So, how does one argue with a drab statistic that we can all wrap our minds around like 80% of millionaires being first generation as a measure of America's high degree of intergenerational income mobility? Or, for that matter, its inverse - the 1 in 18 chance an individual from a non millionaire family has of becoming a millionaire?
How does one deny this sort of data?
Below is what he dubs the Great Gatsby Curve. On the horizontal axis is the Gini coefficient, a measure of inequality. On the vertical axis is the intergenerational elasticity of income — how much a 1 percent rise in your father’s income affects your expected income; the higher this number, the lower is social mobility.
Misdirection -- Don't count the number of 1st generation millionaires and compare that to the total. Look, instead, at how a parent's higher income affects, not their children's income, but their children's expected income and then compare that to the "Gini Coefficient", a measure of inequality, and call the result a measure of intergenerational income mobility or immobility.
So what does a measure of "how much a 1 percent rise in your father’s income affects your expected income" really show?
It shows how much children benefit (relative to their peers) from wealthy parentage. It ignores that those children may be less wealthy than their parents, and that their children may be less wealthy than them -- the real question of intergenerational income mobility.
Knowing that Krueger measures, not intergenerational changes in wealth, but the degree to which children benefit relative to their peers from wealthy parentage, if we turn to the linked chart that compares nations we see that children benefit less from parentage in generally more statist countries (listed low/left on the chart) and benefit more from parentage in generally less statist nations (listed high/right on the graph).
Krugman says more about statist governments breaking the bonds of family than it does about intergenerational income of which it says nothing.
H/T Andrew Brod