Wall Street or Main Street: Who to Bail Out?

Housing crises are characterized by a sharp increase in foreclosure rates that generates losses to mortgage investors. To preserve the solvency of these investors, governments have historically implemented two policies: a) offer them bailouts (Wall Street), and b) subsidize the mortgage refinancing of households to prevent additional foreclosures (Main Street). The implementation of these instruments involves a trade-off, shaped by two frictions. On one hand, houses lose 20% of their value during the foreclosure process because of larger depreciation due to vacancy and vandalism. If the government offers complete bailouts to investors rather than subsidies to households, the economy has to bear the dead-weight loss of the value lost by foreclosed houses. On the other hand, house prices have an idiosyncratic component, so the government does not have perfect information on individual households’ decision to default. Households have incentives to engage in strategic default to qualify for benefits. A subsidy policy transfers resources to households that were not planning to default in the absence of the policy but avoids the dead-weight loss of foreclosures. I quantitatively assess the welfaremaximizing policy in a heterogeneous agents’ economy and find that a subsidy-only policy would have generated welfare gains of up to 0.4%, measured as the consumption equivalent variation, as compared to the baseline calibration that matches the TARP and HAMP programs implemented during the Great Recession. Households on the left tail of the equity distribution, which benefit the most from a subsidy program, obtain the largest welfare gains. In contrast, a bailout-only policy would have generated a welfare loss of 0.8%. JEL classification: E21, E32, E44, G21, H24 ∗Email: [email protected]. I would like to thank Hal Cole, Jesús Fernández-Villaverde, Dirk Krueger and Guillermo Ordóñez for their helpful comments and guidance. For great comments, I thank Rhys Bidder, Serdar Birinci, Minsu Chang, Vasco Cúrdia, Mary Daly, Galina Hale, Juan Manuel Hernández, Juan David Herreño, Nicolás Idrobo, Òscar Jordà, John Krainer, Marianna Kudlyak, Kevin Lansing, Thomas Mertens, Marı́a José Orraca, Pascal Paul, Nicolas Petrosky-Nadeau, Eugenio Rojas, Daniel Wills, Junyuan Zou, and participants at the Federal Reserve Bank of San Francisco seminars, University of Pennsylvania Macro Club and the Student Research Conference at GWU. All errors are mine. The code for the computations can be found on my github site.