TheMReport — News and strategies for the evolving mortgage marketplace.
Issue link: http://digital.themreport.com/i/1463706
M REPORT | 59 O R I G I NAT I O N S E R V I C I N G DATA G O V E R N M E N T S E C O N DA R Y M A R K E T THE LATEST DATA Study: Loan Bias Among Borrowers of Color A new report outlines lending trends in both government and conventional channels when it comes to low- and moderate-income borrowers and neighborhoods of color. T he Community Rein- vestment Act (CRA) was passed in 1977 to fight explicitly racist redlining policies and requires federal bank regulators—the Office of the Comptroller of the Currency, the Federal Reserve, and the Federal Deposit Insur- ance Corporation—to evaluate the performance of banks and thrifts to determine how well each meets "the credit needs of its entire community, includ- ing low- and moderate-income neighborhoods." Independent mortgage banks and credit unions, or nonbanks, are not subject to the CRA. Federal banking regulators are considering revisions to the CRA, and one issue they have raised is whether and how the regula- tions should explicitly take race into account, given the CRA's genesis and purpose. Although current CRA regulations do not consider race, data collected under the Home Mortgage Disclosure Act (HMDA) include information about home mortgage borrow- ers' race or ethnicity, allowing us to examine how well banks perform in lending to borrowers of color relative to the nonbanks not subject to the CRA. Based on an analysis of mortgage loans made to owner-occupant home purchasers, we find that banks substantially underperformed nonbanks in serving borrowers and neighborhoods of color. This shortfall is partially due to banks making fewer Federal Housing Administration (FHA) loans, and government channels tend to lend more frequently to borrowers of color. The analysis shows banks make a lower pro- portion of their loans to neighbor- hoods and borrowers of color through both government and conventional lending channels. Using 2018–19 HMDA data, we found that banks made a lower proportion of loans to borrowers of color—those who identify as Black, Hispanic, Asian, or another nonwhite race. Overall, banks made 23.2% of their owner-occu- pant home purchase mortgage loans to borrowers of color, com- pared with 31.3% for nonbanks. This racial lending gap contin- ues up the income ladder, while middle-income neighborhoods, banks lagged nonbanks by 8.3 percentage points. In upper- income neighborhoods, the gap between banks and nonbanks is 6.3 percentage points. The gap is particularly strong in communities that are both LMI and neighborhoods predominantly of color: 22.2% of bank loans in LMI neighborhoods were made in neighborhoods of color, compared with 26.4% of nonbank loans in LMI neighborhoods of color. Banks make fewer mortgage loans through government chan- nels than nonbanks, which can partially explain the shortfalls. Overall, government channels are significantly skewed toward lend- ing to borrowers of color, as gov- ernment loans tend to allow for more risk layering—a wider com- bination of lower credit scores, higher debt-to-income ratios, and higher loan-to-value ratios. Within each government and conventional channel, banks make a disproportionately lower share of loans to neighborhoods and borrowers of color. We find that banks make 7.2% of their govern- ment loans and 5.2% of their con- ventional loans to neighborhoods of color, compared with 11.7% and 7.3%, respectively, for nonbanks. The data on borrowers of color reveal similar patterns. In govern- ment channels, banks make 28.6% of their government loans to these borrowers; for nonbanks, the share is 37.8%. In both government and con- ventional channels, banks make proportionally fewer loans to LMI neighborhoods of color and borrowers than nonbanks, largely because, within each channel, banks have a narrower credit box. Although bank and nonbank loans tend to have similar loan- to-value ratios, bank loans tend to have higher average credit scores and lower average debt-to-income ratios.