Figure 2: Rate Spreads by Borrower Race/Ethnicity and Sex for Ten Leading National Lenders
Figure 3: Rate Spread Comparisons for Ten Leading National Mortgage Lenders
Figure 4: Range of Rate Spreads by Race/Ethnicity for Ten Leading National Mortgage Lenders
Figure 5: Range of Rate Spreads by Sex for Ten Leading National Mortgage Lenders
Figure 6: Average Rate Spreads in Largest States for Ten Leading National Mortgage Lenders
Figure 7: Average Reported Rate Spreads by State for Ten Leading National Mortgage Lenders
An analysis of 2004 Home Mortgage Disclosure Act (HMDA) data from ten leading national mortgage lenders shows no meaningful differences in the pricing of first lien loans with reported rate spreads to homebuyers of different race and sex. On average, African American and white borrowers paid essentially the same rate spread, and Hispanic borrowers paid less than whites. The rate spreads for men and women were almost identical.
In the two months since lenders have been required to make their 2004 HMDA data publicly available, the press and others have focused on comparing the proportions on loans with rate spreads for different borrower groups.1 Generally, these comparisons conclude that lenders are more likely to report rate spreads on loans to minority and female borrowers than on loans to white and male borrowers. Yet, on the ultimate issue of whether certain groups are victims of discrimination, these comparisons generate more heat than light. The disparities reported may well be the result of borrower credit quality or other loan particulars. Because such information is not publicly available, these analyses are essentially inconclusive.
However, since lenders must report the actual rate spread on loans that exceed certain thresholds,2 detailed pricing data is available for all borrowers with rate spreads. We hypothesized that if, as some have alleged, the disparities in rate spreads mean that minorities and women are being overcharged for loans, overcharging should be evident among loans with reported rates spreads. To test this theory, we undertook a study of borrowers with rate spreads to determine whether minorities or women paid meaningfully higher rate spreads than, respectively, whites or men.
First lien home purchase mortgages were selected for the study, because we believe they comprise the most important category of HMDA loans.3 A home purchase mortgage is made to a borrower who needs a loan to purchase a home, while the other HMDA categories, refinance and home improvement, are comprised of loans that can only be made to an existing homeowner.
Moreover, the most vulnerable Americans are highly represented among home purchase borrowers. Surveys indicate that first-time homebuyers account for 40 percent of all home purchases,4 and that 25 percent of all first time homebuyers are minorities,5 and 21 percent are single women.6 Since the borrowers studied had reported rate spreads-a strong indication that they had deficient or non-traditional credit histories-they are, presumably, particularly vulnerable to overreaching by unscrupulous lenders.
The lenders selected for the study were identified by the American Banker as the "Largest Home Mortgage Originators in 2003." Together, these lenders accounted for nearly 85 percent of all U.S. home mortgage loans.7
Mortgage loans reported with rate spreads by the ten leading national lenders are categorized below according to the primary borrower's race or ethnicity and sex. The average (arithmetic mean) rate spread and number of loans are listed for each borrower category.
| Borrower Category | Average Rate Spread | Number of Loans with Rate Spreads |
|---|---|---|
| All | 3.996 | 90,298 |
| Asian | 3.814 | 1,917 |
| Hispanic | 3.872 | 13,304 |
| White | 4.023 | 49,704 |
| African American | 4.032 | 14,824 |
| American Indian/Alaska Native | 4.084 | 426 |
| Hawaiian/Other Pacific Islander | 4.118 | 290 |
| Race Not Provided | 4.005 | 9,710 |
| Race Not Available | 3.789 | 123 |
| Male | 3.992 | 55,466 |
| Female | 4.006 | 31,622 |
| Sex Not Provided | 3.982 | 3,204 |
| Sex Not Available | 3.335 | 6 |
Figure 2
This study focused on reported rate spreads to white, African American, and Hispanic borrowers, because 97 percent of all rate spread loans with reported race or ethnicity were made to such borrowers. In addition, these borrower categories are at the center of the current debate regarding disparities.8 The study also examined rate spreads to male and female borrowers.
The study found that among first lien home purchase loans with rate spreads, borrower race, ethnicity, and sex were irrelevant. The average rate spread for Hispanic borrowers was 15 basis points lower than the average for whites, and the average rate spread for African American borrowers was less than one basis point higher than the white average. Similarly, the average rate spreads reported for men and women were within about a basis point. These findings are supported not only by comparisons of average rate spreads but also by comparisons of trimmed mean and median rate spreads.
| Borrower Category | Average | Trimmed Mean | Median |
|---|---|---|---|
| Hispanic | 3.872 | 3.774 | 3.620 |
| White | 4.023 | 3.883 | 3.700 |
| African American | 4.032 | 3.929 | 3.770 |
| Male | 3.992 | 3.856 | 3.680 |
| Female | 4.006 | 3.887 | 3.720 |
Figure 3
As illustrated in the graph below, pricing was also consistent among borrower categories when rate spreads were sorted by range. In fact, Hispanic borrowers were consistently less likely than white borrowers to have loans with higher reported rate spreads, and African American borrowers had proportionately fewer rate spreads of 6.5 and above than white borrowers.
The rate spread range data for men and women was similarly consistent. The proportion of female borrowers with rate spreads of 5.5 percentage points or more was no more than .2 percent higher than that of men.
Findings for the five most populous states, set forth below, reinforce the conclusion that borrower race, ethnicity, and sex were not factors in the pricing of first lien home purchase loans with reported rate spreads. In fact, relative to white borrowers, average rate spreads were lower for both African Americans and Hispanics in California, Illinois, and New York. The average Hispanic rate spread was lower than the white rate spread in Texas and the same as the white rate spread in Florida. African American and white borrowers had the same average rate spread in Texas, and the African American average was within nine basis points of the white average in Florida. Women had lower average rate spreads than men in California and Florida and were within 11 basis points elsewhere.
A breakdown of average rate spreads by state, including the percentage point differences between white and minority and male and female borrowers, is set forth on the following page.