NCRC 2020 Home Mortgage Report: Examining Shifts During COVID-19

January 2022

Image: ©oum

— Key Takeaways

  • The private mortgage lending market continued to inadequately serve most households of color.
  • The gap in homeownership rates between Black and White families remained near a 120-year high point.
  • Refinance lending exploded during the pandemic and disproportionately flowed to non-Hispanic White, Asian Indian and Chinese borrowers.
  • Other BIPOC communities saw their share of mortgage lending shrink.
  • Over decades, the private mortgage market and laws intended to expand access to credit have failed to achieve equitable home ownership rates or increase family wealth in communities of color.
  • Policy makers, community leaders and lenders themselves should rally around bold new approaches, race-specific goals and performance requirements.

— Key Findings

  • Non-bank mortgage companies extended their lead over banks in originating home purchase loans overall, and were by far the primary source of credit for LMI and non-Asian BIPOC communities.
  • In 2020, the 25.5 million HMDA records had 14.5 million loan originations plus another 2.9 million records of loan “purchases,” where loans were originated by a third party and bought by the lender.
  • Average refinance interest rates declined from 4.89% in 2018 to 3.09% in 2020.
  • Low interest rates drove a massive increase in homeowners seeking refinances, doubling from 7.1 million applications in 2018 to 15 million in 2020.
  • Loans lacking race and ethnicity data have grown since 2018 to 15% of all applications and 75% of all purchased loans in the 2020 HMDA release.
  • 15 of the top 20 loan purchasers reported race and ethnicity on 0% of the loans they purchased.
  • Reporting of disaggregated race and ethnicity data by borrowers reached a new high of 61% of eligible applicants. Yet home purchase loans lacking race and ethnicity data increased 1.9%.
  • There has been little change in the share of same-sex lending in HMDA across all loan purposes. Less than 2% of loans in the HMDA dataset in 2020 had a same-sex pair of borrowers.
  • Since 2007, there has been a 2.4% increase in market share for BIPOC home purchase borrowers.
  • Most of the increase in BIPOC home purchase lending was due to expanding numbers of Hispanic and Asian applicants, who increased by 1.5% and 1%, respectively. Black borrowers decreased by 0.1%.
  • Asian Indian borrowers were responsible for Asian gains in refinance lending in 2020. Because of their impact on the market, the share of Asian borrowers increased from 29% to 36% from 2018 to 2020.
  • In 2018, 27.5% of refinance loans went to an LMI homeowner, and in 2020 that figure fell to 19% even though about 43% of the population were in LMI households.
  • Originations to majority-minority neighborhoods fell from 19.1% of all loans in 2018 to 18.5% in 2020.
  • Loans in LMI neighborhoods declined from 16.5% in 2018 to 13.7% in 2020.
  • In 2018, more than 45% of refinance loans were to older adults, and in 2020, that figure fell to 31.5%.
  • There was an alarming increase in the number of loan records that lacked demographic data on race and ethnicity. This was thanks to a loophole in reporting requirements which allows the removal of demographic data from purchased loans.
  • The number of reverse mortgages expanded sharply during the pandemic. This might suggest more broadly that older adults were under intense financial pressure.
  • Black and Hispanic borrowers bought less valuable homes than White (and some Asian) borrowers, and they paid more to do so.
  • Despite predictions that the growing Hispanic population would increase their share of the housing market, there was no evidence of this shift.

