An Analysis of Default Risk in the Home Equity Conversion Mortgage (HECM) Program
47 Pages Posted: 20 Jul 2014 Last revised: 31 Aug 2015
Date Written: August 18, 2015
While reverse mortgages are intended as a tool to enable financial security for older homeowners, in 2014, nearly 12 percent of reverse mortgage borrowers in the federally insured Home Equity Conversion Mortgage (HECM) program were in default on their property taxes or homeowners insurance. Unlike the traditional mortgage market, there were no risk-based underwriting guidelines for HECMs through 2014. In response to the relatively high default rate, a variety of policy responses were implemented, including establishing underwriting guidelines. However, there is a lack of data and analysis to inform such criteria. Our analysis follows 30,000 seniors counseled for reverse mortgages between 2006 and 2011. The data includes comprehensive financial and credit report attributes, not typically available in analyses of reverse mortgage borrowers. Using a bivariate probit model that accounts for selection, we estimate the likelihood of tax and insurance default. Financial characteristics that increase default risk include the percentage of funds withdrawn in the first month of the loan, a lower credit score, higher property tax to income ratio, low or no unused revolving credit, and a history of being past due on mortgage payments or having a tax lien on the property. Our estimate of the elasticity of default with respect to credit scores is similar to that for closed-end home equity loans, but higher than that for HELOCs. We simulate the effects of alternative underwriting criteria and policy changes on the probability of take-up and default. Reductions in the default rate with a minimal effect on participation can be achieved by requiring that participants with low credit scores set aside some of their HECM funds for future property tax and insurance payments, a form of escrowing.
Keywords: Reverse Mortgages, Mortgage Default, Senior Housing, Property Taxes
JEL Classification: R21, R31, G21
Suggested Citation: Suggested Citation