Leer en Español

In 2022, Futuro Investigates began digging into mortgage application outcomes in New Jersey. Our investigation was triggered by the case of a Latino couple who reported experiencing discrimination from a local bank. It was also informed by prior research suggesting that white Americans are more likely to be granted mortgages than similarly qualified applicants of color.

To identify possible disparities in lending outcomes between white and Latino New Jerseyans, we analyzed hundreds of thousands of conventional mortgage applications filed in the state from 2018-2022. This data is made publicly available under the Home Mortgage Disclosure Act (HMDA), which requires financial institutions to collect and share information on lending activity with the federal government.

Our analysis revealed a gap in mortgage denial rates between Latino and white residents. After removing specific loan outcomes from our dataset (see our rationale for excluding these outcomes in our full methodology), we found that more than 11% of conventional mortgage applications submitted by Latinos resulted in denials—higher than the overall denial rate of 8% and roughly double the rate for white borrowers.

Some of these differences in denial rates could be explained by factors outside of a borrower’s ethnicity. When making mortgage determinations, financial institutions consider multiple variables, including an applicant’s debt-to-income ratio (their monthly debt weighed against their monthly income), credit history and the presence of any co-applicants.

These variables can put Latinos at a disadvantage when applying for a home loan. Many have non-traditional forms of income that translate poorly to lenders’ underwriting systems; others struggle to build credit due to systemic and cultural barriers. Lenders have long argued that these factors—not a borrower’s race or ethnicity—are responsible for disparities in mortgage outcomes.

To better assess the impact of ethnicity on lending decisions, we built statistical models that adjusted for differences in income, debt and several other variables correlated with application outcomes (see our methodology for a complete list of the variables included in our models). Applying these models to our dataset gave us a clearer picture of the relationship between a borrower’s ethnicity and the odds of an application being denied.

A town in Avalon, New Jersey. (Ted Shaffrey/AP Photo)

We found that these adjustments narrowed the disparities but did not eliminate them. While the estimates varied slightly between different versions of the models and the subsets of loans examined, our overall findings suggested that Latinos applying for conventional mortgages faced odds of denial approximately 1.4 times that of non-Hispanic white borrowers. That’s roughly equivalent to raising an applicant’s probability of denial from just under 6% (the raw, unadjusted denial rate for non-Hispanic white borrowers) to 8%.

Our models had limitations. For instance, they couldn’t consider mortgage applicants’ credit scores because that information isn’t included in HMDA’s public data. For precisely this reason, lenders have criticized previous analyses of mortgage outcomes, saying credit scores play a critical role in lending decisions and help explain potential disparities in denial rates.

Since we couldn’t factor credit score data into our analysis, we reran our statistical models while excluding all mortgage applications rejected for credit-related reasons. If credit history plays an outsize role in mortgage decisions—as lenders claim—removing these applicants should theoretically have narrowed the gap in denial rates between Latino and white borrowers. The gap between Latino and non-Hispanic white applicants barely changed under the adjusted model, suggesting that factors beyond credit history contribute to disparities in mortgage outcomes.

Our analysis also sought to determine whether disparities in mortgage decisions varied among New Jersey counties. By examining application outcomes on a county-by-county basis, we found that Latino mortgage applicants in all 21 counties were rejected at higher rates than white residents. In five of those counties—Essex, Mercer, Gloucester, Burlington and Union—the difference amounted to more than six percentage points.

After adjusting for the same application factors as we did in our other models, two of the aforementioned counties rose to the top: Gloucester County and Burlington County. The models suggested that in both counties, Latinos faced odds of denial approximately double that of non-Hispanic white borrowers (see our methodology for a detailed explanation of our county-level analysis).

While our research focused on mortgage denials, we also examined interest rates for approved mortgage applications. To do this, we analyzed rate spread—a variable included in HMDA data that indicates how much higher or lower a borrower’s interest rate is than the average low-risk borrower for a comparable loan.

Our analysis showed Latinos were more likely to receive higher or less favorable rate spreads on mortgages than white borrowers. Using the same subset of conventional loans as our previous models and removing extreme outliers from the dataset, we found Latinos had a median rate spread of 0.35—roughly double the rate spread for white borrowers. That disparity shrank but persisted directionally after adjusting for income and debt-to-income ratio factors. (For a more thorough breakdown of our analysis on rate spreads, please consult our methodology).

Jusleine and Nour Saudi, Lead Producer. (Peniley Ramírez for Futuro Media)

The results of our statistical analysis suggest that ethnicity may play a greater role in mortgage decisions than lenders would like to acknowledge. Even after adjusting for differences in income, debt and other factors, Latinos in New Jersey were more likely to be denied a conventional mortgage than their white counterparts. Moreover, when Latinos did manage to obtain a mortgage, they paid higher interest rates than white borrowers on comparable loans.

The disparities we uncovered in New Jersey are consistent with what researchers have observed nationally. In a comprehensive analysis of 2019 HMDA data, The Markup found Americans of color were more likely to be denied mortgages than white applicants with similar financial backgrounds. (Our methodology was inspired by and partly based on the nonprofit newsroom’s analysis, which used similar statistical modeling). A 2022 working paper from a researcher at the Federal Deposit Insurance Corporation also identified gaps in denial rates between white Americans and applicants of color, even after adjusting for credit score. Finally, a Consumer Financial Protection Bureau report from 2020 found that accounting for credit scores didn’t erase lending disparities for borrowers of color.

*Banner photo by Associated Press. 

The work of Futuro Investigates is made possible by The W.K.Kellogg Foundation, Chan Zuckerberg Initiative, The Tow Foundation, The Geraldine R. Dodge Foundation, Michelle Mercer and Bruce Golden, and Hispanics in Philanthropy.