The mortgage interest tax deduction (MITD), a portion of the tax code that permits homeowners to deduct mortgage interest from their taxes, has widespread popular support. It is the second most common tax deduction, and the largest federal tax expenditure. It is particularly popular among middle class homeowners and the politicians they vote for. But it is often criticized by academics and policy wonks for benefiting well-to-do families who need it least, for encouraging overconsumption of housing, and for subsidizing sprawl. The map below, from a Pew Institute study released last week, shows the spatial distribution of the tax deduction. MITD claimants are concentrated in middle-class suburbs around (but not in) urban areas, where the houses are big and the homeownership rate is high.
A recent white paper released by Chetty, Hendren, Kline, and Saez (hereafter CHKS) from the Harvard/Berkeley Equality of Opportunity Project comes down in support of the MITD. The authors theorize that "these deductions may impact economic opportunity by providing opportunities for credit-constrained middle and low income families to become homeowners." The study finds that the MITD, along with many other tax expenditures, is correlated with higher income mobility. That is, in places where the MITD is larger, children's success in life is not bounded by their parent's wealth or poverty. It seems they are free to succeed or fail on their own merits.
In this blog post, I examine the way the MITD was measured in the CHKS study and propose an opportunity to improve this measurement. Then I re-estimate their results using an alternative metric. Using the CHKS study data set and our alternate measurement, it appears that the regressive nature of MITD expenditures corresponds with reduced mobility overall and has a large, negative association with the mobility of low-income Americans.
CHKS demonstrated that average MITD expenditures have a positive association with intergenerational mobility, but they were unable to find an effect when evaluating the progressivity of the MITD. This is probably because the metric was noisy. The authors calculated the progressivity of the MITD by subtracting MITD over adjusted gross income (AGI; MITD/AGI) for the top and bottom income cohorts. By this measure, the MITD looks progressive: as a percentage of AGI, the wealthy take a deduction that is, on average, 8.6 percent lower than the deduction taken by the poor.
But is this what the data are really telling us? Below is a chart showing the sum of all mortgage interest tax deductions in blue and the tax deduction as a percentage of adjusted gross income in green. We can see that the bulk of the deduction goes to people making between $100,000 and $200,000 a year.
Looking at the tax deduction as a percentage of income (in green), we can see that by just comparing the tails, the deduction looks progressive. Those with an adjusted gross income of less than $10,000 a year deduct a much higher percentage (15 percent) of their AGI than those who make $200,000 or more a year (just 3 percent). After the tax expenditure, the income distribution should be flatter than before. However, if you look at the middle of the distribution, the opposite story is true. The tax deduction is regressive—income brackets with higher incomes claim higher deductions as a percent of income. After the policy, the income distribution is more unequal than before.
Why do the tails tell a different story than the middle of the distribution? And why do people in the lowest income category—who typically don't own homes and can't qualify for a mortgage—have the highest percentage of deductions? Well, as it turns out, according to statisticians at the Internal Revenue Service (IRS), the bottom bracket is inflated with a number of wealthy folk who declare high losses to reduce their adjusted gross income. This explains why, for example, in well-to-do communities like Coral Gables, Florida, or Nantucket, Massachusetts, the average person with an AGI of less than $10,000 a year deducts close to 100 percent of that in mortgage tax interest.
The top bracket, by contrast, seems understated. Per filer, this group claims a much larger deduction than any other group. But because this bucket includes the Bloombergs and Buffetts of the world, this high MITD registers as just a small percentage of even higher AGI.
Because the data in the bottom income bracket is noisy, and the top bracket is skewed by those with extremely high income, it makes more sense to calculate the progressivity of the MITD by comparing the second-highest and second-lowest income cohorts. Below is a map of the results.
What happens if we plug this new metric into the CHKS study data? The initial results show that where the MITD is more regressive, parents' income is a better predictor of children's income, and mobility is lower. The results also show that a more regressive MITD corresponds with steep declines in mobility for low-income Americans. This finding makes sense, given that the benefit bypasses low-income homes, either because they are not homeowners or because they do not make enough money to itemize deductions.
It's important to note that the results are correlational, not causal. But if they have any interpretation at all, they suggest that the overall regressive structure of the MITD may be reducing equality of opportunity and making it harder for low-income families who do not own homes to achieve the American Dream.
By Elora Raymond, graduate research assistant, Center for Real Estate Analytics in the Atlanta Fed's research department, and doctoral student, School of City and Regional Planning at Georgia Institute of Technology