fuel excises. In order to estimate the revenue effect of fuel excise tax proposals, it is first necessary to split the forecast for baseline highway trust fund excise tax receipts (described above) between fuel excises and other highway trust fund taxes (primarily excises on trucks and trailers, tires and the use tax on heavy vehicles). These non-fuel highway excises are assumed to remain a fixed percentage of GDP (0.03 percent, the ratio for FY1999-FY2013, based on SOI data).28 The total for fuel excises is then converted to a calendar year basis and split between the excises on gasoline and diesel fuel using data on consumption by fuel from the year-by-year tables accompanying the EIA Annual Energy Outlook 2014.29 Projected prices for gasoline and diesel are also derived from the EIA data.
The effect on revenues of proposed changes in motor fuels excises is generally estimated separately for gasoline and diesel (some proposals differentially increase the rates on gasoline and diesel, or index the rates for inflation). The CBO fuel tax elasticity, which increases (in absolute value) over fifteen years following a change in a fuels tax rate from -0.06 to -0.4, is used in making the estimates to account for lower fuel use due to higher prices.30
Carbon tax. Proposals for a carbon tax are generally similar in structure to CBO 2011 revenue option 35,31 but with different effective dates, initial tax rates, and rate adjustment factors. The methodology used for estimating carbon tax revenues under such proposals follows the CBO methodology for carbon taxes on liquid fuels.32 This methodology is based on two equations, the first for the “price” of emissions (of CO2 equivalents, or CO2e):
where P is end use price of emissions (the value of end use consumption divided by total emissions in metric tons of CO2e), E is total emissions (in metric tons of CO2e) from end use of energy, and the subscripts c, n and p are for coal, natural gas and petroleum products.
The second equation is:
where Ebase is from the first equation, P is from the first equation, T is the carbon tax rate (or equivalent for a cap-and-trade policy) in dollars per metric ton of CO2e, and s is a sensitivity parameter (effectively an elasticity) that CBO derived and that varies over time. Since GDP is fixed for purposes of our estimates, the term (GDPpolicy/GDPbase) is always 1, and the equation is the same as used for other excise tax estimates.
Values for E, Ec, En and Ep are provided in the EIA forecast. If a proposal for a carbon tax is included in a broader proposal that also includes proposed increases in motor fuel excises, these values are adjusted to reflect the lower level of emissions due to higher motor fuel excises. However, these values only reflect the amount of end use emissions from combustion of fossil fuels. To account for other greenhouse gas emissions, these values are “grossed up” by the ratio of fossil fuel combustion emissions to total greenhouse gas emissions based on EPA estimates for 2013.33
The values of Pc, Pn, and Pp are derived from the EIA forecast for end users of these fuels, with the values adjusted for proposals that also increase motor fuel excises. The value of P is then computed each year from the first equation above.
CBO only reports the value of s for the first year a carbon tax is in place (-0.4), the fourth year (“nearly -0.6”) and the 38th year (-1.5). For the second and third years, the value of s is assumed to decline (i.e., increase in absolute magnitude) linearly between -0.4 and -0.6, and between the fourth and 38th year to decline linearly between -0.6 and -1.5.
With the value of all variables determined, the second equation is used to determine the level of emissions under the proposal in each year. The tax rate for the year (which may grow at a rate specified as part of the proposal) is applied to derive the calendar year, pre-offset amount of revenue from the proposal. The effect of the carbon tax on motor fuel consumption is also calculated and the corresponding reduction in motor fuels excises netted against carbon tax receipts.
Financial transactions tax. Proposals for a financial transactions tax may include stock trades, bond trades, option premiums, foreign exchange spot transactions, futures, and swaps or the underlying notional values of futures and swaps in the base. Values for these transactions for 2014 were obtained from the various sources cited in Baker, et al. (2009),34 with some differences in which source was used for a particular transaction.
Transactions costs for futures are generally taken from Schulmeister, et al.35 and the same values used for swaps. Foreign currency transactions cost were taken from Matheson.36 Transactions costs for stock trades are assumed to be 100 times the cost for a corresponding futures trade in the same instrument,37 with higher costs assumed in smaller markets (e.g., BATS).
