Every contest for the presidency is also a contest among interest groups, think tanks, and journalists eager to gain attention and influence public policy by persuading candidates to adopt their agenda as a campaign theme.
Income inequality has long been a favorite rhetorical device to promote such disparate policies as tariffs, immigration restrictions, or subsidies to builders of low‐income housing. In the current election cycle, the mantra of “rising inequality” is incessantly used as a rationale for punitive tax policies toward high‐income taxpayers and even middle‐income investors — proposals rarely defended on their economic merits. “Fairness” arguments often seem to drown out serious debate about the potential impact of higher marginal tax rates on economic efficiency, incentives, tax avoidance or economic growth. This campaign for higher tax rates on upper incomes invariably relies on measures of incomes among the “top 1%” as reported on individual income tax returns (a strange average of incomes ranging from about $350,000 to $3.5 billion).
Telling this story always involves picking two dates very carefully, as if that described an ongoing trend.
The Old Two‐Year Ruse (start with 1979)
In June 2007, the Pew Charitable Trust’s “Mobility Project” was launched with a pamphlet written by project director John E. Morton, former foreign policy adviser to Sen. John Kerry’s presidential campaign, and Isabel Sawhill, associate director of the Office of Management and Budget under President Bill Clinton. The first sentence noted that “the convergence of a presidential election cycle” and “income inequalities last seen nearly a century ago” provide “a unique opportunity to refocus the debate.”
The document claimed that the United States “is a society with rapidly growing income inequality,” noting that “the Congressional Budget Office [CBO] finds that between 1979 and 2004, the real aftertax income of the poorest one‐fifth of Americans rose by 9 percent, that of the richest one‐fifth by 69 percent, and that of the top 1 percent by 176 percent.” Yet “rapidly growing” surely implies inequality is growing rapidly in the present, not the past. Comparing 1979 and 2004 shows that something happened between those two dates, but not what happened or when. If we needed only two years to define a trend, then the same CBO estimates would show that the share of income for the top 1% was 12% in 1988 and 12.2% in 2003. It is no accident that the Mobility Project started with 1979.
One thing that happened between 1979 and 1988 is that marginal tax rates on individuals stopped being higher than tax rates on corporations. The textbook Taxes and Business Strategy, by Myron Scholes and others, shows that “pre‐1981 … many doctors, lawyers and consultants incorporated to escape the high personal tax rate.” After the highest individual tax rate fell from 70% to 28%, they added, “many of the corporations converted back” to partnerships, LLCs, and Subchapter‑S corporations.
Shifting income from the corporate tax to the individual tax created an illusory increase in top incomes in studies by the CBO and by economists Thomas Piketty and Emmanuel Saez in their 2003 study “Income Inequality in the United States, 1913 – 1998,” which is regularly updated on Saez’s website. Business income was only 11.1% of the reported income of the top 1% in 1986, according to Piketty and Saez, but that fraction nearly doubled in only two years to 21.2%. The unusually rapid increase in reported top incomes between 1979 and 1988 largely reflects increased incentives to earn more income in taxable cash (rather than perks), and to report that income on individual (rather than corporate) tax returns. But it also reflects the fact that inflation fell from 13.3% at the end of 1979 to 4.4% by 1988, greatly increasing the value of bonds and stocks. That, in turn, added to capital gains and stock option payoffs in the CBO estimates of top incomes‐particularly in 1986, when there were huge sales of appreciated assets to avoid a higher tax in 1987. To now look back on 1979 as an ideal example of low inequality is ironic, because the runaway inflation of 1979 and the subsequent three‐year recession hurt rich and poor alike.
Using the same CBO figures, if the income gains are measured from 1988 instead of 1979, the increase in the income of the top 1% drops from 176% to 47%, the increase for the top fifth drops from 69% to 31%, and the increase among the bottom fifth rises from 9% to 16%. In other words, nearly all of the “rapidly rising” inequality happened between 1979 and 1988 ‑a period that ended 20 years ago. Some rise in inequality still remains in this particular set of data, to be sure, even after eliminating the exaggeration resulting from using 1979 as the base year. Yet other measures of income show a quite different pattern of gains, particularly for low‐income households.
