In early 2020, the stability of the mortgage market seemed at grave risk. Who would buy a home during a pandemic? How would lenders process paperwork while offices were closed? How would mortgage servicers find the funds to pay investors for loans in forbearance?

Ultimately, however, mortgage activity soared during the balance of the year:

  • Four trillion dollars in new mortgages was originated in 2020, a record, and far higher than in any year since 2003.
  • Rates on 30-year fixed-rate mortgages fell below 3 percent, a historic low.
  • Profitability for the largest U.S. mortgage lenders soared.

Mortgage markets overcame stubborn headwinds and limitations during their upward march, but they also showed signs of significant abnormalities. In their paper, “How Resilient Is Mortgage Credit Supply? Evidence from the COVID-19 Pandemic,” Andreas Fuster of the Swiss National Bank and CEPR, Aurel Hizmo of the Board of Governors, Lauren Lambie-Hanson of the Philadelphia Fed, James Vickery of the Philadelphia Fed, and Paul Willen of the Boston Fed and the National Bureau of Economic Research demonstrate that such abnormalities produced a sharp increase in the markups charged by mortgage lenders, meaning that lower market interest rates were only partially passed along to mortgage borrowers. The authors also find evidence of a tighter credit supply in the parts of the U.S. mortgage market that cater to higher-risk borrowers.

To study the mortgage supply pipeline, the authors gathered data on mortgage rate sheets, rate locks, and loans originated across 280 metros as well as rural areas. The data include a rich set of borrower characteristics that influence the interest rate on a loan (such as the borrower’s FICO credit score, discount points, and credits, and whether the borrower is an investor or an owner-occupant). These loan-level data are complemented by financial market data on the prices and yields of mortgage-backed securities, macroeconomic data, and a variety of other data sources.

When studying interest-rate behavior throughout the pandemic, the authors looked at the difference (or spread) between the rate on mortgages and the yield on 10-year U.S. Treasury securities. (The Treasury rate acts as the benchmark risk-free rate for long-term loans in the U.S.) They find that the spread quickly widened in late February, peaking in March before gradually drifting back to prepandemic levels by November. (In other words, mortgage rates declined by less than the baseline rate represented by U.S. Treasuries during the pandemic.) The authors show that, unlike in the 2008 financial crisis, this widening spread between mortgage and Treasury rates is more than accounted for by a sharp increase in the markup charged by lenders as compensation for originating the mortgage.

What explains this sharp rise in the price of intermediation in the mortgage market during the pandemic? The authors' findings suggest that, among other constraints, it was unusually difficult for lenders to expand capacity amid the challenging operating environment presented by COVID-19. Bolstered by a concurrent surge in loan demand, markups edged higher. In effect, mortgage credit supply was unusually inelastic during the pandemic compared to prior periods of high loan demand.

For example, the hiring and training of new mortgage employees such as loan processors and loan officers was particularly challenging during the pandemic, because of the shift to working from home as well as disruptions to the licensing process. (The administration of licensing exams, for instance, paused when testing sites closed.) As a result of the limited pool of qualified labor, the authors show, mortgage employees were hired at a pace that was slow relative to the demand for their services (as measured by the volume of job postings). And labor costs rose as lenders competed for talent, a spike that is unusual during periods of heavy growth in lending. During typical lending booms, labor costs per dollar of loan volume typically fall (due to scale economies), but not so during the pandemic. Instead, labor costs remained unusually high.

Complications in the loan-origination process seem to have constrained supply, too, particularly early in the pandemic. Steps in the mortgage and home-purchase process that typically require face-to-face interaction, such as the property appraisal and the loan closing, became more difficult. And it became more cumbersome to verify borrower income because of the quickly deteriorating labor market and limited access to records as employers shifted to remote work.

With the above headwinds in mind, the authors ask if lenders who rely less on paper-based loan applications (referred to as fintech lenders because they have a fully online origination process) gained market share during the pandemic. Their analysis suggests that fintech lenders indeed gained market share, most markedly among complicated, documentation-heavy loans such as purchase mortgages (that is, mortgages used to buy a home rather than refinance an existing loan) and loans to borrowers with lower credit scores.

As part of their investigation into mortgage availability during the pandemic, the authors thought carefully about possible alternative explanations for the rise in mortgage markups. They tested, for example, whether reduced competition (via the concentration of market share among a small group of big lenders) prevented mortgage rates from falling more. They find no quantitative evidence to support this hypothesis, however.

The authors also investigated whether the heightened risk of delinquency due to forbearance programs and the challenging economic environment were reflected in higher mortgage rates. They find no evidence that this was the case among low- to moderate-risk loans, but rates did rise for borrowers in the Federal Housing Administration (FHA) market, who are often first-time homebuyers and who default more frequently. Rates also rose in relative terms in the jumbo market (measured as the spread between rates on jumbo loans and rates on smaller but otherwise identical loans), where mortgages do not carry a government-backed guarantee. Finally, rates were higher for “super-conforming” loans, which were less likely to be purchased by the Federal Reserve through its quantitative easing program.

“How Resilient Is Mortgage Credit Supply? Evidence from the COVID-19 Pandemic” highlights general lessons about the frictions that can prevent borrowers from receiving lower mortgage rates during periods of falling market interest rates. During periods of high demand, capacity constraints crimp credit supply and ultimately contribute to higher interest rates. The particular features of the COVID-19 pandemic magnified these capacity constraints and also led to higher rates for borrowers on the lower end of the credit scale and those not eligible for government guarantees. Looking toward the future, the authors consider possible changes in mortgage contract design: Perhaps the incomplete pass-through of interest rates to mortgage borrowers can be alleviated if extant mortgages are automatically refinanced (that is, their rates are lowered) during an economic downturn. Such streamlined, automatic processing would also free up resources for originating new loans, increasing the likelihood that home purchasers get lower mortgage rates.