There are days when I think I could rename this blog Stupid Things In SmartMoney and concentrate my efforts solely on that ironically named publication. Today’s inspirationally bad offering is Mortgage Deduction Pads Lender Profits, in which we learn that “A tax perk aimed at homeowners ends up raising mortgage rates, research shows.”
Apparently, the mortgage interest deduction has the unintended consequence of driving up interest rates, harming those poor American homeowners it was meant to help. How does this happen? Well, here is where the story gets a little fuzzy.
Standard economics theory holds that consumer subsidies raise demand for goods, thereby shifting prices higher. With mortgages, the "price" is the interest rate.
I can recast this in a way that almost makes sense. The mortgage interest deduction increases the value of houses, which increases the total amount borrowed with mortgages. That increased demand raises interest rates. But it would raise all interest rates, not just on mortgages but on credit cards and Treasury bonds too. It is a competitive marketplace, and all those potential loans are bidding for the same pool of potentially lent dollars.
But as theoretically sound as this reasoning might be, it seems pretty clear to me that the effect of the mortgage interest deduction on global interest rates is likely immeasurably small, particularly in light of the many other factors that affect those rates.
Enter the academic paper that SmartMoney reports on. The Incidence of the Mortgage Interest Deduction: Evidence from the Market for Home Purchase Loans by Andrew Hanson of the Economics Department of Georgia State University. Even in the context of some of the terrible papers I have discussed here, this one is a travesty.
The author’s approach is to examine interest rates on mortgages above and below the $1M ceiling on deductibility. His reasoning is that since only sub-$1M mortgages benefit from the subsidy, lower rates for bigger mortgages will show that the subsidy is driving up interest rates.
The problem with academics (and journalists) is that they often live lives sheltered from the realities of the marketplace. So they tend to miss points that are obvious to us poor slobs trying to make it in the real world. If you were a bank, why would you write a $1.5M mortgage at a lower rate than you could get on two $750K ones? Everything else equal, you would prefer the two smaller loans for diversification anyway.
Blithely ignoring the absence of reasoning that would support the effect he hopes to find, Hanson soldiers on looking for his snark. His next step is to use questionable data. He has a database of mortgages written in 2004, which is a little stale for a paper published this year, but that is forgivable. If the subsidy effect was there in 2004 it likely existed in other years.
What is not forgivable is that this data contains interest rate information only for loans where the rate was more than 3% above the equivalent Treasury rate. In other words, Hanson uses a skewed subset of mortgages that were particularly expensive and that make up just under 5% of the full database. Given that the mortgage interest rates were the subject of his study, this alone ought to have stopped him in his tracks.
A further problem is what the database lacks: information on the value of the house. Below the $1M limit, buyers have a strong incentive to borrow as much as the bank will let them have, even if they could otherwise put more cash down. So if you are buying a $500K house, you will borrow as much tax deductible interest money as you can, typically $400K, even if you had the full $500K in the bank. But if you are buying a $5M house, even if you can get a $4M mortgage, it is relatively expensive money and you are more likely to borrow less. Thus you would expect that higher mortgage values tended to have lower loan-to-value ratios, and thus lower interest rates.
Of course, Hanson carries on anyway, running quite a few regressions, some using the latest econometric techniques. And then he declares that the snark has been found. Interest rates are higher for the smaller mortgages, thus lenders are benefiting from the subsidy. “… Lending institutions capture between 9 and 17 percent of the subsidy created by the [mortgage interest deduction] in the form of higher interest rates.”
Except that, according to the numerical data in the paper, bigger loans were significantly more expensive. What Hanson actually found with his flawed data is that, everything else equal, mortgages over $1M had interest rates that were about 0.4% higher. Also, for every $100K of mortgage value, the rate was another 0.04% higher up to $1M, and then 0.01% higher per $100K after that.
Incredibly, it is that little kink in the line, the fact that the rate at which mortgages get more expensive slows above $1M, that allows Hanson to declare victory and spin his tale of banks capturing a subsidy meant for consumers.
To be fair, Hanson is just doing his job, albeit badly. His paper is not for general consumption. It is meant for other academics who understand the tables of numbers and equations he uses and can make their own informed judgments about his work.
SmartMoney, on the other hand, is for a general audience. It repeated to that audience a very poorly supported conclusion from an academic paper. Although I suppose it is possible that they did so knowing the conclusion was bogus because it supported some larger agenda, I doubt it. I believe they just reported on research that they did not understand nor were capable of evaluating.
(As an aside, as I have written before, I am no fan of the mortgage interest deduction. But I do not think we need to invent new reasons not to like it, particularly faux-populist anti-bank ones. All we really need is a way to gently wean ourselves from it without the dislocation that going cold turkey would bring. Gradually lowering the $1M limit over 10 or 20 years still works for me.)