Tax reform ideas played an important role in the recent Presidential election. Republican candidate Mitt Romney proposed large tax cuts and other changes that he said could be part of a revenue-neutral tax reform that also retained low rates on savings and investment and would not raise taxes on the middle class. In an earlier analysis, we showed that it was not possible to achieve all of Romney’s stated goals simultaneously. This paper reviews that analysis and critiques several responses to our analysis. Legislating realistic tax reform will require recognition of the difficult trade-offs among these competing goals.
In February, 2012, Republican presidential candidate Mitt Romney laid out a four-part agenda for tax reform that featured cutting income tax rates by 20 percent from today’s levels, promoting “savings and investment for the American people,” repealing the estate tax, and repealing the alternative minimum tax. He also proposed a variety of additional tax cuts, including a 29 percent reduction in the corporate tax rate. Taken together, the proposed tax cuts would reduce revenues by roughly $456 billion in 2015 or about $5 trillion over the next decade. Governor Romney said that the changes would be part of a revenue-neutral reform that would offset the cost of the cuts by eliminating tax breaks and not raise taxes on the middle class, but he did not specify which tax breaks would be eliminated.
In August, the Tax Policy Center (TPC) published a paper we co-authored that analyzed these economic and policy goals. As an illustration of the difficulty inherent in implementing such a substantial tax reform, our paper showed that achieving all of the goals listed above would, under reasonable assumptions about what tax breaks might be on the table, result in tax cuts for households with income above $200,000. Given the goal of revenue neutrality, this would seem to require tax increases on households with income below $200,000, something that Governor Romney has said he does not want to happen.
This conclusion was a mathematical demonstration of the difficulty of accomplishing simultaneously all of the goals Romney had laid out. It was neither a statement of Romney’s intentions nor a political prediction of what he would actually do if elected president. Indeed, Romney had also indicated that he did not want to cut taxes on high-income tax payers and that he wanted to reduce the burden for middle-income households. Thus, our results demonstrated the more general conclusion that something would have to give: Romney’s proposals would fall short on at least one of the goals he had set out. Any practical reform based on his proposals would entail some combination of reduced revenues, tax increases on households with income below $200,000, higher taxes on saving and investment, smaller reductions in income tax rates, or other changes.
Our conclusions held even though we imposed the base-broadening measures in the most progressive manner possible, eliminating them for the highest-income households first before affecting anyone else. This was intended to place an upper bound on what the plan could achieve in terms of progressivity. Our analysis noted that, in reality, practical and administrative challenges of implementing the base-broadening reforms in this manner would mean that any realistic effort to close tax expenditures consistent with Romney’s goals would require higher effective tax rates and would yield even less progressivity, lower revenues, and more tax complexity than we had modeled. Furthermore, we excluded the effects of the corporate tax cuts, despite evidence that these provisions would also result in lower revenues.
Our results held even when we incorporated revenue feedback, not just according to the standard “microdynamic” effects used by TPC, Treasury, and the Joint Committee on Taxation, but also additional feedback effects from potential economic growth, based on estimates from a Romney adviser, even though we believe those estimates are overly optimistic.
In light of several responses to our work, some recent suggestions by Romney on how to finance part of his proposed tax cuts, and ongoing interest in tax reform, this paper reviews the discussion and to respond to various issues that have arisen in response to our paper.
We offer several conclusions. First, we continue to stand by our original results. Although some media outlets and individuals have interpreted a second document that we published as somehow walking back or disavowing our earlier results, that is simply a misinterpretation on their part.
Second, after reviewing several responses to our study, we note that even with the freedom to choose from a much larger set of “pay-fors,” not a single author has offered a plan that generates sufficient revenue to pay for all of Romney’s (non-corporate) tax cuts, exempts households with income below $200,000 from higher taxation, and is administratively feasible. None of the responses actually refutes our study – that is, none of them shows that our conclusions are wrong, given what Romney laid out as goals. Rather, the responses either shift the goal posts – by focusing on tax increases on different households or changing the baseline –
fail to raise sufficient revenue to offset all the tax cuts Romney has proposed; raise taxes on saving and investment, in violation of Romney’s goal of promoting these activities; or do a combination of these things.
Third, Romney subsequently described several ideas to help pay for some of his proposed tax cuts. This was a welcome update to the discussion, but the ideas (which Romney described but did not formally endorse) would not generate sufficient revenue to pay for the tax cuts.
