The Problems With The Piketty-Saez Data Or Why Middle Class Has Done Much Better Since 1979 Than Reported

Click here to read this great post from Greg Mankiw. Here is most of it

"This national conversation has generated renewed attention to the highly influential Piketty-Saez data.  It is worth pointing out, therefore, some limitations of these data, which have been stressed by Cornell economist Richard Burkhauser: The data are on tax units rather than households, they do not include many government transfer payments, they are pre-tax rather than post-tax, they don't adjust for changes in household size, and they do not include nontaxable compensation such as employer-provided health insurance.

Does this matter?  Yes!  Here are some numbers from the Burkhauser paper:

1. From 1979 to 2007, median real income as measured by pre-tax, pre-transfer cash income of tax units rose by only 3.2 percent.  That is a paltry amount for such a long period.  You might conclude that middle class incomes have been stagnant. But wait.

2. Households are more important than tax units.  Two married people are one tax unit, whereas a couple shacked up are two tax units.  We would not want to treat the movement from marriage to shacking up as a drop in income.  If we look at households rather than tax units, that meager 3.2 percent rises to a bit more respectable 12.5 percent.

3. Now consider government transfer payments. If we add those in, that 12.5 percent number becomes an even better 15.2 percent.

4. What about taxes? The middle class received some tax cuts during that period.  Factoring taxes in, the 15.2 percent figure rises to 20.2 percent.

5. But not all households are the same size, and the size of households has fallen over time. Adjusting for household size increases that 20.2 percent to 29.3 percent.

6. There is still one thing left: employer-provided health insurance, an important fringe benefit that has grown in importance. Adding an estimate of that into income raises the 29.3 percent figure to 36.7 percent."