The concept of "income distribution" is fundamentally muddled. Aside from government subsidies and transfer payments, income is not "distributed" at all. Most income is either earned or stolen. If some group's income was earned by legitimate means, then it is their income, not "ours." To complain that 10 percent earned too large a percentage of our income is to forget that they actually earned 100 percent of their own income.

 Unfortunately, the eternal ambition of Robin Hood economics is to steal money from those who earned it and "redistribute" it to those with more political clout. When in pursuit of such a worthy cause, it appears quite respectable to torture innocent statistics. Those deploying statistics in this campaign take special care to select their favorites.

 Washington Post columnist Steven Pearlstein recently confessed: "My favorite statistic comes from a recent study of tax returns by the Internal Revenue Service. It shows that, in 1979, the top 10 percent of households earned 33 percent of all pretax income. By 2003, their share had climbed to 44 percent. The shares of everyone else declined."

 Where did those numbers come from? They certainly didn't come from the Census Bureau, whose surveys show that the top 20 percent of households earned 49.8 percent of all pretax income in 2003.

 The Census Bureau doesn't add taxable capital gains to income, but the Congressional Budget Office does. Yet Pearlstein's statistics obviously didn't come from the CBO, either. The CBO estimates that in 1979 the top 10 percent of households earned 39.3 percent of all pretax income. By 2003, their share had dropped to 38.3 percent (or 33.7 percent after taxes). The shares of everyone else increased.

 It is easy to see why the CBO is not Pearlstein's favorite source of income statistics. It used to be Paul Krugman's favorite in 1992, but he has since switched to a 2001 study by two French economists, Thomas Piketty and Emmanuel Saez. Yet Piketty and Saez cannot be Pearlstein's favorite at the moment, because they showed that "the labor share (of national income) has always been around 70-75 percent, and the capital share has always been around 25-30 percent."

 Pearlstein, by contrast, wants us to believe "the share of the economic pie going to workers in the form of wages, salaries and benefits seems to have fallen, while the share going to holders of capital -- in the form of interest, dividends and capital gains -- has gone up." His thesis is that globalization has made capitalists more effective at exploiting the proletariat.

 As I explained in my last column, labor's share always rises in recessions (it peaked in 2001) because recessions shrink profits and capital gains. That does not mean workers should welcome recessions just to get a larger share of a smaller pie.

 Unlike Pearlstein's untenable claim that "labor's share of the economy is shrinking," the source of his favorite statistic is not a mystery. It comes from a 10-page paper on "retrospective income" prepared for an academic conference by Michael Strudler and Tom Petska from the Statistics of Income division of the IRS, and Ryan Petska from Ernst and Young. But estimates are just estimates, and estimates derived from tax returns (including those from the CBO) are among the worst. Continued...