Tax subsidies for the wealthy, part 287

The current status is that yes, there is a tax deduction for homeownership. The president's tax advisory council currently recommends modifying but not eliminating it. In this context, here is some discussion of the above arguments.

Really, do homeowners vote more?

As for the liberal-conservative questions, this paper faces the classic problem of ferreting out causation: being older, wealthier, and more rural make you more likely to buy a home and to be a Republican at the same time, so to what extent can we attribute Republicanness to homebuying and to what extent is it just becoming old and stodgy? On p 1600, we get a table of 136 or so coefficients, eight of which are coefficients on home ownership. The coefficient on support for spending on inner-city problems indicates that homeowners dislike such spending more than renters, but the authors reject the coefficient (-0.05, versus +0.68 for voting) as too small to be worth caring about. I don't get this, since it's a unitless number and statistical significance means that it is not a question of variance in a small sample (y'know, under the right assumptions).

The positive result about home ownership provides some confidence, but the null results are hard to use without a real live model being tested. We know that age and income and race affect both homebuying and politics, so they just throw 'em in on the right-hand side and let them all duke it out. On top of just being sort of lazy, this method reduces the power of the test: if homebuying and age are highly correlated, then including age in the regression will reduce the coefficient on homebuying, and this is true no matter what is being regressed. If we ran a regression of the form (frequency of trips to Home Depot) = A * (homebuying) + B * (age), then A will be smaller than the A' in (frequency of trips to Home Depot) = A' * (homebuying). So loading up the right-hand side seems sensible, but biases the results toward a null result. In most papers, this is OK because the author is rooting for big coefficients, and so is making it harder for the pet theory to be proven right, but if the author is hoping for null results, then loading up the right-hand side with correlated coefficients is sort of cheating. Not that I could do better: the problem of using statistical techniques to determine causality is not yet solved.

Do tax discounts on home buying induce more home buying?

At the level of detailed rules, the tax deduction is the usual mess. For example, you can deduct interest on a home equity loan. That is, if you take out a loan for $100,000 using your home as collateral, and then spend that on strippers and coke, then you can deduct the interest on that loan. You used your home as collateral so it's a home loan, see. The new plan will eliminate this.

The report below ([2] on p 72) notes that current U.S. owner occupancy is at 69% in the U.S.A., but Canada, the UK, and Australia have no mortgage interest deduction and are at 66%, 69%, and 70%, respectively. That is, the U.S. isn't very much out of line from the non-deduction world.

The first of a pair of relevant papers in the academic literature is Rosen and Rosen [3]. It opens by pointing out that "Since World War II the proportion of the U.S. housing stock that is owner occupied has risen from 48 to 64 percent." (that was in 1980) The question, then, is how much of this is due to the mortgage interest deduction. The deduction was in effect for this entire period, so we don't have a before-after picture to look at; instead the authors went with a simple time series analysis.

I am well-aware that everybody is sick of me ranting about time series models. Fortunately, the authors point out the key critique about a time-series model themselves: "The most important is that it lacks a choice-theoretic foundation." (p 62) That is, there is no serious model of decision making behavior here, just an exercise in numerical correlations. That said, the model finds that: "If all personal income tax benefits for homeownership were eliminated, our model indicates that in the long run the proportion of homeowners would drop by about 4 percentage points." (pp 69-70) But before telling your friends, see below.

They find this by calculating coefficients using net housing prices (with the tax) and then recalculating the rate of homebuying with prices modified to pretend there is no tax rebate. That is, they are estimating an elasticity with respect to price (if we lower prices by 1%, by what percentage will homebuying go up?), and then multiply that by the reduction in price implied by the tax rebate.

This paper also finds no effect for their measure of capital constraints. Don't have the time to investigate this one further, but this is a point of evidence against my claim at the head of this article and below that it's capital in the traditional sense that's keeping people from buying.

But wait! Rosen, Rosen, and Holtz-Eakin [4] re-estimated the change in an article entitled "Housing Tenure, Uncertainty, and Taxation". The article takes into account the investment properties of housing (and as the title suggests, takes into account the riskiness of the investment), and they find that eliminating the deduction would induce not a 4% change, but a 0.4% change. This is the most recent I could find (It's only cited by one other basically unrelated article), so I'm going to say that the cutting edge in the econ literature--bearing in mind all the modeling caveats above--finds a very small effect on homebuying from the interest deduction. Of course, by lowering prices there is some effect, but a 0.4% change isn't much bang for 69 billion bucks.

For the liberal homeowner

First, you may want to go read the Slate article entitled "End the Mortgage-Interest Deduction!: Why the left should embrace the Bush tax commission's most radical proposal." That article makes a number of good points that I will try not to repeat here. The key point is that this, like many tax exceptions, just benefits the wealthy. [Also, don't miss this PDF linked from the above article.]

However, an important point that the Slate article discusses but seems to downplay is that the tax deduction is not being outright eliminated, but is being replaced by a tax credit. This mitigates the benefit to the wealthy, but still subsidizes those who had the capital to help themselves.

