The argument for these proposals has nothing to do with the impact of higher tax rates on incentives and the economy. It is all about “fairness” — defined as reducing the top 1%‘s share of income.
This political exercise invariably begins by citing dubious statistics about pretax incomes among the top 1% (1.3 million tax returns) as an excuse for raising tax rates on the top 5%, among others. Echoing speeches from Sen. Clinton, Business Week recently exclaimed, “According to new Internal Revenue Service data announced last week, income inequality in the U.S. is at its worst since the 1920s (before the Great Depression). The top percentile of wealthy Americans earned 21.2% of all income in 2005, up from 19% in 2004.”
These statistics are extremely misleading.
First of all, the figures do not describe the top percentile’s share of “all income,” but that group’s share of “adjusted” gross income (AGI) reported on individual tax returns. For one thing, thousands of professionals and business owners who used to report most of their income under the corporate tax responded to lower individual income‐tax rates after 1986 and 2003 by reporting more income under the individual tax as partnerships, LLCs and Sub‐S corporations.
Peter Merrill of PricewaterhouseCoopers found that “since the Tax Reform Act of 1986 … the share of business income earned through pass‐through entities has increased by 75% from 29% in 1987 to 52% in 2004.” Business profits accounted for just 11.1% of the income reported by the top 1% in 1986, according to economists Thomas Piketty and Emmanuel Saez, but that business share leaped to 21.2% by 1988 and to 29.1% in 2005.
It is this bookkeeping shift, moving business income from the corporate to the individual tax, not CEO pay, which raised the top 1%‘s share on individual tax returns. Income reported on W2 forms — salaries, bonuses and exercised stock options — accounted for only 57.2% of total income among the top 1% in 2005, down from 63% in 2000 and 65.7% in 1986. Real compensation among the top 1% actually fell 7% from 2000 to 2005.
Turning to the denominator of this ratio (“all income”), a huge portion of middle and lower income is no longer reported on tax returns. A larger and larger share of middle‐class investment income is now accumulating outside of AGI because it is inside IRA, 401(k) and 529 savings plans.
The CBO reckons the top 1% accounted for more than 59% of all capital gains, interest, dividends and rent reported on individual tax returns by 2004. Yet estimates of the share of national wealth of the top 1% range from 21%-33%.
If the top 1% own 21%-33% of all capital, how could they be collecting 59% of the income from capital? They can’t and they aren’t. The top 1% is simply reporting a rising share of capital income because those with more modest incomes are keeping a rising share of their capital income unreported — in IRA, 401(k) and 529 accounts. Millions also shrink their “adjusted” incomes by subtracting contributions to IRAs unavailable to the rich.
Another huge swath of middle and lower income is excluded because AGI includes only the taxable portion of Social Security benefits and totally misses most other transfer payments such as Medicaid, food stamps and the Earned Income Credit. The Canberra Group, an international group of experts on income statistics brought together from 1996–2000 by the OECD, World Bank, U.N. and others, insisted household income must include everything that “increases the recipients’ potential to consume or save.” Government transfers amounted to $1.5 trillion in 2005 — more than the total income of the top 1% in the basic Piketty and Saez estimates ($1.2 trillion).
As a result of such huge omissions, and tax avoidance, the AGI of $7.5 trillion in 2005 was $3.7 trillion smaller than pretax personal income (personal income was $10.3 trillion in 2005, after subtracting $875 billion of payroll taxes). Anyone suggesting AGI is a more accurate measure than personal income is obliged to argue that GDP in 2005 was exaggerated by 29.4%.
Estimated income shares from the IRS or Messrs. Piketty and Saez are not about income per household, but income per tax return. That matters because the top fifth of households average two salaries per tax return. The Census Bureau reports that the top fifth accounted for 26.8% of all full‐time works last year while the bottom fifth accounted for just 5.7%. In fact, 64.5% of the households in the bottom fifth had nobody working, not even part time for a few weeks. When labor economists discuss income inequality, they habitually switch to speculating about skill‐based differences in hourly wages, totally ignoring differences in hours worked.
Third, the latest IRS figures are not comparable with those of 1986, much less with 1929, because the definition of AGI changes with changes in tax law. Such estimates differ greatly, with the IRS saying the top 1% received only 11.3% of income in 1986 (because AGI then excluded 60% of capital gains) while Messrs. Piketty and Saez put that year’s figure at 13.1% and the CBO says it was 14%.
The IRS figures only go back to 1986, so the Business Week comparison with the 1920s is invalid. The new figure is from the IRS but the old one is from Messrs. Piketty and Saez. Their recent estimates are also not comparable to their prewar estimates. Before 1944, their figures were obtained by dividing top income shares by 80% of personal income. Their estimates for 2005 were obtained by dividing top incomes by the $6.8 trillion left on tax returns after excluding even taxable transfer payments.
If total income for 2005 was defined as it was for 1928, then the share of the top 1% would have dropped to 13.3% in 2005, compared with 19.8% in 1928. Besides, as Messrs. Piketty and Saez explained, “our long‐run series are generally confined to top income and wealth shares and contain little information about bottom segments of the distribution.”
A fundamental problem with all tax‐based income data involves “taxable income elasticity.” Numerous studies, some by Mr. Saez, show that the amount of top income reported on individual tax returns is highly sensitive to changes in marginal tax rates on individual income, corporate income and capital gains. After the tax on dividends was reduced in 2003, for example, top‐bracket investors held more dividend‐paying stocks in taxable accounts (rather than in nontaxable accounts) and fewer tax‐exempt bonds.
When the top tax on capital gains was cut in 1997 and 2003, investors reacted by trading stocks more frequently and realizing more capital gains in taxable accounts. In the Piketty‐Saez data, capital gains accounted for only 10.8% of the top 1%‘s income from 1987 to 1996, when the capital gains tax was 28%. By contrast, capital gains accounted for 16.9% of the top 1%‘s reported income from 1997 through 2002, when the tax was down to 20%.
Even if estimates of the top 1%‘s income share were not so sensitive to changes in tax rates, they would still tell us nothing about what happened to incomes among the other 99%. The top 1%‘s share always falls in recessions, even aside from capital gains. But that certainly doesn’t mean recessions raise everyone else’s income.
“It is a disputed question,” wrote Messrs. Piketty and Saez, “whether the surge in reported top incomes has been caused by the reduction in taxation at the top through behavioral responses.” In fact, their data clearly suggest that higher tax rates on top incomes, dividends and capital gains would sharply reduce top incomes, dividends and capital gains reported on individual tax returns. Such behavioral responses would have little impact on actual income or wealth at the top, while nonetheless leaving middle‐income taxpayers stuck with a much larger share of the tax burden.