You're quite correct that I didn't account for the cost of housing paid by those making between $65,000 and $125,000, which certainly reduces the marginal rate on those making over $125,000 (not by reducing the actual burden on those people but rather by increasing the burden on those paying between $65,000 and $125,000).
But there are many, many things I didn't account for. The point was not to make a full accounting, the point was to present a back of the envelope calculation demonstrating what can hypothetically happen when programs are structured this way. If you want better researched numbers I encourage you to read this:
Those numbers also point out something else somewhat misleading about my example -- although it is possible (particularly with regard to education programs) for the upper middle class to be ravaged by excessive marginal tax rates, those most hard hit by means tested programs are really the working poor. Have a look in the first graph at the marginal rate on someone making minimum wage.
Your 'back of the envelope calculation' was in error, and in a way that affected the conclusions you drew. You claimed that the new Stanford FinAid policy introduced a 'cliff' (negative returns for increased family income), but it doesn't.
I'm not making any other claims about the merits of progressive taxation, price discrimination by universities, or economic policies affecting the working class. I'm just pointing out your error so that other readers don't rely on your calculations.
From the fine article:
"Stanford will expect no parental contribution toward tuition [~$45K for 2015-6] from parents with annual incomes below $125,000.... And there will be zero parental contribution toward tuition, room or board [~$65K for 2015-6] for parents with annual incomes below $65,000"
So your table should read:
28% federal income tax ($28,000)
9.3% California state income tax ($9,300)
Reduced Stanford FinAid ($45,000)
Total: ($82,300)
This is admittedly a very high marginal rate, but it is not negative.
Of course the calculation was wrong. The correct calculation would require consideration of the entire tax code and every means tested benefit the hypothetical family might be eligible for. But your assumption that the loss of $45,000 rather than $65,000 makes it not a cliff is equally erroneous, because many of the other things not accounted for go the other way. The loss of financial aid may still exceed $45,000 for a family with more than one child in college, or if the housing expenses in question are eligible for some other source of means tested financial aid. The additional income may destroy various previously permissible tax deductions because of the Alternative Minimum Tax. There are any number of totally unrelated benefits that the family may lose eligibility for at the same time.
There is a realistic hypothetical family that loses more than they make in total regardless of whether the loss of Stanford financial aid is $65,000 or $45,000.
But there are many, many things I didn't account for. The point was not to make a full accounting, the point was to present a back of the envelope calculation demonstrating what can hypothetically happen when programs are structured this way. If you want better researched numbers I encourage you to read this:
http://johnhcochrane.blogspot.com/2012/11/taxes-and-cliffs.h...
Those numbers also point out something else somewhat misleading about my example -- although it is possible (particularly with regard to education programs) for the upper middle class to be ravaged by excessive marginal tax rates, those most hard hit by means tested programs are really the working poor. Have a look in the first graph at the marginal rate on someone making minimum wage.