Homeownership is a cornerstone of the American dream; however recent data published by the Consumer Financial Protection Bureau (CFPB) raises important questions about how equally attainable that dream is across ethnicities. In 2013, middle-class African-Americans were 84 percent more likely to get denied for a mortgage than Caucasians.

Despite federal laws established to protect against housing and mortgage discrimination, analysis by Credio and FindTheHome shows persistent relative disparities in mortgage approval rates between ethnicities in the United States.

As expected, the meltdown in mortgage lending caused a significant increase in mortgage denial rates across all ethnicities in 2007 and 2008. Relative disparity between groups, however, was a constant across years.

African-Americans and Native Americans were denied at nearly twice the rate of their Caucasian and Asian-American counterparts in 2013. These disparities held across income brackets.

Tap the play button to see how denial rates changed from 2007 to 2013.

As you can see from the visualization, African-American applicants are denied at higher rates than Caucasian applicants across all income levels.

For example, in 2013, middle-class African-Americans ($50K-$100K) had a denial rate of 35 percent, while Caucasian middle-class applicants only had a denial rate of 20 percent. African-Americans in the highest income brackets (above $150K) were almost twice as likely to get denied for mortgages than Caucasian applicants (28 percent vs. 15 percent).

Disparities between Caucasian and African-American denial rates vary greatly by state. To capture the disparity in denial rates, Credio and FindTheHome calculated a disparity ratio equal to the African-American denial rate divided by the Caucasian denial rate for those with $50K-$100K annual incomes.

For example, a score of 1.0 would signify an equal denial rate for African-Americans and Caucasians, whereas 1.5 would mean that African-Americans get denied 50 percent more than Caucasians. The higher the ratio, the greater the disparity.

Null states on the map above are places where there were fewer than 50 applicants in one of the ethnic income brackets.

In 2013, regional variation in disparity was stark. African-Americans in Utah were 11 percent more likely to get denied than Caucasians, whereas in Nebraska they were 125 percent more likely to get denied.

Income is only one metric used in accepting or denying mortgages: credit scores, debt-to-income ratio, down payment and credit history all play a role.

The CFPB did not include these metrics in released data. The mortgage approval process involves many steps that rely on a combination of computer processing and human discretion.

“… Mortgage approvals are driven by the two automated underwriting systems—Desktop Underwriter for Fannie Mae and Loan Prospector for Freddie Mac,” said Joe Parsons, a senior loan officer at PFS Funding. “Those systems are both colorblind.

“If an applicant’s financial profile (credit score, income, assets and liabilities) is below a certain threshold, he or she will be rejected.”

According to Fannie May, these government-sponsored enterprises that underwrite loans require a minimum credit score of 620 and a debt-to-income ratio of 45 percent or lower to approve an applicant.

Automated underwriting systems evaluate candidates and return one of two results: approve/eligible or refer/eligible. In his 25 years of experience, Parsons has never seen an auto rejection. Those who receive the refer/eligible response are manually underwritten by a loan officer, who could reject them.

The CFPB’s data included general reasons for denial. Credit history was the most common reason for denials of African-American and Native-American applicants.

Gender offers another opportunity for discrimination to occur throughout the lending process.

According to the CFPB, in 2013, women from all ethnicities were denied 21 percent more often than men from all ethnicities (26 percent to 22 percent, respectively). The gender of applicants could be affecting the ethnic disparity we’re seeing.

Without more granular data on the purported financial drivers of the mortgage approval process, determining the source and extent of bias is elusive. However, discrepancies in denial rates across ethnicity, income and region raise important questions for policymakers about the efficacy of federal legislation intended to remove discrimination from the process.