Key Points
Budget scoring conventions used by government agencies (i) do not include estimates of additional revenues associated with new IRS enforcement spending and (ii) would count 2022 as the first year of the 10-year budget window. Under these assumptions, we project that the Ultra-Millionaire Tax Act introduced by Senator Elizabeth Warren raises $2.1 trillion over the 2022-2031 budget window.
Allowing for the effects of greater enforcement spending, our projection increases to $2.4 trillion. In addition, delaying the 10-year window by one year, our projection increases to $2.7 trillion between 2023-2032. This delayed projection window corresponds to the $3 trillion estimate recently cited by Senator Warren.
The above “conventional” estimates do not include economic feedback effects. Including macroeconomic effects, we project that the Act would raise $2.0 trillion over the 2022-2031 budget window and $2.3 trillion over 2023-2032.
We estimate that the Act would reduce capital by 3.1 percent, average hourly wages by 1.2 percent, and GDP by 1.2 percent in 2050.
Budgetary and Economic Effects of Senator Elizabeth Warren’s Wealth Tax Legislation
Senator Elizabeth Warren (D-MA) recently introduced the Ultra-Millionaire Tax Act of 2021 (the “Act”), a legislative proposal for a federal tax (the Ultra-Millionaire Tax, or UMT) on the wealth of high-net worth families. The bill follows Senator Warren’s presidential campaign proposal to institute a progressive wealth tax, which PWBM analyzed in 2019.
The current Act would assess a tax of 2 percent on net worth above $50 million and 3 percent on net worth above $1 billion.1 The UMT would be implemented in 2023 on the value of net worth calculated in 2022. The Act also would allocate $100 billion in funding to the Internal Revenue Service (IRS) to strengthen enforcement efforts, improve taxpayer services, and modernize their IT systems. Among taxpayers subject to the UMT, a minimum 30 percent audit rate would be implemented. For these taxpayers with wealth above $50 million, a 4 percent “exit tax” for any who renounces their citizenship.2 The bill also includes additional anti-evasion and avoidance measures including systematic third-party reporting.
Budget scoring conventions used by the Congressional Budget Office (CBO) and Joint Committee on Taxation (JCT) do not include any increase in revenue due to the Act’s additional spending on IRS enforcement. The first row of Table 1 presents PWBM’s estimate of year-by-year revenue estimates for fiscal years 2022-2032 under this CBO/JCT scoring convention. The second row of Table 1 shows net revenue estimates but does include revenue from the additional IRS outlays specified in the Act. These first two estimates are “conventional” and do not account for economic feedback effects. The third row of Table 1 shows dynamic revenue estimates which incorporate macroeconomic effects of the Act.
| 2022 | 2023 | 2024 | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 | Standard budget window, 2022-2031 | Delayed budget window, 2023-2032 | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Conventional | Standard scoring conventions | -10 | 232 | 236 | 230 | 223 | 216 | 220 | 227 | 235 | 244 | 254 | 2,053 | 2,317 |
| Including new IRS Enforcement | -9 | 220 | 238 | 249 | 260 | 269 | 279 | 286 | 294 | 303 | 315 | 2,390 | 2,714 | |
| Dynamic | Including new IRS Enforcement | -16 | 162 | 222 | 227 | 231 | 235 | 237 | 239 | 243 | 245 | 251 | 2,025 | 2,291 |
Without accounting for revenue effects of the Act’s new IRS enforcement spending—the CBO/JCT scoring convention—we project that the UMT would raise roughly $2.1 trillion over fiscal years 2022-2031. When these enforcement effects are included, we project that instead the UMT would raise a net $2.4 trillion over the budget window 2022-2031. Both of these estimates include the effects of tax avoidance and evasion, which directly lower the amount of revenue raised, but do not include macroeconomic effects. Incorporating macroeconomic effects, we estimate that the UMT would instead raise $2.0 trillion over 2022-2031. The technical appendix below discusses the sensitivity of net revenue estimates to assumptions about avoidance. Because the UMT does not come into effect until 2023, it raises no revenue in calendar year 2022 under either scoring convention.
