Two French economists, Thomas Piketty and Emmanuel Saez, can count on a flood of publicity every time they release a new estimate of the share of U.S. income supposedly received by the top 1 percent. Even veteran Washington Post columnist Robert Samuelson approached their latest "astonishing" estimates as unquestionable scripture. "The biggest gains occurred among the richest 1 percent," he exclaimed. "Their share of pretax income has gradually climbed from 8 percent in 1980 to 17 percent in 2005."

Gradually? On the contrary, half of that increase happened in just two years, 1987 and 1988. The top 1 percent's share (of what?) was 13.2 percent in 1988, 14.9 percent in 2003.

To calculate the top 1 percent's share of total income, we need a definition of total income. For postwar data, Piketty and Saez use a modified version of adjusted gross income (AGI). Unfortunately, the Bureau of Economic Analysis calculates that AGI is not even a good measure of AGI -- it was missing $1.1 trillion dollars in 2004, called the "AGI Gap." It is also missing income of non-filers, estimated at $479 billion in 2000.

Transfer payments of $1.5 trillion are arbitrarily excluded, too. Benefits from private pensions qualify as "market income," yet benefits from Social Security do not.

The famed Canberra Group of experts insisted that household income must include cash transfers, food stamps and everything else that "increases the recipient's potential to consume or save." Most or all transfers are included in every official measure of pretax household income from the Congressional Budget Office, Bureau of Economic Analysis, Census Bureau, Bureau of Labor Statistics and the Fed. My family will collect more than $3,000 a month from Social Security next year, but Piketty and Saez say that's not income (the IRS disagrees).

In the American Economic Review last May, Piketty and Saez explained that their top 1 percent figures for other countries "are obtained by dividing top income shares by personal income." Their U.S. figures for 2005, however, are obtained by dividing top income shares by "market income" of $6.8 trillion -- a figure 38 percent smaller than pretax personal income. Income of the top 1 percent ($1.2 trillion) was 10.8 percent of pretax personal income ($11.1 trillion).

Everyone imagines the increase in the top 1 percent share, from about 13 percent in 1988 to 17 percent in 2005, must reflect the lavish salaries of a few thousand CEOs and celebrities. Samuelson tells us, "There were about 18,000 lawyers, 15,000 corporate executives, 33,000 investment bankers (including hedge fund managers, venture capitalists and private-equity investors) and 2,000 athletes who made roughly $500,000 or more in 2004." But that totals 68,000 -- less than 5 percent of the 1.4 million in the top 1 percent.

Such lists of a few high salaries illustrate a pervasive fallacy that the top 1 percent's reported income has been driven up by labor income -- salaries, bonuses and "nonqualified" stock options reported on W2 forms. On the contrary, labor earnings fell from 66 percent of all income reported by the top 1 percent in 1986 to only 57 percent by 2005. Investment income dropped from 23 percent to 14 percent of top 1 percent income over the same period.

Increases in the top 1 percent's income after the 1986 Tax Reform came from more business income being reported on individual tax returns, rather than corporate tax returns. The share of the top 1 percent income coming from business profits jumped from 11 percent in 1986 to 21 percent in 1988, and continued rising to 27 percent in 1994, the year after individual tax rates were increased.

The business share was still 27 percent in 2002, but it rose to more than 29 percent in 2005 after individual tax rates were once again reduced. How and why that happened is a textbook example of why tax return data cannot be used to measure income distribution. And the example is not just in my own textbook, "Income and Wealth." Continued...