This paper examines contemporary trends in mortgage refinancing among low-income and minority homeowners.  Interest rates on mortgages in the U.S. are at historic lows; however, analysis of data from the Home Mortgage Disclosure Act (HMDA) for 2009 – 2012 shows that rates of mortgage refinancing have been substantially  lower among low-income and minority homeowners.  Indeed, refinance activity appears to be least prevalent among low-income and minority homeowners who bought during the subprime boom and are currently paying off higher-priced loans – on average, these borrowers are in loans with interest rates that are more than 4 percentage points above the current rate on a 30-year, fixed-rate mortgage.  As a result, subprime lending has led not only to a disproportionately negative impact on low-income and minority communities as a result of increased foreclosure rates, but also through the longer-term asset stripping resulting from the high cost of credit for these borrowers.

According to the option-based model of mortgage termination, borrowers currently paying off subprime and/or higher-priced loans should be exercising their option to refinance.  However, low-income and minority borrowers may face significant barriers to doing so, including credit, collateral, and knowledge constraints.  To shed light on racial and income differences in refinance activity, this paper examines the factors that are correlated with differences in the propensity to refinance.  Specifically, what is the relative importance of borrower-level (e.g. credit status, ability to repay) and neighborhood-level (e.g. house price changes, lender market segmentation, Community Reinvestment Act (CRA) eligibility, and concentration of foreclosures) characteristics in predicting the likelihood of a refinance? To answer these questions, we rely on a unique dataset that merges loan performance data from a large, mortgage servicing database with data on borrower race and income from HMDA, as well as data on lenders (from the HMDA Lender file) and neighborhood level variables (including zip code level house price data and socio-economic and demographic variables from the U.S. Census).  We specify a multinomial logit model of mortgage terminations that accounts for the joint nature of the borrower’s options to prepay or default to estimate the influence of credit and income constraints, collateral constraints, the institutional lending environment, and neighborhood level characteristics on the likelihood that a borrower refinances their mortgage.  

The findings of this research are important for contemporary policy debates bearing on access to credit and lending for lower-income and minority households. Disparities in refinance rates have significant implications for both household wealth and financial well-being over time, and will shape the degree to which homeownership leads to positive spatial, social and intergenerational externalities. Moreover, there is evidence that mortgage refinancing under the Home Affordable Refinance Program (HARP) has led to reductions in default risk, which suggests that reducing barriers to refinancing may be an avenue for reducing the social and economic costs associated with foreclosures. More broadly, a better understanding of the factors that shape mortgage borrowing decisions among low-income and minority homeowners can inform the development of more equitable housing finance policies and programs going forward.