Our paper proceeds as follows: Section II describes Romney’s proposals. Section III summarizes our earlier analysis. Section IV discusses and critiques the responses to our work. Section V discusses Romney’s recent ideas for paying for his proposals. Section VI concludes. An Appendix provides a detailed response to Rosen (2012).
II. Romney’s Proposals
In the Wall Street Journal on February 23, 2012, Romney published an op-ed laying out his economic plan, focused on four items:
- Cutting income tax rates by 20 percent below their 2012 level,
- Promoting saving and investment (including eliminating taxes on capital gains, qualified dividends and interest incomes for households with income below $200,000; and retaining the current 15 percent top rates on capital gains and dividends for high-income households),
- Repealing the estate tax, and
- Repealing the alternative minimum tax.
He also proposed a series of other tax changes over the course of the campaign, including:
- Repealing the surtaxes on investment and earnings of upper-income taxpayers that were enacted in the Affordable Care Act,
- Extending all of the 2001/03 tax cuts,
- Allowing the 2009/10 tax cuts to expire, and
- Reducing the corporate tax rate to 25 percent from 35 percent.
III. Our Analysis
The purpose of our analysis was to examine the trade-offs inherent in tax reform. We drew on the Tax Policy Center microsimulation model, which was developed to provide revenue and distributional estimates of tax policies that had otherwise only been available from government sources, like the Joint Committee on Taxation (JCT) or the Department of the Treasury.
Much like the models used at these government agencies, the TPC model uses large samples of individual households from multiple data sources, projects the income and other tax-related variables of these households over the budget window, and estimates the taxes owed by each sample household under a specified proposal. The model includes estimates of the ‘microdynamic’ effects on revenues that occur when, for example taxpayers increase reported taxable income or capital gains in response to tax rate cuts, but not ‘macrodynamic’ effects of faster economic growth. (Our paper, however, discussed potential macrodynamic effects of the Romney proposal.)
While using a microsimulation model is more complex than using published tables from the IRS, estimates based on a large sample of individual income tax returns are capable of assessing the interaction of different tax provisions (for example, between changes to the AMT and to the regular income tax), and determining tax liability for households with different levels and sources of income, uses of funds, and family circumstances. This is the reason federal agencies and some private groups (including TPC and the National Bureau of Economic Research) use this type of model to estimate revenue and distributional effects of tax reform proposals. In short, by drawing on the TPC model, we can provide timely analysis of important policy-relevant issues using the basic methods of tax analysis used by the government agencies that provide the official scores of tax proposals.
Regarding Romney’s proposals, TPC estimated in March that, in addition to extending the entirety of the Bush tax cuts, Romney’s explicit proposals would reduce revenues in 2015 by $481 billion on a static basis and by $456 billion after accounting for microdynamic behavioral responses. (The commonly cited figure that Romney proposed $5 trillion in gross tax cuts, with unspecified revenue raisers, is based on the $456 billion estimate for 2015, adjusted for projected baseline growth in the size of the economy over the next decade.)
In our paper, we compared the Romney proposals to a current policy baseline and did not aim to finance the corporate tax cuts — we only included the tax cuts listed in the first five bullets above. With these parameters, we estimated that Romney’s proposed cuts would reduce revenues by $360 billion in 2015, after including microbehavioral responses.
A key consideration of our analysis was how such cuts would be financed. In the absence of any specific proposals from the Romney campaign, we looked to existing tax expenditures that could be closed in order to raise revenues. Similar to a previous TPC study, we divided tax preferences into several groups: (1) exclusions of income from sources that are administratively difficult to tax; (2) tax preferences for saving and investment; and (3) all other tax expenditures.
Of these possible revenue-raisers, we excluded tax expenditures in the first group. These preferences–including imputed rent from owner-occupied housing, and other items–are typically excluded from tax reform proposals due to their administrative complexities (and sometimes for other reasons as well). We also excluded consideration of closing preferences aimed at saving and investment, because of the prominence of promoting saving and investment in Romney’s economic proposals. (We discuss this further in the next section.) We considered all other tax expenditures (including those related to itemized deductions for mortgage interest, charitable contributions, and state and local taxes; the exemption of employee income in the form of health insurance; various other exclusions, above the line-deductions, and tax credits) fair game to pay for the proposed tax cuts.
A key aspect of our analysis is that we closed the available tax expenditures “from the top down.” That is, we offset revenue losses from tax rate reductions by first eliminating tax expenditures for the highest-income groups. If that did not generate enough revenue to pay for the tax cuts, we then closed all available tax expenditures of the next highest-income group and so on. Although it would be both administratively and politically impractical as well as economically damaging (because of the high marginal tax spike it would create) to eliminate all tax expenditures only above a given income threshold, it made the financing of the tax proposals as progressive as could be, for those tax expenditures.
