Before we let PSZ go, we've not talked about their egregiously misleading result that the bottom half's income has risen by only $500 since 1980. /1
If you look at the @SentierRes numbers for median household money income and adjust them for inflation using the PCE deflator, since its previous cyclical peak in December 2007, median hh income has risen by $5,000. It rose by over $13,000 from 1980 to 2007. 13+5=$18K richer /2
Money income includes cash transfers, which isn't included in PSZ national income (except UI and Social Security benefits). But the $$ from other cash transfers isn't huge (SSI, TANF, Workers' Comp, veteran payments), especially at the median. /3
(Note that noncash transfers have grown more generous over time, though those aren't in money income.) Do we really think there has been a drop in *unobserved* national income big enough to counter all this? What's going on? /4
We should note that Piketty & Saez's data was always egregiously misleading for income trends below the top. Their CURRENT estimates show that the bottom 90% was richer in 1969 than in 2018. If you believe that, you have no business writing or thinking about income trends /5
One problem with the PS estimates is avoided in the PSZ ones: PSZ include Social Security in income. One of the problems with PS estimates is they don't include ANY transfers, so as the population ages, more of them are retirees without much/any market income. /6
One thing driving PS's and PSZ's finding is that they are looking at tax units and (in the case of PSZ) splitting the income between spouses filing jointly. Tax units are like tax returns, with added would-be returns from nonfilers (imputing income to them). /7
Tax units (in the PSZ data) include dependents (think college kids). Tax units (in the PSZ data) also include the institutionalized (treated as tax units with $0 in income). So they're padding the bottom w nursing home residents, prisoners, & members of psychiatric facilities /8
By splitting incomes between spouses, they're not allowing for economies of scale. A married couple doesn't pay two mortgages, though two people dating and living apart each have housing expenses. /9
There's also the problem of under-reporting of income (addressed potentially badly by PSZ) and of well-off people with negative or low reported income due to taking business losses, which Auten and Splinter highlight davidsplinter.com/AutenSplinter-… /10
(Incidentally, the bus loss issue affects the CBO data, too, when you look only at the market incomes of non-elderly people, so save yourself some time & don't go there. If you do, you'll see business income makes up a suspiciously large share of the bottom fifth's income) /11
Nor do PSZ include child support or alimony. These are all defensible positions given the fetishization of national income that has overcome PSZ, but they result in estimates that fundamentally and profoundly understate improvement in living standards over time /12
Which adds to the perception that (overstated) rising income concentration has siphoned off money below the top. Maybe that's true. But you can't credibly make that case with the PSZ estimates. /13
Bottom line: quite apart from whether you shld believe PSZ or Auten-Splinter, different income definitions are useful for different purposes, and equal-split individual pretax national income is a bad, bad way to gauge trends in living standards of the poor and middle class. /fin