— Recommendations

  • The Community Reinvestment Act should be expanded to explicitly evaluate racial/ethnic equality and how banks serve people of color and majority-minority communities.
  • Congress, bank and housing regulators, community leaders and lenders should rally around a 60% homeownership goal for African Americans and Latinos, and take the necessary actions to achieve that goal.
  • Expand support for first-generation buyers, with a focus on helping families with down payments and closing costs to remove the largest financial barrier to start building wealth via homeownership.
  • Increase funding to the National Housing Trust Fund and Capital Magnet Fund.
  • Expand fair housing enforcement, including HUD’s Affirmatively Furthering Fair Housing and Disparate Impact Rules.
  • Strengthen and better evaluate racial and ethnic representation and investment in financial institutions.
  • The Consumer Financial Protection Bureau (CFPB) should engage the public, consumer groups, and lenders to assess best practices for encouraging loan applicants to volunteer demographic data.
  • The CFPB should require lenders that report loan purchases to report any demographic data collected by the loan originator. Close the loophole that allows no reporting of demographic data for purchased loans.
  • Expand the disaggregated race options that are available to borrowers to include alternatives to African American, Native American and White to improve our new understanding of the nuances in racial disparity in the mortgage market. Add options for borrowers to identify in other ways as well; for example, veteran status, sexual orientation and educational attainment.
  • In order to identify predatory practices such as churning, steering or overcharging, the CFPB should include the universal loan identifier and borrower credit score in the public HMDA data.
  • The CFPB and the prudential regulators should explore using the disaggregated race and loan cost data in HMDA to better understand issues such as steering, redlining and discrimination in the mortgage market.
  • The CFPB should release HMDA each quarter on a rolling basis. Lenders that take more than 60,000 applications per year are required to report quarterly and more current data would greatly expand the impact of HMDA with no additional burden on reporting entities.
  • The Federal Reserve should deepen its knowledge and understanding of the relationship between the racial wealth and homeownership gaps and mortgage interest rates. Ensuring a steady flow of safe, affordable mortgage credit is key to the Federal Reserve’s goal of maximizing employment.
  • The interest rate for fixed rate mortgages is set based on the yield of the 10-Year Treasury note. This rate can be impacted by economic conditions, including the pandemic, climate change and natural disasters. The Federal Reserve should consider how fluctuations in this market impact the flow of mortgage credit, especially to LMI and BIPOC communities, and find a method to insulate the housing finance system from this disparate impact.

Jason Richardson, Senior Director, Research & Evaluation
Joshua Devine, Director of Racial Economic Equity
Jamie Buell, Racial Economic Equity Coordinator
Dedrick Asante-Muhammad, Chief, Membership, Policy and Equity

This report was produced with support from the W.K. Kellogg Foundation.


In this paper the National Community Reinvestment Coalition presents a broad and detailed analysis of 2020 Home Mortgage Disclosure Act (HMDA) data. Our analysis reveals how the private or conventional mortgage lending market is failing to serve economically insecure households of color.  This re-enforces the racial divide in homeownership which is a pillar of the growing racial wealth divide.

Whites and Asian Americans were the major racial/ethnic categories whose demographic representation was about equally reflected in their mortgage lending. In 2020 whites originated 65% of all mortgages, close to their representation of the adult population of 64%. Asian Americans, who are 6.1% of the population, were close to their demographic representation, originating 6.5% of home lending. Latinos share of the home mortgage lending market was about half of their demographic representation with 9% of home mortgage lending for 17% of the adult population. Blacks and Native Americans share of the home lending market, 5.2% and .4% respectively, was less than half their demographic representation. 

Latinos, African Americans and Native Americans are being even more poorly served by conventional, private lenders than what has been previously understood. Almost 60% of Black home purchase applications are through federally-backed FHA, VA and RHS loan programs, and for Latinos it is about 50%. The conventional/private mortgage market, even with government backed loans, is maintaining the homeownership disparities that are at the center of the American racial wealth divide.

The recently released HMDA data also shows that the conventional/private mortgage market is under-serving low- and moderate-income (LMI) neighborhoods as a whole, providing only 13.7% of loans to LMI neighborhoods, although these areas represent 30% of neighborhoods. NCRC has proposed a series of policy reforms to help address these issues and make homeownership as asset of equality verse its current role as carrying the effects of historic discrimination into the present and future.

