A new book by economists Emmanuel Saez and Gabriel Zucman, “The Triumph of Injustice,” makes the case that economic inequality in the U.S. has worsened significantly over the last few decades. They explain that the share of total national income earned by the top has greatly increased, while their effective tax burden has dropped. As a result, the top 400 richest people in the country faced an effective tax rate in 2018 that was lower than that of the bottom 50% of the population.
This result rests on several unconventional methodological assumptions, which, in our opinion, biases the outcome towards finding relatively lower tax rates on the wealthy compared to the poor and overstates the decline in top earners’ tax burdens. We discuss these in detail below. However, very broadly, a big reason for their finding is the trend decline in corporate income taxation, and the role of the tax and transfer system in addressing inequality and poverty.
It is true that the taxation of capital income, a primary form of income for the wealthy, has declined over the last several decades, and this is a trend that has gained prominence historically. Yet, there are good economic reasons for why this has happened, not just in the U.S., but also worldwide. One reason for this trend is that capital income taxation, and in particular, corporate taxation, has been shown to negatively affect capital formation and worker wages. Therefore, when countries aim to attract capital, they have often done so through cuts in the corporate rate. Naturally, if we assume, as Saez-Zucman do, that this reduction in corporate taxes is a cut in tax rates for rich shareholders only, then this will show up as a decline in their tax rates over time. But it’s important to understand that the primary reason for engaging in these rate reductions is that countries are competing for globally mobile capital, which they believe will result in higher investment and higher levels of economic growth.
Second, it is also true that personal income tax rates have come down over time in the U.S., and that precisely in an effort to reduce tax burdens on lower income households, we have adopted (refundable) tax credit programs like the earned income tax credit (EITC) and the child tax credit (CTC), that not only result in tax savings for poorer households but directly transfer cash income to them. By not accounting for these transfers, Saez and Zucman overstate tax rates on poor households and make them appear relatively worse off compared to richer households.
In the next section, we provide a detailed assessment of the main assumptions that we think are driving their result.
METHODOLOGY AND ASSUMPTIONS
Most of the arguments in their book are based off their “distributional national accounts” (DINA). DINA takes 100% of national income (the total amount of income earned by the U.S. in a year) and total taxes collected by the federal, state, and local governments and allocates it to individuals. Their goal is to measure how much of national income each income group is earning and what taxes they are paying in any given year and how that has changed over time.
The task of distributing all income and taxes to individuals in the U.S. is a useful, but difficult task. We know how much income people report to the Internal Revenue Service for tax purposes because that data exists, but that doesn’t include all national income. Tax data exclude things like retained earnings of corporations, imputed rental income, non-taxable retirement benefits, and their share of income used to pay excise taxes and other taxes on production. In addition, we may know who remits a tax, but that is seldom who ultimately bears the burden of the tax. Thus, taxes such as the corporate income tax, sales taxes, and excise taxes similarly need to be imputed for individuals, which requires making assumptions about who bears the tax.
Ultimately, the results of this analysis are driven by the underlying assumptions. As we show below, the assumptions made in the book are at odds with decades of research in public finance about the economic incidence of such taxes.
The Analysis Excludes the Effect of Transfers and the Refundable Portion of Tax Credits
An important feature of Saez and Zucman’s analysis is that it excludes the additional income low-income households receive in transfer payments, including the benefit of the refundable portion of tax credits such as the Earned Income Tax Credit (EITC) and the Child Tax Credit (CTC). As mentioned above, Saez and Zucman distribute total national income each year. National income excludes the impact of transfers because they represent money shifting from one household to another with no corresponding production. In addition, their definition of tax (mostly) conforms to the definition used by the Bureau of Economic Analysis’s National Income and Product Accounts (NIPA). An important result of this is that income tax liability cannot go below zero, which excludes the impact of the refundable portions of tax credits.
Excluding transfers from income and the refundable portion of tax credits diverges significantly from convention. None of the major groups that produce distributional analyses of taxes excludes the refundable portion of tax credits from their distributional tables. In addition, more conventional methods employed by the Treasury’s Office of Tax Analysis (OTA), Joint Committee on Taxation (JCT), and the Tax Policy Center (TPC) tend to use broad definitions of income that include not just income related to production (wages, salaries, capital income), but also many benefits from the government, including Social Security benefits. A broad definition of income somewhat better reflects the resources available to households to pay taxes than national income. This is especially the case for low-income households, which receive about 40% of their income from transfers, according to the Tax Policy Center. Without these sources of income and the benefit from the refundable portion of tax credits, effective tax rates are overstated.
