3 ways to improve the mortgage market

It’s not a great time to be a homebuyer. Expensive houses and high mortgage rates are sidelining many would-be shoppers.

Rates on a 30-year mortgage are lower than this time last year, but it’s anyone’s guess what will happen with a new presidential administration incoming. If high prices and high borrowing costs persist with low inventory, affordability will be hard to come by, market watchers say.

At the MIT Golub Center for Finance and Policy’s annual conference this fall, researchers offered suggestions that could bring relief to one part of the homebuying process — financing.

Like many industries, the mortgage market is benefiting these days from the use of artificial intelligence and big data, whether that’s using additional sources of information to create a more complete picture of a borrower or streamlining the mortgage underwriting process.

“We have an embarrassment of riches now of data to study,” said Paul Willen, senior economist and policy adviser in the Federal Reserve Bank of Boston’s Research Department.

During a panel session, several academic experts discussed research projects that explore how data could be put to work to support homeownership and offered insights on how their findings apply directly to the mortgage market. Among their suggestions:

1. Reexamine the tax assessment and home valuation process.

In the U.S., it’s generally accepted that the higher the value of your home, the more you should pay in property taxes. However, that’s not always the case. New research shows that property tax revenues adjust at a pace that is inconsistent with the growth and decline of property values in the U.S.

“Everyone knows the assessor is supposed to ‘mark to market,’ and basically everybody knows they don’t,” said Willen.

In the research paper “Assessing Assessors,” Lauren Cohen and Huaizhi Chen write about how property tax reassessments spike during positive markets (thereby boosting tax revenue) but don’t show the same sensitivity to negative markets.

“Shocks to public expenses actually drive part of the valuation of assessment properties across the U.S.,” said Chen, an assistant professor of finance at the University of Notre Dame. For example, passing a referendum for the construction of schools or police stations “immediately makes it more likely that property prices are going to be revised upward within the next three years,” he said.

On the flip side, prices tend not to be revised downward in weak markets because that would decrease the amount of property tax revenue, which is many municipalities’ single largest source of discretionary revenue, the authors write in their research paper.

Golub Center executive director Edward Golding suggested that with the help of big data, it would be possible — and beneficial — to merge the house appraisal and tax assessment processes into one value calculation.

“We’d have a fair process if we integrated the two,” said Golding, a senior lecturer at MIT Sloan. “Not easy, but I think the paper suggests it might be beneficial.”

Separately, Cohen and Chen found that local tax assessors have tax assessments on their own properties that are significantly lower than those of neighboring properties.

2. Rethink how mortgages are designed.

Homebuyers who refinance their mortgages more often end up paying less overall compared with people who rarely refinance. Despite this, the most popular mortgage contract in the U.S. is a 30-year fixed-rate mortgage. New research shows that approximately 20% of unconstrained U.S. borrowers would benefit financially from refinancing but fail to do so.

“People don’t optimally refinance,” said David W. Berger, associate professor of economics at Duke, even when they are presented with information showing the move to be favorable. “It helps, but not completely. There are people who have [interest rates] that are, say, 3 percentage points higher than the current mortgage rate. That is thousands of dollars they’re leaving on the table.”

In “Refinancing Frictions, Mortgage Pricing and Redistribution,” Berger and co-authors Konstantin Milbradt, Fabrice Tourre, and Joseph S. Vavra present a model for assessing what would happen if mortgages were automatically refinanced when rates declined, without requiring active borrower intervention.

While this scenario encourages refinancing among borrowers who are slow to react, the authors caution that doing so can also have unintended consequences: Lenders make money when borrowers continue at higher interest rates. To recoup those losses in an automatic-refinance plan, they may charge higher interest rates on newly originated mortgages and thus reduce mortgage credit access for a number of borrowers.

Another idea the authors explored: a mortgage contract that prevents people from refinancing during the first few years of a mortgage. The aim would be to cut down on the repeated “churn” of refinancing and reduce the dead-weight costs associated with mortgage origination.

It might seem that introducing an initial limit on refinancing would hurt borrowers, the authors write in their research paper, but since these contracts would be temporary, they wouldn’t permanently lock borrowers into high rates. This approach could result in savings that would ultimately pass through into mortgage rates at origination, thereby benefiting borrowers more broadly.

3. Increase reliance on algorithms to benefit lower-income borrowers.

Algorithmic underwriting is making waves in the mortgage market. Traditionally performed by humans, underwriting has become more and more automated in recent decades. Advantages include fewer errors and less exposure to agency conflicts, but the downside is that algorithmic underwriting cannot collect and interpret “soft” information — a constraint that may hurt borrowers who have unconventional income and an opaque credit history.

A paper co-authored by Janet Gao, associate professor in finance at Georgetown University, studies the impacts of algorithmic underwriting in a high-risk segment of the U.S. mortgage market. The research examines a Federal Housing Administration policy that transitioned human underwriting to human-augmented algorithmic underwriting for high-leverage borrowers with low credit scores.

The policy change spurred lenders to increase the amount of money they were lending to such borrowers, which allowed the homebuyers to migrate to higher-quality neighborhoods (defined by the quality of the school district) without experiencing significantly higher default rates.

“We show that increasing reliance on algorithmic underwriting can increase the financial inclusion toward lower-income households without sacrificing too much on the risk control,” Gao said. “Algorithms can alleviate many lending frictions related to humans. They’re better at processing complex information, free of agency conflicts, and they don’t face the same level of capacity constraints.”

Gao’s co-authors on the paper “Algorithmic Underwriting in High Risk Mortgage Markets” are Hanyi (Livia) Yi and David Zhang.

Watch: Golub Center for Finance and Policy 11th Annual Conference