Essays in Mortgage and Household Finance
Name
Wilson-wilsonjr-PhD-Management-2026-thesis.pdf
Size
1.71 MB
Format
Adobe PDF
Checksum (MD5)
7481a863d4b6ebe596856802e23643e3
Author(s)
Wilson, John
Advisor(s)
Lucas, Deborah
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
This thesis comprises three chapters on mortgage and household finance. The first chapter consists of joint work with Deborah Lucas and Paul Willen and examines the impact of the cessation of LIBOR on the mortgage market. Approximately $800 billion of adjustable-rate mortgages (ARMs) indexed to LIBOR remained outstanding when LIBOR was published for the last time on June 30, 2023. For the many ARMs that did not have contracts which specified an alternative index, The Federal Reserve mandated replacing LIBOR with the corresponding Term Secured Overnight Funding Rate (Term SOFR) plus a fixed spread. The conversion rule created the possibility of significant wealth transfers between ARM borrowers and lenders because the mandated spreads neglected to account for the difference between rate spreads in the ARM market and in the LIBOR market. To evaluate the size and incidence of realized transfers, we develop a valuation framework that takes the historical spread between LIBOR-indexed and Treasury-indexed ARMs as the best estimator of the neutral spread adjustment, and that projects future cash flows for the affected universe of ARMs using auxiliary models of prepayment, default, and term structure dynamics. We estimate that mortgage lenders benefited significantly at the expense of borrowers, with a total present value transfer of $248.73 million as of the conversion date, representing 0.49% of the balance of our sample ARM pool. The size of individual transfers varied with mortgage and borrower characteristics. Future unintended wealth transfers could be made less likely by routinizing the inclusion of appropriate fallback language in floating rate contracts offered to households, and with greater transparency on the part of regulators about the redistributive consequences of their policy choices. The second chapter examines the determinants, performance, and policy significance of ARMs in the U.S. mortgage market using detailed loan-level data. I show that ARMs are disproportionately chosen by higher-income borrowers financing larger homes, with affordability constraints—rather than down-payment limitations—driving selection when fixed-rate mortgage (FRM) rates are high. Contemporary ARMs, particularly 5/1, 7/1, and 10/1 products, exhibit default rates comparable to or lower than FRMs once differences in origination cohorts are controlled for, challenging the perception that ARMs are inherently riskier. ARMs provide more symmetric pass-through of monetary policy than FRMs due to the high proportion which is floating, mitigating mortgage interest rate lock-in effects in aggregate. I also document a “precautionary refinancing” behavior in the year before ARMs float, reflecting borrower aversion to interest rate risk even at the expense of higher payments. These results imply that regulation should focus less on phasing out risky products and more on better credit screening and transparent contract design and that lenders can strategically offer ARMs to expand credit access for DTI-constrained but creditworthy borrowers. The third chapter tests a prediction of the Permanent Income Hypothesis that households who are illiquid may wish to borrow against future income when there is a positive shock to their expectations of future income. I develop a simple two-period model to demonstrate this prediction and then test it using Equifax data. I demonstrate that households identifying as Democrats experience a strong, positive shock to their expectations about their future real income around the 2020 presidential election. A difference-in-difference analysis finds that in response households in geographies with a high Democrat voter share were 0.08% more likely to take on debt to buy a car, and that their outstanding auto loan balance increased by $104 on average relative to households in strongly Republican geographies. Compared to the rate at which households in geographies with a high Democrat voter share purchased cars via loan prior to the election, this 0.08% increase represents a 11.69% increase in the purchase rate. I validate my results by finding a similar result holds for installment loan purchases. I show that this result is robust to my empirical assumptions.
MIT Department
Sloan School of Management
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