A few of NCRC recommendations include:

  • CRA should be expanded to explicitly evaluate racial/ethnic equality and how banks serve people of color and majority-minority communities
  • NCRC has called for the nation to rally around a 60% homeownership goal for African Americans, Latinos and Native Americans and take the necessary action to achieve that goal
  • Strengthen and better evaluate racial and ethnic representation and investment in financial institutions
  • CFPB should require lenders that report loan purchases to report any demographic data collected by the loan originator. End the loophole allowing no reporting of demographic data for purchased loans.


As this report highlights, there is much important data in the Consumer Financial Protection Bureau release of HMDA data. It is up to the financial industry, nonprofit advocates, community activists, regulators and government officials to analyze and act upon this important information. Hopefully this NCRC paper will help instigate action to create a nation where homeownership is a more broadly held and progressive asset.

Introduction and Overview

Home equity – the difference between the value of a home and the amount of mortgage debt on the home – is an important component of overall household wealth. Corelogic

In June 2021, the Consumer Financial Protection Bureau (CFPB) released the 2020 data from the Home Mortgage Disclosure Act (HMDA). This valuable dataset is the most comprehensive and detailed glimpse into mortgage lenders’ ability to provide credit to American homebuyers and homeowners. In this report, NCRC reviews 2018 through 2020 HMDA data and discusses trends in lending, the impact of COVID-19, the refinance wave, and interesting changes in lending among disaggregated racial groups. The impact of the racial wealth divide on the way that borrowers encounter the mortgage market is important to understand in order to address the growing disparities in homeownership and wealth. HMDA is also one of the very few datasets that offers a granular look at family wealth in America over time, by offering snapshots of their home equity.

Black and Hispanic families are far less likely to own homes compared with White families. NCRC’s analysis shows that the gap in homeownership rates for Black and White families is now near its highest point in 120 years. From home appreciation rates to neighborhood valuation, researchers have categorically documented the discriminatory origins and outcomes of our mortgage finance system. The assets of Black and Latino households are largely concentrated within their home equity. 47% and 51% of respective Black and Latino households’ wealth portfolio is concentrated in their home equity, compared to White households’ wealth portfolio in which 25% is concentrated in home equity. The ongoing racial/ethnic homeownership divide and growing concentration of wealth amongst the wealthy maintains the racial wealth divide today and will continue to expand it into the future.

Data such as HMDA allows regulators, consumer advocates and the public to examine lenders for evidence of disparate impact stemming from structurally racist elements of our housing finance system. Redliningsegregation and displacement have imbued a legacy of discrimination into our housing market and on the health of our communities. By advocating for greater transparency and access to data on lenders and lending and strengthening homeownership rates among economically marginalized groups, NCRC seeks to address the growing racial wealth divide. To that end, NCRC is advocating for the collection of small business loan data as required in the Dodd-Frank Wall Street Reform and Consumer Protection Act, and we continue to support greater data collection and transparency under the Community Reinvestment Act (CRA). The CFPB has proposed a rule to implement Section 1071 of the Dodd-Frank Act, which requires the collection of the race and gender of small business applicants. The proposed rule is comprehensive. However, we will be asking the agency to include some categories of lenders that are proposed to be exempted. NCRC also supports the efforts of the CFPB to make Section 1071 consistent with HMDA data in important areas such as action taken on applications and disaggregated racial and ethnic categories. NCRC has also created a 60% homeownership goal for African Americans, Latinos and Native Americans and is asking for the public and private sectors to work toward those goals.

In this report, we will review trends in mortgage lending from 2018 through 2020. In 2018, the CFPB took over the collection and reporting of HMDA data and updated it to a new format with many new variables. This HDMA report is the first time we have had the opportunity to review three years of this new data, including information on borrower age, disaggregated race and ethnicity, loan interest rates, closing costs and features. In this report, we looked at those loans along multiple dimensions including the race, ethnicity, gender and age of the applicants.

In addition, we were particularly interested in the interaction of BIPOC applicants with low- and moderate-incomes and in LMI and majority BIPOC census tracts (neighborhoods). Special attention was placed on how BIPOC and LMI people experienced the mortgage market during a year in which COVID-19 ravaged the health and economic stability of communities, particularly communities of color.