We generally assume that the price elasticity for all transactions is -1.25. (The “price” of a transaction is the estimated transactions cost as a share of the dollar value plus the tax). This elasticity assumption appears reasonable based on the various elasticity estimates reported in Matheson.38
Transactions are forecast for 2015 and later years assuming they all grow from their 2014 level at the same rate as nominal GDP, rather than at the much higher rate preceding the financial crisis. Transactions costs are assumed to decline at a uniform rate to half their 2010 levels by 2035, continuing (at a much slower rate) the decline in transactions costs over the past several decades. Transactions, transactions costs, tax rates and elasticities are then used to produce calendar year estimates of gross receipts from the tax. As with other excises, the offset is then applied and the estimate converted to a fiscal year basis.
 See “Brief Description of the Tax Model”
 TPC’s methodology for estimating the revenue effect of a value-added tax is described in Eric Toder, Jim Nunns and Joseph Rosenberg, Methodology for Distributing a VAT, Urban-Brookings Tax Policy Center and the Pew Charitable Trusts, April 12, 2011. Estimates for other broad-based consumption taxes (such as an X-tax or flat tax) start with this methodology and use on-model estimates for the effects of household-level exemptions, deductions, credits, and (for the X-tax) graduated rates.
 Congressional Budget Office. The Budget and Economic Outlook: 2015 to 2025, January 2015.
 Congressional Budget Office. The 2015 Long-Term Budget Outlook, June 2015.
 Congressional Budget Office. The 2014 Long-Term Budget Outlook, July 2014. CBO did not include a forecast for individual income tax receipts as a percentage of GDP in its 2015 long-term projections.
 These include Federal Reserve remittances, customs duties, and miscellaneous receipts (which include the fees and fines imposed by the ACA). We project these baseline amounts because some proposals modify or repeal these other sources of revenue.
 Joint Committee on Taxation, Estimates of Federal Tax Expenditures for Fiscal years 2014-2018, JCX-97-14, August 5, 2014.
 Office of Management and Budget, Analytical Perspectives, Federal Receipts, Tax Expenditures (Chapter 14), Budget of the United States Government, Fiscal Year 2016, February 2015.
 For 2015 through 2025 the calendar year GDP forecast is from the January 2015 CBO forecast. After 2025 calendar year GDP is assumed to grow at the same rate each year at the annual compound growth rate of GDP over fiscal years 2026 through 2040 as forecast by CBO in its June 2015 long-term forecast.
 The top corporate rate is generally used as the EMTR on corporations. The EMTRs on other sources of income are computed on TPC's microsimulation model as follows (using wages as an example of the calculations): The wages of all workers is increased by $1,000, the change in income tax (and payroll tax, for wages and business income subject to SECA only) is computed, then the tax change as a percent of $1,000 (i.e. the effective rate on the marginal $1,000 of wages) is computed; and finally these effective rates are weighted by current wages.
 Estimates for most "extenders" were taken from the supplemental materials for Congressional Budget Office, An Update to the Budget and Economic Outlook: 2014-2024, August 2014, and then adjusted for the estimates by the Joint Committee on Taxation (JCX-107-14R, December 4, 2014) for the "Tax Increase Prevention Act of 2014" (H.R. 5771), which extended most items through 2014. Estimates for the President's FY2015 Budget proposals are from Joint Committee on Taxation, Estimated Budget Effects of the Revenue Provisions Contained in the President's Fiscal Year 2015 Budget Proposal (JCX-36-14, April 15, 2014) and for the President's FY2016 Budget proposals are from Joint Committee on Taxation, Estimated Budget Effects of the Revenue Provisions Contained in the President's Fiscal Year 2016 Budget Proposal (JCX-50-15, March 6, 2015).
 For some provisions, there may be a revenue effect in the preceding year due to taxpayers' anticipatory behavior.
 An exception is that for tax expenditure estimates JCT excludes outlay effects (but reports them separately in footnotes).
 Changes in the payroll tax EMTR could be the result of proposed payroll tax rate or base changes (such as raising the OASDI wage cap or repealing exclusions from the payroll tax base such as cafeteria plans), or changes to provisions affecting employer-provided health insurance, which would change the share of compensation entering the payroll tax base.
 The payroll tax EMTRs are computed on TPC's microsimulation model.
 See Joint Committee on Taxation, JCX-20-14, February 26, 2014.
 The foreign tax credit is generally limited to the amount of U.S. tax attributable to a corporation's foreign-source income. The Statistics of Income (SOI) Division of IRS published a table showing relevant items by industry for corporations that were in an excess credit position in 2010 (available athttps://www.irs.gov/uac/SOI-Tax-Stats-Corporate-Foreign-Tax-Credit-Table...). We use this SOI data to make our estimates for 2015 and future years.
