Title

The Mortgage Interest Deduction: Revenue and Distributional Effects

Author(s)

Austin J. Drukker Austin Drukker (Brookings Institution)

Ted Gayer Ted Gayer (Brookings Institution)

Harvey S. Rosen Harvey Rosen (Princeton University)

Abstract

Conventional estimates of the size and distribution of the mortgage interest deduction (MID) in the personal income tax fail to account for potentially important responses in household behavior. As noted by Gervais and Pandey (2008) and Poterba and Sinai (2011), among others, were the MID to be eliminated, households would sell financial assets such as stocks and bonds to pay down their mortgage debt, and the smaller holdings of these taxable assets would offset some of the revenue gains from taxing mortgage interest. Conventional estimates therefore overstate the increase in revenues associated with eliminating the MID. Conventional estimates also overstate the progressivity of eliminating the MID, because households with higher levels of non-residential assets would respond by selling their taxable, non-residential assets. This paper builds on previous work that estimates the consequences of removing the MID using a framework that allows for the possibility of portfolio rebalancing. Unlike previous studies, we analyze data for several years — every third year from 1988 to 2012, inclusive. This reduces the likelihood that our estimates are due to the idiosyncrasies of some particular year, and allows us to investigate how and why the differences between estimates with and without a portfolio response have evolved over time. We then turn to the distributional implications of eliminating the MID, again looking at multiple years. A noteworthy feature of our distributional analyses is that we focus on both wealth and income as classifying variables. Our main findings are: (i) The revenue loss associated with the MID is smaller if one allows for rebalancing, with the ratio of the rebalancing-adjusted revenue loss to the conventionally estimated revenue loss varying from 76 percent in 1997 to 90 percent in 2009. While not dramatic, these are non-trivial effects. (ii ) During our sample period, changes in the ratio of the 2 revenue loss estimates were due primarily to changes in the relative stocks of assets to mortgage debt as opposed to changes in rates of return and the tax system, (iii) Portfolio rebalancing attenuates the increase in progressivity associated with elimination of the MID.

Creation Date

2017-10

Section URL ID

Paper Number

251

URL

https://gceps.princeton.edu/wp-content/uploads/2017/10/251rosen.pdf

File Function

Jel

H24, H31

Keyword(s)

Suppress

false

Series

3