The one-year delay in implementation was not included in Warren’s 2019 wealth tax proposal, and so the estimate presented above is not directly comparable to our 2019 analysis. For a more direct comparison we delay the 10-year window by one year, estimating that—on a conventional basis—the Ultra-Millionaire Tax Act would raise $2.7 trillion over 2023-2032; $60 billion greater than our estimate for the 2019 proposal of $2.64 trillion over 2021-2030, and roughly $30 billion less than the 2023-2032 revenue estimate cited by Senator Warren. On a dynamic basis, we estimate that the Act would raise $2.3 trillion over 2023-2032, $700 billion less than the estimate cited by Senator Warren. The specific effects and modeling choices leading to these differences in estimates are discussed in the technical appendix below.
Table 2 presents the economic effects from using the new revenue from the UMT to reduce federal budget deficits. This estimate includes updates to our 2019 estimate’s modeling and estimation of bequests and inheritances.
| Year | GDP | Capital stock | Average hourly wage | Hours worked |
|---|---|---|---|---|
| 2031 | -0.6 | -1.4 | -0.7 | 0.0 |
| 2040 | -0.8 | -2.2 | -1.0 | 0.1 |
| 2050 | -1.2 | -3.1 | -1.2 | 0.0 |
Smaller federal deficits translate into less crowding out and thus increased national saving and greater capital accumulation. However, wealthy households that face a tax on their savings choose to save less and thus accumulate less capital. The net effect is a decline in the total capital stock of 1.4 percent in 2031 and 3.1 percent in 2050. This decline in capital in turn makes workers less productive, which is reflected by a decline in wages of 0.7 percent in 2031 and 1.2 percent in 2050. Lower private capital leads GDP to decline by 0.6 percent in 2031 and 1.2 percent in 2050.
The analysis presented above contains differences from PWBM’s previous analysis of a wealth tax proposal from Senator Warren. Some of the differences result from changes in the policy proposal and others result from PWBM model updates. Below, we explain each difference and describe its effect on the ultimate budget estimate. Please refer to the Technical Appendix of our previous analysis for detailed information on our modeling strategy.
Senator Warren’s campaign version of a wealth tax proposal set the top bracket at 6 percent. The UMT sets the top rate at 3 percent instead. A lower rate reduces revenue raised, but we project that wealth tax avoidance among billionaires will be smaller under a 3 percent rate, which increases taxable wealth and thus revenues. On net, these effects reduce the budget estimate relative to our 2019 analysis.
The previous analysis assumed that the wealth tax enactment year was 2021; the Act specifies that the UMT begins in 2023. This change increases the budget estimate relative to our 2019 analysis.
The Act specifies that any wealth tax paid will be deductible from gross estate in the year of a decedents’ death. PWBM assumed wealth tax payments would not be deductible in our previous analysis. This change slightly reduces the budget estimate relative to our 2019 analysis.
Wealth taxes mechanically slow the accumulation of wealth for those who are subject to the tax, assuming that consumption patterns remain fixed. This reduction in wealth will result in lower capital income for those subject to the tax, which means lower income tax revenues. As noted in Leiserson (2020), the fixed GNP convention for budget estimates implies that this wealth loss would end up owned by households not subject to the wealth tax, perhaps increasing taxable income. But the proliferation of nontaxable investment accounts, foreign owners of US assets not subject to US income taxes, and the progressive structure of income taxes means any counterbalancing effects will be negligible.
Over time, the decumulation effect of wealth taxes would also reduce the value of estates at death, which would lead to lower estate tax revenues.
PWBM did not incorporate these “offset” effects in the 2019 analysis. Including these effects reduces the budget estimate relative to our 2019 analysis.
PWBM’s previous analysis relied on the 2016 Survey of Consumer Finances (SCF), the most recent survey of wealth microdata at the time of that writing. Since then, the 2019 SCF has been released; this updated data serves as the basis of our model. Updating the data reduces the budget estimate relative to our 2019 analysis.
