How Do Mortgage Refinances Affect Debt, Default, and Spending? Evidence from HARP (with Andreas Fuster; American Economic Journal: Macroeconomics, 2021)
We use quasi-random access to the Home Affordable Refinance Program (HARP) to identify the causal effect of refinancing into a lower-rate mortgage on borrower balance sheet outcomes. Refinancing substantially reduces borrower default rates on mortgages and other debt. Refinancing also causes borrowers to expand their use of debt instruments, such as auto loans, home equity lines, and other consumer debts that are proxies for spending. Borrowers that appear more constrained ex ante grow these debts more strongly after refinancing but also pay down credit card balances by more. These borrowers also have lower take-up of the refinancing opportunity.
The Measurement and Behavior of Uncertainty: Evidence from the ECB Survey of Professional Forecasters (with Robert Rich, Joseph Song, and Joseph Tracy; Journal of Applied Econometrics, 2016)
We examine matched point and density forecasts of output growth, inflation and unemployment from the ECB Survey of Professional Forecasters. We construct measures of uncertainty from individual histograms, and find that the measures display countercyclical behavior and have increased across all forecast horizons since 2007. We also derive measures of forecast dispersion and forecast accuracy, and find that they are not reliable proxies for uncertainty. There is, however, evidence of a meaningful co‐movement between uncertainty and aggregate point predictions for output growth and unemployment. These results are robust to changes in the composition of the survey respondents over time.
Equilibrium Mortgage Design with Heterogeneous Borrowers (Revise & Resubmit, Review of Corporate Finance Studies)
Conventional mortgages price discriminate: by refinancing, sophisticated borrowers pay less than do unsophisticated borrowers, even if their original mortgage terms were identical. I present a model of the mortgage market in which such discriminatory products are offered by lenders to maximize rent extraction from borrowers, who have heterogeneous search costs during mortgage origination. This can justify why mortgages that, for example, refinance automatically are not observed, despite the large and varied social benefits they could generate, as argued in previous literature. The menu of interest rates and "points" that borrowers are typically offered can also be rationalized by this mechanism.
The Joint Dynamics of Moving House and Mortgage Refinancing (working paper)
Market interest rates affect homeowners' decisions of when to move and when to refinance, but it has been standard to study those two options in isolation. This paper studies their interaction with a dynamic model of a household with both options when interest rates and idiosyncratic housing match quality are stochastic. The options "warp" each other and their coexistence can create an inaction region relative to models in which they are studied in isolation. The nature of their interaction depends on the level of interest rates: when interest rates are low, the option to refinance deters moves, but when interest rates are high, the option to refinance encourages households to move, as they know they can subsequently refinance. An implication of this is that streamlined refinancing opportunities can lessen volatility in mobility stemming from interest rate fluctuations, a concern in the current `"lock-in" episode in the housing market, which is explored with a calibrated version of the model.
-
Recipient of Best Paper award at 2025 Liberal Arts Macro Conference (under the title “Mortgage Lock-In and Home Sales Volume Dynamics”)
The Home Sales Volatility Puzzle: An Empirical Exploration (working paper)
The recent housing cycle in the United States saw a large swing not only in home prices but in the number of home sales as well. This uses a comprehensive dataset on US home sales to investigate two popular explanations for the cyclicality of selling activity: “house lock,” whichconjectures that falling prices cause down-payment constraints to bind and prevent current homeowners from selling their homes; and nominal loss aversion, which proposes that cognitive frictions prevent homeowners from selling when doing so would not garner a price as high asthe one they originally paid for the house. I find that while there is evidence that both of these mechanisms are active at the household level, they explain a fairly small portion of the decline in sales from boom to bust: likely no more than 10%.