The central results are summarized in Figure 2 from our paper, reproduced below. Among households with income above $1,000,000, Romney’s tax cut proposals (again, ignoring the corporate tax) would reduce tax payments by $106 billion. However, even the complete elimination of all the available tax expenditures (i.e., excluding those administratively difficult to tax or those affecting saving and investment) would only raise $54 billion in revenue from this group. As a result, they would receive a net tax cut of $52 billion. Likewise, each group of households with income of $200,000 or above would receive a net tax cut under Romney’s proposals, even if all available tax expenditures were eliminated from the top down. In total, households with income above $200,000 would receive a net tax cut of $86 billion (consisting of a gross tax cut of $251 billion, partially offset by base-broadeners of $165 billion) even if all of their available tax expenditures were closed.
As a matter of arithmetic, in order for the overall tax package to be revenue neutral and preserve incentives for saving and investment, the $86 billion in lower taxes paid by those with income above $200,000 would require a $86 billion tax increase that would be spread out in some fashion over the 95 percent of the population with income below $200,000. This would require a 58 percent reduction in the value of tax expenditures that go to these households. We did not specify how this burden would be distributed among households with income below $200,000. We did note that if the increased burden were shared equally (as a percentage of income) among all households in this group, after-tax income would fall by an average of 1.2 percent. (The average tax increase would need to be about $500 per household, and about $2,000 for households with children.)
This conclusion is mathematical in nature – it is what the analysis shows would have to happen to meet Romney’s stated goals. It was not a prediction of Romney’s actions, were he elected, and Romney has said, before and since, that he does not want to raise taxes on the middle class. That implies an inherent mathematical contradiction between the goals of his proposals: one or more of those goals would have to be compromised.
Finally, we included an analysis of the potential revenue feedback effects from economic growth. We used a model that was developed by Romney advisor Greg Mankiw, and included a significant feedback effect that allowed 15 percent of the tax cut to be paid for through higher economic growth. This implies that the $360 billion cost would fall to about $307 billion. It is not clear how to allocate the increase in income, and hence the added tax payments, across income groups, but it is highly likely that income gains from faster economic growth would increase revenues from both high- and low-income taxpayers. What is clear is that, under any allocation of the changes across income classes, households with income above $200,000 would still receive a large tax cut that would require tax increases on households with income below $200,000 in order to maintain revenue neutrality.
Since the publication of our paper, several responses have been issued. None of the analyses contradict the results in our study – that is, none shows that our results do not hold, given the goals laid out by the candidate. Moreover, none of the authors of the responses argues that the alternatives they considered would be desirable economic policy. The authors are focused instead on attempting to generate the existence of a plan that could meet all of Romney’s goals. Even so, none of the responses offers a plan that generates sufficient revenue to pay for all of Romney’s (non-corporate) tax cuts, exempts households with income below $200,000 from higher taxation, and is administratively feasible.
We divide our discussion of the responses into three subsections. First, we discuss two overarching issues in the responses: raising taxes on the return to saving and the impact of the Romney proposals on economic growth. Second, we discuss three responses that build directly off of the TPC results. Third, we discuss three responses that use 2009 IRS data to examine the issues.
A. Overarching Issues in the Responses
1. Taxing saving and investment
As noted above, Romney made promoting saving and investment one of his four main planks for pro-growth tax reform, on a par with the income tax cuts and repeal of the estate tax and AMT. In Believe in America: Mitt Romney’s Plan for Jobs and Economic Growth, the major heading for the tax policy section is “Mitt Romney’s Plan: Promote Savings and Investment.”
The claim of promoting savings and investment is an imprecise policy goal and is therefore open to interpretation. It could mean to preserve every incentive for savings and investment or it could mean on net to preserve broad incentives while leaving open the possibility of eliminating more targeted incentives. In the Wall Street Journal op-ed, Romney wrote that “I will promote savings and investment by maintaining the low 15% rate on capital gains, interest [sic] and qualified dividends, and eliminate the tax entirely for those with annual income below $200,000. These low tax rates will create powerful incentives for Americans to save and invest, while encouraging business investment and economic growth.” This does not rule out imposing new taxes on some forms of saving.
Nevertheless, we did not impose new taxes on saving in our calculations because his language and goals do not seem consistent with the idea of raising taxes on any specific types of saving. Specifically, we do not see how raising taxes on various forms of saving is consistent with the goal of promoting saving.