Table 1 shows two estimates of the changes in before‐tax income gains for the poorest, richest, and middle fifths of households. The first column is from a May 2007 CBO study, “Changes in the Economic Resources of Low‐Income Households with Children,” and it begins with the 1991 recession and ends with 2005 (http://www.cbo. gov/ftpdocs/81xx/doc8113/05 – 16‐Low‐ Income.pdf). This report was based on Census Bureau income plus transfer payments, notably the earned income tax credit (EITC). The second column covers all households, using the Federal Reserve’s triennial Survey of Consumer Finances from the cyclical peak of 1989 to 2004 (https://www.federalreserve.gov/pubs/oss/ oss2/2004/bulletin.tables.pub.xls ). The SCF uses a broader measure of income than the usual Census Bureau definition, including realized capital gains, business income, and transfer payments.
Estimated Growth of Real Income
Both surveys show relatively large increases in real, inflation‐adjusted income among both the poorest and richest fifths, with smaller gains in the middle. The percentage gains for the top and bottom fifths appear much larger in the CBO figures partly because that study begins with the 1991 recession, which exaggerates cyclical change. In the SCF figures, the 1989 – 2004 income gains among the poorest fifth (23.4%) and second‐poorest (20.9%) exceeded the gains among the top 10% (18.3%) , which does not meet any definition of rising inequality.
The Pew report, as well as sensational news stories this year in The New York Times (April 9) and the Wall Street Journal (April 19), suggested there was something unusually bad about the fact that real median household income in 2006 had not yet gotten back to the peak level of 2000. Yet in the last business cycle median income was also well below the 1989 peak in 1995, after six years, and only $34 higher in 1996 (despite cheap oil). Moreover, the relatively slow growth of middle income in CBO and SCF data shows why the similar measure of median income is not a reliable approximation of how real income in general has grown. In both studies, 80% of the population experience more rapid real income growth than the middle did.
A New Two‐Year Ruse(start with 2002)
Those who cite rising inequality as the reason for undoing the most economically significant tax rate reductions of 2000- 2003 must realize they need evidence about what happened since 2000, not since 1979. Lacking the required proof that the “Bush tax cuts” resulted in a dramatic increase in after‐tax inequality since 2000 (and therefore should be repealed), proponents of higher tax rates resorted to misleading statements about before‐tax incomes since 2002.
Writing in the May 1 issue of Time, Justin Fox said, “According to economists Thomas Piketty and Emmanuel Saez, 75% of all income gains from 2002 to ’06 went to the top 1%-households making more than $382,600 a year.” Noting with approval that “Obama in particular has been explicit about wanting to shift more of the income‐tax burden … onto those making more than $200,000 a year,” Fox blamed inequality on President Reagan’s tax cuts and cited a recent poll showing 51% favor “heavy taxes on the rich.” Yet his figures, from Piketty and Saez, refer to income before taxes.
Citing the same Piketty and Saez calculations, Wall Street Journal columnist David Wessel added “the trend didn’t begin with President Bush’s election, but he didn’t do much to arrest it.” What could he have done? Raising tax rates on the rich or transfers to the poor could not change these numbers, because Piketty and Saez exclude taxes and transfers. “There is significant disagreement among politicians and voters about … how much the tax code should redistribute income,” wrote Wessel, noting that “Sen. Obama would wield the tax code more aggressively than Senator McCain.” This alleged policy debate relies on statistics that exclude taxes in order to rationalize the use of higher tax rates to reduce high incomes. That makes no sense.
The statistics were lifted uncritically from a March 15 memo by Emmanuel Saez of the University of California at Berkeley, “Striking It Richer.” Saez compares the 1993 – 2000 Clinton expansion with the Bush 2002 – 2006 expansion, concluding that “during both expansions, the incomes of the top 1% grew extremely quickly at an annual rate over 10.1 and 11.0% respectively.” But that is only because he starts with 2002, after stock prices had collapsed, and includes capital gains. Realized capital gains fell from 6% of GDP in 2000 to 2.4% in 2002, and then recovered to 5.2% in 2006.
The first column of Table 2 shows the Piketty and Saez estimates of average (mean) income for the top 1%, measured in 2006 dollars and including capital gains. These figures supposedly show that real incomes of the top 1% have been growing very rapidly under Bush, just as they did during the tech stock boom of 1997 – 2000.. In reality, top incomes doubled from 1993 to 2000, but then fell 31% by 2002. For the remaining 99%, by contrast, real income fell by only 4% from 2000 to 2002. Saez claims incomes of the top 1% continued to rise “extremely quickly” after the Clinton years, both in absolute terms and also relative to other incomes. For the whole period from 2000 to 2006, however, his figures show that real income of the top 1% rose by 9% (from $1.185 trillion to $1.243 trillion), while the real income of everyone else rose by 9.8% (from $6.22 trillion to $6.83 trillion).
