Capping the mortgage interest deduction.

INTRODUCTION AND BACKGROUND

public policy designed to encourage home ownership in the United States operates primarily through incentives contained in the federal income tax system. The largest housing-related subsidy in the federal income tax code is the mortgage interest deduction (MID), which, as estimated by the Office of Management and Budget (OMB), reduced income tax revenues by $79.9 billion (1) in fiscal year 2007 (Executive Office of The President, 2007). This makes the MID the second largest tax expenditure, exceeded only by the exclusion of employer contributions for medical insurance premiums and medical care.

The MID alters the user cost of owner-occupied housing for taxpayers, making it more attractive to purchase a home. However, its effectiveness at encouraging home ownership has been a point of contention. Since the subsidy is based on the amount of interest paid on a mortgage, larger subsidies are provided to those purchasing more expensive homes (subject to the current cap of $1 million). While this certainly provides an incentive for people to increase their consumption of housing, it may not be particularly effective in altering the choice between renting and owning (i.e., tenure choice). Policies like those considered here, which lower the cap on the number of mortgage dollars that qualify for the MID may better target the subsidy towards those on the margin between owning and renting, and away from inframarginal households that would choose to own a home whether they receive a subsidy or not.

In order to examine the economic implications of capping the MID, we extend the standard user-cost model of the rental price of housing to include a cap on the amount of mortgage that receives tax-preferred status. Although a $1 million cap currently exists, this feature has not been included in earlier user-cost models. (2) We use this model to show that capping the MID changes the user cost of housing through the share of the mortgage exceeding the cap.

The issue of capping the MID is becoming increasingly important for two primary reasons. First, the current-law cap is not adjusted for inflation and, hence, will affect more mortgages over time as home-price inflation pushes increasing numbers of tax payers over the $1 million ceiling. Second, the recommendation by the President's Advisory Panel on Federal Tax Reform (the Panel) to create partially regionally adjusted caps for the MID (based on median home prices) makes it clear that policy advisers have an interest in changing current law. Our paper examines several ways of capping the MID: lowering the national limit, creating a set of limits that are fully adjusted for local home prices, and the partially regionally adjusted limits recommended by the Panel.

We use data on individual mortgages to simulate the average share of each mortgage in excess of the caps that apply through current law as well as our three alternative policies. For the current-law cap, our national estimates show that less than one--half of one percent (0.39 percent) of mortgage originations exceed the cap, and on average just 0.13 percent of mortgage dollars are subject to the cap. The caps recommended by the Panel would raise the number of mortgages subject to the cap to approximately 13 percent, and the average share of mortgage dollars subject to the cap to 3.44 percent. Regional variation in the effect of the Panel's caps is striking, ranging from negligible effects in some MSAs to a high of 67 percent of the mortgages (and, on average, 23 percent of mortgage dollars) subject to the cap in the San Francisco, CA, Metropolitan Statistical Area (MSA). We also provide MSA-level simulations for a uniform national cap and a fully regionalized cap, both designed to affect the same share of mortgage dollars as the caps in the Panel's proposal to facilitate comparison.

Our analysis shows that estimates that do not consider caps can significantly understate user costs for mortgages (both under current law and alternative policies). For the average mortgage in excess of the current-law cap, for example, not accounting for the cap would understate the user cost by about 4.4 percent (around $4,000 annually).

The remainder of the paper begins with a summary of previous studies that have examined the tax treatment of housing in general, and the MID in particular. Then we present our extension of the standard user-cost model. The main section of the paper then discusses three alternative proposals for changing the cap applied to the MID. The simulations in this section show the fraction of mortgages and the average share of mortgage dollars in excess of each cap. We then provide estimates of how each cap would affect the user cost of housing. The final section of the paper summarizes the main results.

PREVIOUS STUDIES

Economists have long been concerned about the efficiency and equity effects of housing subsidies. Rosen (1985) reviews the early literature, showing that these subsidies have significant effects on both the tenure choice and on the quantity of housing consumed. Conditional on home ownership, Rosen (1979a) estimates that without tax subsidies U.S. residents would have lived in homes that were nine to 17 percent less valuable than their current homes in 1970, depending on their income level. King (1981) similarly estimates that the elimination of tax subsidies for housing in the United Kingdom would reduce the quantity of housing consumed by about 13.7 percent.

Rosen (1979a, 1985) also translates these consumption distortions into efficiency loss estimates. He calculates that his estimates for 1970 translate into an average annual excess burden of about $192 in 1980 dollars. Rosen (1985) also notes that some of the early literature expressed a concern that subsidy-induced housing consumption comes at the expense of business investment (e.g., Summers (1980)). However, he notes that prior to 1985 the econometric evidence on this issue was insufficient to establish this relationship conclusively.

Mills (1989) provides evidence on the efficiency of the allocation of the capital stock in the United States. He uses national income data over the period 1929-1986 and computes the return to housing capital compared to all other private fixed capital. His results indicate that the real returns to capital have been smaller in housing than for non-housing fixed capital, suggesting overinvestment in housing. The President's Advisory Panel report (2005) includes a comparison of the effective tax rates on different types of investment, as calculated by the Treasury Department. Their estimates report that the marginal effective tax rate for owner-occupied housing is zero percent, while the rate is 17 percent for non-corporate business, 26 percent for corporate business, 22 percent for the business sector as a whole, and 14 percent economy-wide. Such differences in effective tax rates are ultimately bound to distort the allocation of capital.

Rosen (1985) also discusses two significant early studies on the tenure choice between owner-occupied and rental housing. Rosen and Rosen (1980) estimate that tax subsidies for owner-occupied housing raised the home-ownership rate by about four percent in 1974. Hendershott and Shilling (1982) provide slightly higher estimates in the range of five to 6.5 percent, depending on the assumed average marginal income tax rate.

On the equity front, both Rosen (1979a) and King (1981) extend their analyses to assess the distributional implications of eliminating housing tax subsidies. Both studies find that the elimination of these subsidies would tend to reduce income inequality relative to the status quo. Rosen (1979a) shows that this result would hold in the United States (although to a lesser degree) even if total tax revenues are held constant via a proportional tax cut. Anderson and Roy (2001) examined the distributional impact of eliminating housing tax preferences, including the elimination of both the mortgage interest and property tax deductions. Their simulations show that elimination of these housing tax preferences would increase the progressivity of the income tax significantly, as measured by the change in the Suits index with a bootstrapped confidence interval.