 The average rate is slightly below the top statutory marginal rate of 35 percent because current corporate income tax rates are graduated. The benefits of graduated rates are phased out for the largest corporate taxpayers, but smaller corporations pay an average rate lower than 35 percent.
 We generally use a behavioral response of 10 percent, which appears consistent with the response implicit with JCT and Treasury estimates for at least modest changes in the corporate rate.
 Specifically, our calculation is based on the data in NIPA Table 2.7, "Investment in Private Fixed Assets, Equipment, Structures, and Intellectual Property Products by Type," adjusted for investment by households and nonprofit organizations as shown in NIPA Table 4.7 "Investment in Private Nonresidential Fixed Assets by Industry Group and Legal Form of Organization."
 MACRS is the Modified Accelerated Cost Recovery System. MACRS generally allows faster depreciation than the ADS (Alternative Depreciation System), which for most machinery, equipment, vehicles, furniture, buildings, and other structures is the default depreciation method.
 James B. Mackie III and John Kitchen, "Slowing Depreciation in Corporate Tax Reform," Tax Notes, April 29, 2013. The percentages for structures (MACRS 27.5 year and 39 year) classes are grossed up to account for sales of existing structures that become depreciable to the new owner.
 These accounts (formerly labeled the Flow of Funds Accounts) are produced quarterly by the Board of Governors of the Federal Reserve System. We use the debt and loan information in Tables L.103, Nonfinancial Corporate Business, and L.104, Nonfinancial Noncorporate Business.
 For example, we assumed that for all mortgages n=20 and the ai are the percentage of principal paid in year i on a conventional mortgage.
 For example, the (weighted average) rate on corporate bonds in each year starting in 2015 is assumed to be the 10-year Treasury rate (as forecast by CBO) plus 2 percentage points.
 The ACA excises modeled on-model are the excise on high-premium employer plans (the "Cadillac tax"), the employer mandate, and the individual mandate.
 Because GDP and the total price level are held fixed, imposing an excise tax generally reduces net business revenues that can be paid to workers and owners as wages and profits. However, if an excise is included in taxable consumption for a proposal that includes a consumption tax, there is no consumption tax offset.
 Internal Revenue Service, Statistics of Income Division, Table 20. Federal Excise Taxes Reported to or Collected by the Internal Revenue Service, Alcohol and Tobacco Tax and Trade Bureau, and Customs Service, by Type of Excise Tax, Fiscal Years 1999-2013.
 U.S. Department of Energy, Energy Information Administration, Annual Energy Outlook 2014 (with Projections to 2040), April 2014.
 In general, pre-offset changes in excise tax revenues (ΔT) were estimated using the following formula: ΔT=q*Δp*Δ(1+Δp/p)e where Δp is the change in price (the change in the tax rate), p is the price (before the tax rate change, but including any existing excise), and e is the elasticity (which might vary over time).
 Congressional Budget Office, Reducing the Deficit: Spending and Revenue Options, March 2011.
 Congressional Budget Office, How CBO Estimates the Cost of Reducing Greenhouse-Gas Emissions, Background Paper, April 2009.
 The 2013 gross-up factor was derived from EPA estimates of greenhouse gas emissions in Tables 2-1 and 2-6 in Environmental Protection Agency, DRAFT Inventory of U.S. Greenhouse Gas, 6 Emissions and Sinks: 1990-2013, February 11, 2015. The 2013 gross-up factor was used in all subsequent years.
 Dean Baker, Robert Pollin, Travis McArthur and Matt Sherman, The Potential Revenue from Financial Transactions Taxes, Center of Economic and Policy Research, Political Economiy Research Institute, University of Massachusetts Amherst, Issue Brief, December 2009.
 See Table A4, pg. 67 in: Stephen Schulmeister, Margit Schratzenstaller-Altzinger and Oliver Picek, A General Financial Transaction Tax: Motives, Revenues, Feasibility and Effects, Österreichisches Institut Dür Wirtschaftsforschung, March 2008.
 See: Thornton Matheson, Taxing Financial Transactions: Issues and Evidence, International Monetary Fund, IMF Working Paper WP/11/54, March 2011.
 Schulmeister, et al. state, "...transactions costs in 'traditional' spot markets are roughly 100 times higher than in 'modern' derivates markets" (pg. 48), and using this factor provides estimates that appear reasonable relative to the older estimates cited, for example, in Pollin et al.
 See Table 5, pg. 17.