Via Fausta comes this insightful article by economist David R. Henderson of the Hoover Institution:

“How often have you heard that the vast majority of families’ incomes in the United States are rising little or not at all, that the middle class is shrinking, that real wages are stagnating, that the top 20%, or 5%, or 1% are getting the lion’s share of the gains in the U.S. economy, that average CEO pay is getting to be a couple of orders of magnitude larger than average people’s pay, or that mobility across income groups has declined? Princeton economist and New York Times columnist Paul Krugman has made a good part of his living credulously repeating most of these claims. Wall Street Journal reporter David Wessel has also often written long articles laying out some of these claims. It seems that not a month has gone by in the last few years that a major respected newspaper hasn’t made such statements as if they were well-established facts.”

Henderson’s article is actually a review of fellow economist Alan Reynolds’ new book, “Income and Wealth.” There, Reynolds picks apart the data regularly cited by Krugman and his epigones,especially a report by Thomas Piketty and Emmanuel Saez containing the oft-cited statistic that the median family income of the bottom 90 percent of Americans fell between 1973 and 2000. As Henderson notes, that’s true only if you count as a separate “family” your college-student offspring who work summers as lifeguards:

“As for the Piketty-Saez study that Krugman and many others have cited, Reynolds points out just how implausible their data are as a measure of family income. Piketty and Saez write that in 2000, ‘the median income, as well as the average income for the bottom 90% of tax units is quite low, around $25,000.’ Note the use of the term ‘tax units.’ ‘Tax units’ are not the same as families. In my family, for example, we have two tax units: my wife and I file our taxes jointly and our daughter files on her own. But that has not stopped people, including Krugman, from writing as if ‘tax unit’ and ‘family’ are synonymous. Reynolds points out that if tax units were the same as families, highly implausible implications would follow. Given the meaning of the word ‘median,’ 45 percent of families (half of 90 percent) would have had to make less than $25,000 in 2000. But U.S. Census data show that for 2000, median family income was $50,732, which means that we know, to the extent we can trust Census data, that 50 percent of U.S. families made more than $50,732 in 2000. That means that the 5 percent of the family income distribution not accounted for (100 percent minus 50 percent minus 45 percent) would have had to be the people with incomes above $25,000 but below $50,732. While that is mathematically possible, it is empirically virtually impossible. The problem stems from the equation of ‘tax unit’ with family.

“How about the ‘vanishing middle class’? To say that the middle class is vanishing, one must have a definition of the middle class. A sensible person would probably define it as the group of people in the middle, say the middle quintile or the middle three quintiles. But that?s not how the commentators who have made the claim have defined the middle class. Instead, they take the group of people making income within a fixed range in inflation-adjusted dollars, say, between $35,000 and $50,000, and show that the percent of the overall number of families within this range is falling. In commenting on this way of defining the middle, Reynolds points out an obvious but, nevertheless, often completely overlooked fact:

“‘Such a fixed definition ensures that the proportion of households in the middle group must decline with a rise in general prosperity, because rising prosperity causes a rising percentage of families to earn more than $50,000.? (emphasis his)”

Moral: It’s probably not a good idea to believe everything you read by Paul Krugman.