More broadly, the choice of excluding all transfers, including the refundable portion of tax credits makes little sense. First, in evaluating income inequality, the role of the transfer system is hugely important. Data on the EITC and the CTC show that combined these programs lifted approximately 9.1 million Americans out of poverty in 2018. Estimates of poverty and income inequality typically account for the role that these programs, as well as food stamps and other benefit programs, play in addressing post-tax and transfer incomes. For instance, the CBO shows that accounting for the entirety of the federal tax and transfer system significantly reduces inequality. The Gini coefficient, a measure of the income distribution, is reduced from 0.51 before taxes and transfers, to 0.42 after taxes and transfers. According to other analyses, economic security programs cut the poverty rate from 25 to 13.9% in 2017.
Others have shown that if you were to at least include the refundable portion of tax credits, the overall tax system would be much more progressive than Saez and Zucman estimate. While data from the CBO show that the federal tax system is progressive, data from ITEP shows that sales and local taxes tend to be more regressive. Even so, combining these data (and including transfers) show that average tax rates are nearly 40% for the richest households and just over 10% for the poorest households—a clearly progressive system. Therefore, it is misleading to ignore these impactful transfers in the analysis. Ignoring transfers is a big reason why tax rates at the top are lower than those for lower income households.
Their treatment of sales and excise taxes also overstates the tax burden on low-income individuals. Saez and Zucman’s method distributes 70% of the sales tax to consumers based on current-year consumption and 30% of the sales tax to labor and capital income (minus saving). Placing this much weight on the amount a taxpayer pays in sales tax in a given year can either greatly overstate or greatly understate their effective sales tax rate. This is because a taxpayer in peak earning years may consume a small share of their income, and pay little sales tax, as they save for retirement. However, this same taxpayer may end up paying significant sales tax burden as a share of income while in retirement, evening things out.
A more consistent accounting of sales taxes would allocate the burden based on taxpayer income. This method seems to align more closely to Saez and Zucman’s methodology in their 2018 paper, which arguably better accounts for the ultimate economic burden of the sales tax and is more consistent with how they distribute other taxes, such as payroll taxes and the corporate income.
Comparing their sales tax distribution in their book to their previous published data shows how much this matters. In the book, the bottom 50% of taxpayers face a sales tax burden above 8% while the top 1% face an effective tax rate of 2.3%. In their previously published data, which uses a more appropriate method for distributing the sales tax, this translates to a 5% effective sales tax burden on the bottom 50% and a 4% effective tax burden on the top 1%.
Interaction between sales tax and refundable credits
Their method for distributing the sales tax, while excluding transfer payments, also results in extreme and counter-intuitive results. In their method, individuals pay sales tax on goods they consume currently. That current consumption is funded not just through wages and capital income, but also transfer income. However, since transfer income is not considered income in their methodology, individuals that may only earn transfer income end up having a positive tax liability, but no income.
This has two effects. First, this results in infinite effective tax burdens for some taxpayers. The possibility of infinite effective tax burdens isn’t just a hypothetical result of their method. According to Saez and Zucman “at the very bottom of the distribution, people do not earn labor, capital, or pension income but only transfer income.” As a result, they limit their population of individuals to those earning more than half the minimum wage to prevent showing these extreme tax burdens at the bottom.
Second, refundable tax credits actually increase tax burdens at the bottom. In their method, the EITC and CTC would not increase income, or reduce income tax liability, but it would increase current consumption. This increase in current consumption would result in higher sales and excise tax burdens.
Overstatement of income at top
In addition to overstating the tax burden on low-income individuals, Saez and Zucman also make methodological choices that understate the tax burden on high-income individuals. When imputing income to individuals, Saez and Zucman tend to attribute more income to very high-income individuals than other researchers. For example, there is around $1 trillion in business income that is counted as national income but is not reported on tax returns. Saez and Zucman choose to allocate this income in proportion to reported business income, imputing more income to very high-income individuals. Given any level of taxes, this drives down effective rates more for high-income individuals.
Researchers Gerald Auten and David Splinter, who have also done the same exercise with the same data, show that differences in how Saez and Zucman allocate unreported income, retirement income, and the overall definition of the income reduce Saez and Zucman’s effective tax rate on the top 0.01% by nearly 20 percentage points and raise the effective tax rate on the bottom 50% by 10 percentage points, relative to Auten and Splinters findings for 2014. Further, as economist Wojtek Kopchuk points out, in comparing Saez and Zucman’s earlier paper published in Quarterly Journal of Economics, with their current book, the average effective tax rates on the top 0.01% look very different. Between 1962 and 2014, the QJE series shows a drop of 3.3%, while their recent book series shows a drop of 18.7 percentage points.
