How much revenue would scaling back the Mortgage Interest Deduction raise?

Less than static analysis would suggest. Even in the absence of the mortgage interest deduction, homeowners get a tax preference because imputed rent goes untaxed. If the MID were scaled back or eliminated, some households would sell assets with taxable returns to pay down their mortgages, thus reducing the net tax revenue arising from the policy change.

Jim Follain was the first (I think) to arrive at this insight, and Pat Hendershott and I found that people in Australia (where there is no MID) pay off their mortgages faster than Americans. The reduction in leverage is probably a good thing, and I know of no compelling argument for the mortgage interest deduction (it does almost nothing to encourage homeownership), but we should not delude ourselves into thinking that its elimination would produce a great revenue windfall.