PWBM previously assumed a 6 percent annual growth rate in net worth. Our assessment of the current state of macro-financial conditions leads us to adjust this assumption to 5 percent. This change reduces the budget estimate relative to our 2019 analysis.
The Act includes several enforcement provisions not clearly specified in Senator Warren’s campaign proposal. Specifically, the Act would appropriate an additional $10 billion dollars per year from 2022-2032 towards IRS funding, 70 percent of which would be directed towards enforcement efforts for the UMT specifically.
As per our previous analysis, we begin with a semi-elasticity of taxable wealth of –13.3 This elasticity implies that a one percentage point increase in the wealth tax rate leads to a 13 percent reduction in taxable wealth. This elasticity reflects both tax evasion (illegal means of reducing taxes such as misreporting wealth) and tax avoidance (legal means, like shifting wealth towards lower-taxed asset classes). The Act’s provisions for IRS enforcement would directly reduce evasion, as additional audits would bring in would-be evaded revenue.
To analyze the quantitative effect of the enforcement provision, we first estimate the would-be size of the wealth tax gap -- the difference between taxes owed and taxes paid -- under the current-law level of IRS funding. Administratively, the element of the income tax most similar to a progressive wealth tax is private business, estate and trust, and capital gains income – categories of income that are concentrated among the richest taxpayers and where information reporting is comparatively more difficult to implement. The IRS estimates the 2011-2013 tax gap for this type of income to be about 17 percent.
We then calculate the implied evasion elasticity with respect to the wealth-tax-equivalent tax rate. Historical TAXSIM data show that these types of income faced a roughly 30 percent marginal income tax rate on average in those years. Assuming a nominal rate of return of 8 percent, the implied evasion semi-elasticity is about –8.4 This number represents the baseline evasion response to wealth taxation; with additional enforcement efforts, it would fall in absolute value.
The next step is to estimate the size of this reduction under the Act. Multiple analyses suggest that at the current margin, a dollar of IRS funding would lead to more than one additional dollar in revenues collected; that is, a return on investment (ROI) of greater than 1. To inform our estimates of how much additional wealth tax and income tax revenue would be raised under the Act, we begin with estimates calculated by Holtzblatt and McGuire (2020) of the ROI specifically on marginal audits of high-income business returns. The ROI numbers vary over time which reflects factors like time spent training new agents, the increasing difficulty of audits at the margin as “low-hanging fruit” projects are taken on first, and taxpayer behavior changing in response to enforcement patterns.
We make two adjustments to these ROI factors. First, we account for the tax rate differential between the current income tax and proposed wealth tax. Second, we reduce the ROI factors to account for the diminishing marginal returns to additional funding, as the proposed IRS funding is substantially larger than what was estimated in Holtzblatt and McGuire (2020). The final step is to re-estimate the wealth tax gap under the proposed enforcement regime, re-calculate the implied evasion elasticity, and subtract the difference from our total wealth tax elasticity. The result is a total semi-elasticity of evasion and avoidance ranging from –8.4 to –9.4 depending on the year. Finally, following CBO, we convert tax-year liability resulting from audits to receipts in order to project when the additional revenues are actually collected.
Accounting for these enforcement measures increases the budget estimate relative to our 2019 analysis.
This analysis was conducted by Jon Huntley and John Ricco under the direction of Richard Prisinzano.
-
The proposal includes a trigger that would raise the 3 percent rate to 6 percent in the event that the United States creates a federal universal health insurance program and bans private health insurance. For the analysis contained herein, PWBM assumes current law for federal health insurance programs. ↩
-
An “exit tax” of sorts already exists under current law in the form forced realization of accrued capital gains. For the analysis herein, we assume that taxpayers do not change their behavior in 2021 in anticipation of the Act being enacted in 2022. While this assumption is standard in scoring, taxpayers might be able to avoid the additional exit tax under this Act by renouncing their citizenship prior to the Act’s passage into law. ↩
-
Refer to the Technical Appendix of our previous analysis for a detailed discussion of how we arrive at this value. ↩
-
Calculated as ln(1 - 0.17) / (0.3 * .08). ↩