In contrast, all of the authors who have criticized us have suggested including revenue raised from subjecting some forms of saving and investment to higher taxes – either ending the exclusion of interest on state and local bonds, eliminating the tax deferral on long-term saving in life insurance products (often referred to as taxing the inside build-up); or changing the treatment of capital gains on assets that are transferred at death. Including new taxes on forms of saving as a way to pay for the Romney proposals seems to imply either that (a) the authors are ignoring Romney’s goal of promoting saving and investment, or (b) the authors believe that taxing a particular form of saving will not reduce saving and investment.
2. Effects of tax reform on economic growth
We included a discussion of how economic growth would affect our results and we showed that our central results do not change even after accounting for revenue feedback using growth effects taken from Mankiw and Weinzierl.
However, we are skeptical that the growth effects of tax rate cuts – whether taken in isolation or as part of a revenue-neutral package – are as large as is sometimes claimed. While a full debate about taxes and economic growth is beyond the scope of this paper, it is worth noting that estimates that do incorporate macro adjustments by the Congressional Budget Office (CBO) and JCT show smaller effects than Mankiw and Weinzierl. Recent papers by analysts at the Congressional Research Service also highlights a weak relation between lower tax rates and economic growth.
Moreover, history has not been as kind to the verdict that tax cuts generate economic growth as substantial as the campaign’s rhetoric would suggest. For example, the massive and permanent tax increases associated with World War II do not appear to have had an impact on U.S. economic growth rates. Likewise, the 1981 tax cuts, which cut the top rate from 70 percent to 50 percent, accounted for only a very small share of the growth of the economy between 1981 and 1986, according to Martin Feldstein, who advises the Romney campaign today and who was Chair of the Council of Economic Advisers under Reagan. Feldstein and now-CBO director Douglas Elmendorf analyzed the 1981 tax cut and concluded that:
“Our evidence contradicts the popular view that the [early 1980s] recovery was the result of a consumer boom financed by reductions in the personal income tax. We also find no support for the proposition that the recovery reflected an increase in the supply of labor induced by the reduction in personal marginal income tax rates. … The driving force behind the recovery of nominal demand was the shift to an expansionary monetary policy. The rapid response in nominal GDP can be explained by monetary policy without any reference to changes in fiscal and tax policy.”
The 1986 tax act is the most recent example of base-broadening, tax-rate reducing, revenue-neutral reform. In that act, the top personal income tax rate fell by 44 percent and the corporate rate fell by 26 percent. Auerbach and Slemrod document tepid economic growth responses to the 1986 tax act.
Most recently, the 1993 hike of 8.6 percentage points in the top income tax rate did not stop a remarkably strong decade of economic expansion and budget balancing from occurring. The 2001 tax cuts had no appreciable impact on growth during the rest of the decade – through most of the decade (before the Great Recession) growth occurred mainly in housing and finance, two sectors that were not favored by the cuts. Gale and Potter estimate that the deficit-increasing effects of the 2001 tax cut, which reduced long-term economic growth, outweighed the marginal-tax-rate-cut effects, which increased growth, so that the net impact of the tax cut on long-term growth was negative. It is difficult to translate this result, however, into an analysis of the effects of revenue-neutral reform without knowing how the base-broadeners needed to generate revenue-neutrality would affect incentives.
B. Discussion of Studies that build off of the TPC results
Three authors presented responses that start with the estimates we obtained, described above, and modify the results in different ways.
Jensen suggested that eliminating the exclusion of interest on state and local bonds and the exclusion of inside-buildup on life insurance vehicles could raise “upwards of $90 billion” in 2015. In response, we showed that an upper bound on the revenue obtained from eliminating these two exclusions is on the order of $49 billion ($29 billion for tax-exempt interest, $20 billion for life insurance), at least $4 billion of which would come from families with income under $200,000. Jensen obtained a much higher estimate by including corporate changes and by confusing the tax expenditure estimates for these items with revenue estimates of repealing the provisions. In fact, because taxpayers respond to such changes, the tax expenditure estimates are significantly larger than the revenue implications of repeal.
Dubay argues that the $86 billion shortfall could be made up by a combination of taxing municipal bond income, taxing the inside build-up in life insurance vehicles, ending the step-up of basis on assets with capital gains that are transferred at death, and incorporating the effects of economic growth.