Average Before‐Tax Income of the Top 1%(in 2006 Dollars, Including Capital Gains)
The second column shows the declining share of top incomes received from salary, bonuses, and stock options. Saez, however, says, “A significant fraction of the surge in top incomes since 1970 is due to an explosion of top wages and salaries.” That was true from 1966 to 1988, and from 1994 to 2000, but not since then. His data show that real labor income of the top 1% was flat from 1988 to 1994, rose 65.7% from 1994 to 2000, and then fell 8.4% from 2000 to 2006.
The relatively small increase in top 1% incomes from 2000 to 2006 is mainly because of income shifting‐the business share of the top 1% of income rose from 27.4% in 2002 to 30.1% in 2006, after individual and corporate tax rates became the same. Lower tax rates on dividends also raised taxable top incomes. Dividends accounted for only 4.2% of top percentile incomes in 2002 (aside from capital gains), but 7.4% in 2006. When capital gains were included, they accounted for 5.8% of the income of the top 1% in 2000, 2.3% in 2002, and 4.4% in 2006. It is not difficult to imagine why Saez chose to include capital gains and to measure top income gains from 2002 rather than from 2000.
Imagine There’s No Taxes…
The Brookings Institution began a new policy program in 2006, the Hamilton Project‐founded by former Treasury secretary Robert Rubin. The first director was Peter Orszag, now head of the Congressional Budget Office. The project’s seminal statement on tax policy, “Achieving Progressive Tax Reform,” was coauthored in June 2007 by former Treasury secretary Larry Summers; Jason Furman, now economic director for the Obama campaign; and Jason Bordoff, the new project director.
The paper acknowledged that “excessively high tax rates distort economic behavior by changing the incentives to work, save, and invest, which can harm economic performance.” Yet the key theme was to assert that “rising inequality strengthens the case for progressivity.” Proof of such rising inequality again relied on CBO estimates of income changes since 1979. Like Saez, Summers, Furman, and Bordoff emphasized before‐tax data: “In 1979,” they wrote, “the before‐tax income of the most affluent 1 percent of the U.S. population already equaled that of the bottom 26 percent. That share has since risen nearly continuously, reaching 45 percent in 2004.”
Even if those figures were credible, they are irrelevant to the policies being prescribed. It is clearly illogical to point to income before taxes to suggest that the rich do not pay enough taxes. It is also illogical to point to income before transfer payments to prove that the poor do not receive enough transfer payments. The before‐tax CBO estimates exclude the refundable earned income tax credit (EITC). The authors’ estimates also rely on Piketty and Saez, who exclude all transfer payments. If transfer payments are excluded, the Census Bureau calculates that incomes of the poorest fifth in 2006 averaged only $5,208 but their disposable income‐including transfers‐was $12,329. Similarly, the beforetax income of the richest fifth averaged $182,505, but their disposable income — after taxes — was $137,293.
The Hamilton Project paper claims that “the tax system itself has become considerably less progressive. Reductions in taxes have been particularly dramatic for very high income taxpayers.” That comment cannot refer to individual income taxes, as presidential policy debates assume. CBO estimates the individual income tax rate among the top 1% fell modestly from 21.8% in 1979 (when the top tax rate was 70%) to 19.9% in 1989 (when the top tax rate was 28%) and 19.4% in 2005 (when the top tax rate was 35%).
By contrast, reductions in taxes have been “particularly dramatic” for the middle fifth, whose average tax was slashed from 7.5% in 1979 to 5% in 2000 and 3% in 2005. For the poorest fifth, the tax rate fell to minus 1.6% in 1989, minus 4.6% in 2000, and minus 6.5% in 2005‐reflecting the expansion of refundable tax credits. In the same paper, Summers, Furman, and Bordoff argue that the progressive 1986 – 2003 reductions in income tax for the bottom 40% have now made it “regressive” to reduce anyone else’s income tax: “The bottom 60 percent of households pay less than 1 percent of total income taxes,” they write, so “any income tax cuts that do not include expansions in refundable credits such as the EITC are therefore necessarily regressive.” Making that comment even more ironic, the authors exclude the EITC when adding‐up incomes of the bottom 40%.
