I have explained how the FDIC-HASP mortgage modification plan massively distorts the supply of income-earning efforts, because its mortgage modification is large and means-tested: its formula implies that an action taken by a borrower to increase his income would increase his housing payment obligation by 31 percent of the income increment. If the affordable payment (i.e., the payment that would comprise 31 percent of income) were re-evaluated monthly, this would amount to a 31 percent marginal tax rate in each month that a modification could occur.
Standard practice determines an affordable payment based on the most recent year’s income, and puts that payment in place for five years. Thus, a marginal dollar earned in the base year raises mortgage payment obligations by 31 cents in each of the following five years, and may also raise payment obligations beyond the five-year modification period.
In what I have previously written on this subject, I ignored the later year terms as would be appropriate if modifications were achieved soley by reducing interest payments (that is, leaving the time path of principal payments unchanged), and interest payments were permitted to jump up to the originally contracted amount when year 5 was over (but note that the U.S. Treasury, 2009, has said "[the] lower interest rate must be kept in place for five years, after which it could gradually be stepped up to the conforming loan rate in place at the time of the modification.")
At the other extreme, when followed by a modification of purely principal, a marginal dollar earned in the base year raises mortgage payment obligations by 31 cents in every year the loan is outstanding. For example, with 25 years remaining and an interest rate of 6 percent per year, this amounts to a 396 percent marginal tax rate!
It is my impression that interest reductions are the margin used most to modify mortgages, in which case the 396 percent marginal tax rate would not apply. But it does illustrate the point that the 131 percent rate I have been using could be significantly understated.