Research

Financially Constrained Mortgage Servicers

Revise and Resubmit, Journal of Financial Economics

Cubist Systematic Strategies Ph.D. Candidate Award for Outstanding Research (2018 WFA)

Best Paper Award, Ohio State University PhD Conference on Real Estate and Housing (2018)

Financially constrained mortgage servicers destroyed substantial MBS investor value during the financial crisis through their management of delinquent mortgages. Servicers are obligated to advance to investors monthly payments missed by borrowers. This paper shows that, to minimize this obligation to extend financing to distressed borrowers, constrained servicers aggressively pursued foreclosures and modifications at the expense of investors, borrowers, and future mortgage performance. IV regressions suggest that servicers' financial constraints caused 440,712 additional foreclosures. On average, a servicer's financial constraints were responsible for 20.51% of the total investor value destroyed per defaulted loan–causing aggregate investor value destruction of $84 billion.

Competing for Deal Flow in Mortgage Markets

Under Review

(with Mark J. Garmaise and Gabriel Natividad)

The U.S. mortgage market exhibits competitive instability in which some lenders emerge rapidly from the fringe to substantial market shares. Using inferred discontinuities in application acceptance models to generate local lending shocks, we analyze the impact on a lender of a surge in originations by its competitors. We show that the quickest-growing (not the largest) competitors divert applications and originations from other lenders. Facing a quickly-growing competitor, banks charge higher interest rates, partially due to the increased risk of their loans. Loan performance suffers for other lenders as the quickest-growing competitor's originations increase.

The Perceived Cost of Pension Short(Wind)falls

(with Asaf Bernstein, Mahyar Kargar, Ryan Lewis, and Michael Schwert)

State pensions are underfunded by trillions of dollars, making these deficits larger than all other household non-mortgage liabilities combined, but their temporally distant and uncertain nature make the current burden of such shortfalls unclear. We quantify the incidence of state pension short(wind)falls using a design that i) compares house purchase decisions across state borders where households can choose varying exposure to pension underfunding, and ii) plausibly exogenous returns on pension assets. Consistent with a model of inefficient taxation, we find that a dollar of pension windfalls generates one to two dollars of housing wealth. These findings are not driven by differential property characteristics, even holding within just repeat sales, nor proxies for coincident economic conditions. In fact, the finding that the inclusion of controls for current housing consumption, as proxied by rental rates, do not substantially alter our estimates, suggest that price effects are likely driven by future costs, consistent with the expected timing for the burden of such shortfalls.

Information Exploitation? A Pre-Crisis RMBS Issuer's Private Information

New version coming soon

This paper describes an important borrower risk factor observed privately by the issuer of non-agency RMBS. The private information available to the issuer is drawn from behavioral cues exhibited early in the life of the loan. Mortgage borrowers that make their first six payments at least a day prior to the due date are 14.8 percentage points less likely to become delinquent (equivalent to a 91-point increase in FICO score). This effect is persistent, unobservable at loan origination, and privately observed by the issuer prior to securitization. Both the credit rating agencies and the investor do not appear to be aware of this risk factor. Surprisingly, issuers are quicker to securitize loans with positive private signals rather than less promising loans.