In reality, the authors’ claim that taxes have become “less progressive” cannot and does not refer to the reduction in individual income tax rates. Instead, it mainly refers to a CBO estimate that the effective corporate rate on the top 1% fell from 13.8% in 1970 to 6.7% in 2000 — before rising to 9.9% in 2005. That timing makes it difficult to blame the 2001 – 2003 tax cuts for making taxes “less progressive.” Besides, the CBO’s technique for distributing the corporate tax is seriously flawed.
Adding Two‐Thirds of Corporate Tax to Top Incomes
Believe it or not, the CBO adds twothirds of corporate taxes to the before‐tax income of households of the top 1%. They do that to estimate what share of that tax is borne by the top 1%. Yet adding most corporate taxes to top incomes makes beforetax CBO figures a particularly untenable way to measure changes in top incomes. CBO economists reason that the corporate tax is borne by owners of capital in general. Unfortunately, they then make an indefensible leap by estimating the ownership of capital by looking at only the dwindling portion of capital gains, dividends, interest income and rent still reported on tax returns. As a result, the estimated share of corporate taxes added to top 1% incomes rose from 34% in 1979 to 66.4% in 2004. And that, in turns, accounts for a sizable portion of that 1979 – 2004 increase in top “incomes” (including corporate taxes) used to defend the Hamilton Project’s uneasy case for steeper marginal tax rates on incomes well below the top 1%.
The top 1% could not possibly have received 66.4% of the nation’s investment returns in 2004. Their share of wealth was 33.4% according to the SCF, and closer to 21% according a study co‐authored by Saez. Neither study finds any upward trend in wealth inequality, so assigning 66.4% of capital income to the top 1% is literally unbelievable. The only reason the top 1% accounts for a rising share of taxable savings is that a rapidly increasing share of everyone else’s savings is now sheltered in tax‐free retirement plans. Returns on those investments will never be reported to the IRS as dividends, interest, or capital gains, because distributions from deferred plans are reported as ordinary income while capital gains or dividends from Roth IRAs are never reported. Most capital gains on home sales also vanished from tax returns since 1997.
The CBO’s statistical blunder of using taxable investment returns to guess the share of corporate taxes for the top 1% resulted in nearly $150 billion being added to the “before‐tax income of the most affluent 1 percent” in 2004. That huge miscalculation, in turn, was critical to the Hamilton Project’s unexplained assertion that “rising inequality strengthens the case for progressivity.”
The Brookings‐Urban Tax Policy Center also uses the CBO’s erroneous method to estimate how the corporate tax is distributed. As a result, they estimate that top 1% would receive two‐thirds of the benefit from Senator McCain’s proposal to cut the corporate income tax. Such estimates are not credible.
Unpaid Taxes Don’t Pay Bills or Redistribute Income
The fundamental problem with trying to measure income from individual tax returns is that people try to minimize their taxes. Many studies find that an increase in marginal tax rates on capital gains or high incomes causes income reported from those sources to drop significantly. Economists call this “the elasticity of taxable income.”
If the tax rate on dividends is increased, investors will shift many of their dividendpaying stocks into tax‐exempt pension funds or sell them to tax‐exempt domestic and foreign investors and hold more taxexempt bonds. If the tax rate on capital gains is increased, investors will hold fewer shares in taxable accounts and avoid selling winning stocks unless they have offsetting losses. If the highest individual income is raised far above the corporate tax rate, thousands of professionals and businesses currently filing as partnerships, Subchapter S corporations, and LLCs will reincorporate to shelter income under the corporate income tax.
Because of such well‐documented responses, income tax data provide incorrect information about income distribution when tax rates change. A large body of evidence finds large and sustained increases in reported income among high‐income taxpayers in the wake of major reductions in tax rates on salaries (1988) or capital gains (1997) or both (2003). It follows that we could expect sustained reductions in reported income among high‐income taxpayers if there were major increases in marginal tax rates on high individual incomes and/or capital gains. Such defensive moves by targeted taxpayers would indeed appear to reduce before‐tax top incomes reported on individual tax returns. But it also means the higher tax rates would be largely or entirely ineffective in reducing after‐tax income or in raising additional revenue.
Dubious estimates of relative income changes since 1979 or 2002, usually constructed from before‐tax IRS data, are being widely cited as a sufficient justification for embracing tax policies that are undeniably harmful to economic progress and prosperity. Bad statistics are never a good excuse for advocating bad policies.