Abstract
This article uses national data disclosed as part of the Home Mortgage Disclosure Act (HMDA) to examine variations in loan costs based on type of loan, borrower, purpose (purchase, improvement, or refinance), and neighborhood characteristics. Loan costs are generally higher for nonconventional conforming loans with higher levels of credit risks (loans with higher combined loan-to-value, higher debt-to-income ratios, and for investment properties). This implies that product and borrower risk impact loan costs. However, borrower characteristics such as income and race/ethnicity are also associated with differences in loan costs even after controlling for loan characteristics, location, and lender fixed effects. Total loan costs are higher both in dollar terms and as a share of the loan amount for Black borrowers and Hispanic borrowers, and total loan costs represent a higher share of the loan amount for lower income borrowers. These disparities are larger in neighborhoods with higher levels of lender concentration and implicit racial bias. These findings suggest that in addition to access to mortgages and interest rates, loan costs can represent a barrier for access to homeownership with a disparate impact for Black and Hispanic borrowers, which contributes to perpetuate the homeownership gap.
JEL CODES:
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1 Using a nonrepresentative sample of loans collected by the National Consumer Law Center, loan costs were on average 7% of the contract sales price (Collins, Citation2011).
2 The median LTV for all purchase mortgages originated in October 2018 was 95%, and it has persisted near this level for the last 10 years. See Urban Institute, p. 15, https://www.urban.org/sites/default/files/publication/99701/january_chartbook_2019_2.pdf
3 A sample of the loan estimate form is available at https://www.consumerfinance.gov/owning-a-home/loan-estimate/
A sample of the closing disclosure form is available at https://www.consumerfinance.gov/owning-a-home/closing-disclosure/
4 Since 2010, if the final costs on the closing disclosure differ from those shown on the loan estimate by more than the allowable amounts, the lender is responsible for paying the difference (Collins, Citation2011).
5 Results are similar if HMDA 2018 data are used, but 2019 data are used for this analysis because they comprise the most recent prior to the pandemic disruptions, and lenders are likely to have more accurately incorporated new variables in their reporting processes after the first year of implementation.
6 HMDA data capture about 90% of lending activity, but low loan-volume lenders (fewer than 25 qualified closed-end mortgages annually) and smaller and nonprofit lenders are exempt from HMDA reporting requirements (GAO, Citation2021).
7 The following loans are not subject to reporting requirements: Home Equity Line of Credit (HELOC), reverse mortgages, and closed-end loans made primarily for a business purpose or loans secured by manufactured homes but not secured by land.
8 The party of the transaction that selects the settlement and title agent varies by state and county, and in some places the seller makes this selection.
9 Defined as “the spread between the Annual Percentage Rate (APR) and a survey-based estimate of APRs currently offered on prime mortgage loans of a comparable type utilizing the Average Prime Offer Rates’ fixed table or adjustable table, action taken, amortization type, lock-in date, APR, fixed term (loan maturity) or variable term (initial fixed-rate period), and reverse mortgage.” (Federal Financial Institutions Examination Council, CitationNA) The rate spread is therefore a measure of both assessed borrower risk and differences in pricing (due to borrower or lender behaviors), and we are not able to disentangle these components.
10 We thank an anonymous reviewer for suggesting these interactions. We also considered interactions between DTI and CLTV and between individual race/ethnicity and local share of residents of color as well as including local house price trends over the previous 5 years and unemployment rate among individuals 16 years of age and older in the civilian labor force, but these were not significant, and their inclusion did not affect other coefficients in the model.
11 The Avery Federal Housing Finance Agency (FHFA) HMDA lender file is used for the identification of distinct lenders.
12 We also produced estimates separately for home purchases and refinancing, with similar patterns.
Additional information
Notes on contributors
Arthur Acolin
Arthur Acolin is an associate professor in the Runstad Department of Real Estate at the University of Washington. His research focuses on housing finance and access to homeownership.
Rebecca J. Walter
Rebecca J. Walter is an associate professor in the Runstad Department of Real Estate at the University of Washington. Her research focuses on housing policy, housing markets, and crime and place.