The official recommendations from the President's Advisory Panel on Tax Reform begin on page 70 of this PDF [2]. The story given for cutting the homeowner tax credit is wonderfully Republican: Why favor investment in residential real estate over other forms of investment? Noncorporate business is taxed at 17%, corporate at 26%, home investment at 0%. What kind of bleeding-heart welfare is this!?, the report asks. The claim is that people are making the decision to either invest their capital in a business (via stocks, no doubt) or in a home, and we shouldn't bias them toward investing in a home. Making what is no doubt a wise political decision, the report ignores the arguments above about how homeowners are better citizens, or that the deduction is inequitable.

The details

The distortions that deductions suffer turn what looks on the surface to be an equitable tax discount into a regressive tax discount (i.e., one which favors the wealthier). You have two choices with deductions: go through the tax code and find all the deductions that match your life and itemize the full list, or just take the standard deduction, which is fixed and can not be augmented with new revisions. The trouble with deductions, therefore, is that most of us po' folk don't bother and just take the standard deduction. Thus, deductions are more likely to be taken by the wealthier. Additionally, $1,000 less net income for someone in a 35% tax bracket means $350 in pocket, and for someone in a 15% bracket it means $150 in pocket, so the deduction is almost by definition regressive.

po' folk get no deduction; the wealthier brackets
save about $5,000

Figure 1: On the surface, it is an equitable tax discount, but the details create bias. From the above Tax Panel PDF. double-sized, legible figure

There's a problem with credits too, by the way: if your tax burden falls below zero, you don't get a refund on the credits. [The one exception: the Earned Income Tax Credit, which is actually a form of welfare in the guise of a tax rebate. If you have zero tax burden, the fed will send you a check for remaining EITC.] So a person who has a near-poverty job but had the wherewithal to buy a house won't be benefiting from mortgage interest credits, but probably wasn't benefiting from the current system anyway. Nor is this an absurd proposition: 27% of the homeowners in the data used by Gilderbloom and Markham [1, p 1594] earned under $11,250/year [in early 90s money, or about $17,000 in 2005 dollars, which is pretty close to the poverty line for a family of three].

So converting to a credit is a major shift away from the regressive nature of the tax. Therefore, I side with Slate's recommendation that we equity-oriented liberals should support the switch to a credit. But the reader will note that this has nothing at all to do with housing: any time you see a recommendation to convert a tax deduction into a basically equivalent tax credit, the system will be more equitable. I could not find a source that would state whether the mortgage interest credit system will cost more or less than the deduction system, so I am maintaining a neutral prior and assuming no change. One question that those better with trench-level politics may be able to answer: why are we switching the mortgage interest deduction to a credit, but not the thousand other deductions?

The deduction/credit switch will make any tax more equitable, but it doesn't address the basic inequity to the mortgage interest deduction: why are we subsidizing people who own over those who rent--and why are we doing it with a tax rebate? The problems with making the transition from renting to owning are all about hard constraints people face in the immediate present: no cash to pay down payments, escrow fees, inspection fees, insurance fees, and the six percent cut to the National Association of Realtors(R); no month free to be spent drowned in paperwork, and no time to get an education in legalese and accounting. A promise that at some point in May of next year the Fed will hand you $1,000 does nothing to mitigate any of these present constraints. One can directly address these constraints via cash up front, when people actually need it; there are programs (PDF) that cost a whole lot less than $69 billion/year that use these methods to directly target those who could not buy a home without help.

The President's Advisory Panel decided on a half-measure, eliminating the deduction but replacing it with a credit, which is probably as much as is politically feasible. As such, we should support it, even though the entire $69 billion cash transfer to homeowners does nothing but benefit people who write a monthly check to a mortgage company instead of a landlord.

[1] @articlegilderbloom:politics, author = Gilderbloom, John I. and Markham, John P., citeulike-article-id = 399252, journal = Social Forces, number = 4, pages = 1589-1607, priority = 2, title = The Impact of Homeownership on Political Beliefs, url = <a href="http://links.jstor.org/sici?sici=0037-7732%28199506%2973%3A4%3C1589%3ATIOHOP%3E2.0.CO%3B2-5">Jstor link</a>, volume = 73, year = 1995

[2] @booktax:report, author="President's Advisory Panel on Tax Reform", title = "Simple, Fair, and Pro-Growth: Proposals to Fix America's Tax System", year =2005, month="November", publisher="Government Printing Office", url="<a href="http://www.taxreformpanel.gov/final-report/">here</a>"

[3] @articlerosen:obsoletehousing, author = Rosen, Harvey S. and Rosen, Kenneth T. , citeulike-article-id = 399217, journal = The Journal of Political Economy, number = 1, pages = 59-75, priority = 2, title = Federal Taxes and Homeownership: Evidence from Time Series, url = <a href="http://links.jstor.org/sici?sici=0022-3808%28198002%2988%3A1%3C59%3AFTAHEF%3E2.0.CO%3B2-J">Jstor link</a>, volume = 88, year = 1980

[4] @articlerosen:uncertainhousing, author = Rosen, Harvey S. and Rosen, Kenneth T. and Holtz-Eakin, Douglas , citeulike-article-id = 399303, journal = The Review of Economics and Statistics, keywords = homeownership, number = 3, pages = 405-416, priority = 2, title = Housing Tenure, Uncertainty, and Taxation, url = <a href="http://links.jstor.org/sici?sici=0034-6535%28198408%2966%3A3%3C405%3AHTUAT%3E2.0.CO%3B2-5">Jstor link</a>, volume = 66, year = 1984