Overstatement of decline in tax burden
Lastly, Saez and Zucman distribute the corporate tax in a nonstandard way that overstates the decline in effective tax rates for the very wealthiest householders. Saez and Zucman make the case that reductions in the corporate tax rate and changes to provisions like expensing that provide a boost to capital are 100% favorable for rich shareholders by assuming that their entire incidence is on owners of corporate capital. This is a deviation from the literature (and their previous work). An emerging literature has shown that corporate taxes are not only borne by all owners of capital (corporate and noncorporate), also borne to some extent by workers as well. Government agencies like the JCT and CBO regularly allocate about 18-25% of corporate tax incidence on workers. The Saez-Zucman estimate of a 100% incidence on corporate shareholders is hard to reconcile with any current analysis of these tax burdens.
While the choice to allocate 100% of the corporate tax to corporate shareholders suggests a more progressive corporate income tax than other estimates, it also shows a more remarkable decline in the tax rate of the top. From 1950 to 2018, the burden of the corporate tax fell from 5.9% of GDP to 1% of GDP. According to their data, this reduced the effective tax rate of the top 0.1% by 27.7 percentage points over this period. Contrast this with their previous estimates which distributed the corporate tax to all capital. That estimate shows a seven-percentage point drop in the burden of the corporate tax for the top 0.1%.
Hence, the fact that corporate tax rates have been declining globally, not just in the U.S., combined with the assumption that 100% of this reduction is a reduction in tax rates on owners of corporate capital, naturally leads to the conclusion that effective tax rates for rich owners and shareholders have declined over the last several decades. But what this analysis neglects to mention is that there are sound economic reasons for why countries have cut rates. The public finance literature shows that corporate tax rates matter for investment decisions. And competition for this investment has been the reason that countries have cut corporate rates.
Saez and Zucman argue that they make a moral case for addressing inequality by focusing on the “world as it should be,” as they see it. They moralize that companies are taking advantage of tax avoidance opportunities, and they shouldn’t be allowed to. They argue that corporate rate cutting by countries in an effort to attract capital is a bad outcome, and instead countries should coordinate to tax companies at high rates. They argue against the existence of tax havens. They argue that seemingly profitable companies like Google and Amazon should pay their fair share of taxes. All of these arguments are valid, but are not new.
In ignoring the practical realities of the world as it is, however, they also dismiss the more practical efforts being made to address these challenges. The OECD’s Base Erosion and Profit Shifting Project is a massive undertaking that aims to address the issue of non-taxation of business profits. There have also been smart ideas to address the issue that unrealized capital gains should not remain untaxed, through switching to carryover taxation rather than step-up in basis at the time property is passed onto heirs. There have been proposals to raise estate taxes and broaden the base for capital taxation. And economists have also talked about how a progressive consumption tax would help address inequality in a better way than the current system of taxation.
But instead, Saez and Zucman put forth the proposal for a wealth tax to combat inequality as well as the political influence of the rich. As several economists have stated, there are numerous problems with a wealth tax. First, even if we tax the wealthy using such a tax, it is hard to make the argument that this would reduce their political power, since we are talking about the super wealthy with billions of dollars in assets. Second, not only is such a tax difficult to administer, the fact that we cannot properly value assets that are not marketable, leaves the proposal subject to even more tax avoidance and evasion opportunities than the current income tax. Economists Lawrence Summers and Natasha Sarin point out, the revenue estimates from Saez and Zucman are wildly optimistic, and they will likely only collect one-eighth of what they project.
Finally, the people that the wealth tax would hit hardest are the most successful entrepreneurs and business owners—people who have heavily invested in companies that are now profitable. We need to worry about how such high taxes would impact their decisions to save, invest, and hire workers. It is important to remember than even a small 2-3% annual wealth tax is equivalent to very high income tax rates of about 40-60% over a decade because the same asset is taxed every year. Why not learn from the experience of other countries who have tried and since repealed wealth taxes, and aim for something better and perhaps more practical, such as reforming capital gains taxation or extending the estate tax?
The Triumph of Injustice is provocative reading, but makes less than a compelling base argument. The real injustice is its lack of use of standard assumptions and methodologies because it seems that it is these unconventional assumptions that drive their particular result. Economic discourse on the issue of inequality is important, as is debate on data and methodologies that aim to assess the state of inequality in the world, and policies to address it. We hope that researchers analyze the data, question the findings and assumptions, and continue to focus on the need to adopt sensible policies that address wealth and income inequality and economic mobility.
This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.
Aparna Mathur (@aparnamath) is a Resident Scholar in Economic Policy Studies at the American Enterprise Institute. Kyle Pomerleau (@kpomerleau) is the Chief Economist at the Tax Foundation.