Got Hurt for What You Paid? Revisiting Government Subsidy in the U.S. Mortgage Market

I evaluate the subsidized default insurance policy (implemented through the guarantee for Government-Sponsored Enterprises, GSEs) in the U.S. mortgage market both empirically and theoretically. Empirically, I find that the subsidy has raised mortgage interest rates for all loans strictly eligible for the subsidy (conforming loans), which is contrary to conventional wisdom. I do so by applying regression discontinuity designs and using the exogenous variation in mortgage rates generated by a mandate of U.S. Congress. My empirical strategy circumvents the endogeneity problem in conventional studies. Then I set up a screening model with asymmetric information, which explains my empirical results. Moreover, the model implies that the subsidy has hurt borrowers it was intended to help, even without considering the higher tax burden imposed on borrowers to finance the subsidy. The observed positive jumbo-conforming spread can also be explained by the model through incentive compatibility constraints associated with asymmetric information. My paper cautions regulators against interpreting the observed jumbo-conforming spread as an indication that the subsidy necessarily decreases mortgage rates and benefits conforming borrowers. Moreover, the paper shows that the subsidy raises household leverage, increases mortgage default rate, and ultimately undermines financial stability, calling for deeper housing finance reforms in the U.S. beyond the Dodd-Frank Act.