The novel factor in this study relative to the others is the notion of ending basis step-up on capital gains transferred at death and replacing it with basis carry-forward. He uses the tax expenditure estimate from the fiscal year 2013 Federal Budget that suggests eliminating step up in basis would yield $19 billion. However, Dubay overestimates the revenue that could be collected from this policy, for two reasons. First, the tax expenditure figure he cites comes from a comparison of basis step-up and full taxation of capital gains at death. Replacing basis step-up with basis carryover (under which the tax basis does not get changed, but taxes are not due at the transfer at death – they are due only if and when the asset is sold) would generate significantly less revenue.
Second, the figure he cites is based on current-law projections of tax rates, which for long-term capital gains are more than 50 percent higher than they would be under Romney’s proposals (23.8 percent under current law compared to 15 percent under Romney’s proposals), thus inflating the tax expenditure estimates relative to what they would be under Romney’s proposals.
Brill argues that the $86 billion shortfall could be made up by a combination of taxing interest income from municipal bonds, taxing the inside buildup in life insurance, dropping Romney’s proposal to repeal health care reform from the baseline, and accounting for economic growth. Relative to the other responses, the novel aspect of Brill’s response is his choice to omit health care taxes from the baseline. This raises several issues.
First, Brill suggests that Romney’s tax proposals should be evaluated relative to a baseline that excludes revenues from the ACA taxes—revenues which are highly progressive.
We believe that modifying the baseline to reduce overall revenue and progressivity amounts to moving up the goal line to make the goals artificially easier to achieve. Describing a policy as “revenue neutral” or suggesting that high-income households will pay the same share of taxes implies a commonly accepted baseline. (It’s not “revenue neutral except for those revenues.”) Our analysis draws on the same assumptions for the current policy baseline used by the Congressional Budget Office (whose “current policy” baseline is called the “alternative fiscal scenario”) and by other groups (e.g., Bowles-Simpson and Domenici-Rivlin). The ACA taxes are scheduled to begin on January 1, 2013. Those taxes will thus be in effect when the next president is inaugurated and are rightly considered not only part of “current law” (which takes the law as written) but also “current policy” (which attempts to reflect the trajectory of existing policies—including the likely extension of many expiring tax cuts—and is the baseline we used to analyze Romney’s proposals). Like them or not, the ACA taxes are now part of existing tax policy, and repealing them would be a tax cut—one that particularly benefits high-income taxpayers. More generally, selectively omitting certain disadvantageous features of the baseline to make a policy look better undermines the stated goals of revenue neutrality and the credibility of an analysis.
Second, Brill also suggests that the budget effect of repealing ACA taxes “should be analyzed in the context of the repeal of the various other health care provisions.” Romney wants to repeal the health care reform act (ACA), which is approximately revenue-neutral and very progressive. ACA raises coverage for low- and middle-income households and raises taxes for higher-income households. Both individual elements are progressive. Repealing each provision would thus be unambiguously regressive. We are unable to include repeal of the coverage/spending changes in the tax model, which therefore leads us to understate the regressivity of Romney’s proposals in our model. Removing the revenue portions of Romney’s ACA-repeal proposal, as Brill would like to do, simply leads to a further understatement of the regressivity of Romney’s stated proposals.
From the perspective of households’ wallets, it is a matter of semantics whether ACA repeal is considered “tax policy” or “health policy.” (Similarly, whether curtailing the tax preference for health insurance is “tax reform” or “health reform” is a semantic issue not a substantive one.) What matters is that Romney proposed ACA repeal and doing so would be roughly revenue-neutral but would hurt middle- and low-income households and help high-income households. Hence, we believe that if there is a shortcoming of our analysis of Romney’s proposal to repeal ACA, it is not the inclusion of the revenue changes, but the inability (because we are using a tax model) to incorporate the spending/coverage changes. Had we been able to do so, our analysis would have shown it to be even more difficult to meet Romney’s stated goals.
Finally, for Brill’s plan (or Jensen’s or Dubay’s) to work, of course, one would have to eliminate all tax expenditures for people with income above $200,000 and this would require an enormous spike in the marginal tax rate at that income level. (We know this because Brill worked off of our analysis, and as noted above, our analysis was meant to provide an upper bound on the progressivity of the base broadeners, not an implementable version of tax reform. The same issue applies to Jensen’s and Dubay’s analysis.) That is, a household with income of $200,000 would receive all of its deductions, exclusions, exemptions, etc. But if the household members earned one more dollar, they would immediately lose all of their itemized deductions, their health insurance would be fully taxed, and so on. This would create a massive marginal tax rate at the $200,000 level. We believe it is fair to say that every sensible economist (and we include Brill