26/05/2026

Marius Gruenewald

Researcher

"Where Have All the Lemons Gone? Signaling and Cosigning in the U.S. Mortgage Market"

Abstract

Housing affordability has declined sharply over the past two decades, and family transfers have become an increasingly important. I assemble the first large-scale dataset linking U.S. mortgage borrowers to their cosigners by combining web-scraped county-level deed records with HMDA data, revealing that roughly five percent of originations involve intergenerational cosigning—well below the 10–20 percent observed in other developed countries. This paper argues that the gap reflects a distinctive feature of U.S. mortgage underwriting: GSE pricing schedules condition on the primary borrower's income alone, even when a cosigner's income is considered for loan approval. A structural model interprets intergenerational cosigning through the lens of signaling under this institutional asymmetry: parents privately observe a child's repayment type and lenders extract a signal from observed parental help. The resulting equilibrium shows two key areas - one where signalling works, one where adverse selection eliminates the signalling power. A central prediction is decoupling: cosigned and non-cosigned rates respond asymmetrically to a lender cost shock. Exploiting the 2022 FHFA fee revision in a regression-discontinuity-in-differences design, I confirm the pattern in the data. Lastly, I embed the signaling structure in a quantitative OLG model that endogenizes loan demand and adds a cash-gift alternative, and find that the lender's signal endogenously cancels roughly 90% of the gross adverse-selection premium. Counterfactual joint debt-to-income pricing—closer to the international standard—raises cosigning rates, thereby homeownership, and closes roughly 20% of the cosigning gap; welfare gains are concentrated at the bottom of the income distribution.

For past presentations, click here.