To explore this data in greater detail in your community, NCRC offers the Fair Lending Tool as an NCRC member benefit. Find out more about the tool with this sample version, view this webinar, or contact NCRC if you are already a member and want to learn more about the tools available in the NCRC member hub. NCRC also manages a public Slack channel for researchers that use HMDA or other related data to discuss data, ask questions or seek emotional support. If you would like more information about this resource, please contact Director of Research Jason Richardson.

2018-2020 General Trends of HMDA Data

The 2020 HMDA dataset includes 25.5 million individual records. The CFPB estimates that HMDA covers 88% of mortgage applications each year. In 2020, the 25.5 million records had 14.5 million loan originations plus another 2.9 million records of loan “purchases,” where loans were originated by a third party and bought by the lender.1

The number of home purchase loans continued to increase in 2020, continuing a trend since 2012. However, the real driver of the massive increase in mortgage activity was refinance lending driven by low interest rates. As the table shows, the average interest rate for refinance loans in 2020 was just 3.09%, a decline from 4.89% in 2018. Overall, the 2020 mortgage market appears to have been strong, with a steady increase in home purchase lending and roughly twice the number of refinance loans as in 2019. This was despite the global COVID-19 pandemic. This seeming disconnect between the economic hardships caused by the pandemic and the lack of discernible effect on the mortgage market may be due to the relatively privileged position homeownership offers in terms of tenure. Or, mortgage forbearance and low interest rates might be masking a timebomb of foreclosures. Homeowners in 2020 had available to them broadly effective (so far) forbearance programs if their loans were among the vast majority of loans securitized by mortgage giants Fannie Mae and Freddie Mac (with separate provisions for HUD/VA owned loans). In addition, homeowners may be less likely to be impacted by COVID-19 for various environmental, economic or job market factors. Future research could explore this dichotomy, but if renters have been economically damaged disproportionately by COVID (as seems likely), they will also find the barriers to homeownership to be even higher in the years to come. Careful attention should be paid to the outcome of forbearance programs, including enforcement of servicing guidelines and consumer outreach to prevent unnecessary foreclosures.

Overview of HMDA collection on race and ethnicity data

According to the provisions of the Dodd-Frank Act, mortgage lenders are expected to collect extensive data on the race, ethnicity, age, gender and income of applicants (and co-applicants) in HMDA. This data allows applicants to identify five different ethnic and racial identities. NCRC published a methodology that we used in this report to determine the borrower’s race and ethnicity. The choices offered to borrowers are as follows.

For ethnicity, if the borrower selects Hispanic or Latino, they are offered three other choices. They can enter a freeform response that is coded as “other Hispanic or Latino” in the public HMDA data. They can select up to five ethnicities should they choose to. Other choices include “Not Hispanic or Latino,” and “Information not provided by applicant in mail, internet or telephone application,” and “Not applicable.”

Hispanic or Latino

  • Mexican
  • Puerto Rican
  • Cuban
  • Other Hispanic or Latino (borrower writes in another answer)

For race, borrowers can select up to five of any of the following options.

American Indian or Alaska Native


  • Asian Indian
  • Chinese
  • Filipino
  • Japanese
  • Korean
  • Vietnamese
  • Other Asian (borrower writes in another answer)

Black or African American

 Native Hawaiian or Other Pacific Islander

  • Native Hawaiian
  • Guamanian or Chamorro
  • Samoan
  • Other Pacific Islander (borrower writes in another answer)


Using this methodology, NCRC has identified a growing trend in the HMDA data. More loan records each year are reported by lenders lacking race or ethnicity data on the loan applicant. From 2018 to 2020, loans without this data rose from 10.8% to 13.6% of all loan originations. For applications on forward mortgages on owner-occupied, site-built, 1-4 units loans in 2018, there were 12.1% that did not include race or ethnicity. By 2020 that figure rose to 14.9%.

The reporting of this data is voluntary for the applicant. However, in most cases, the lender is required to assign a race or ethnicity based on visual observation or surname. The exception to this is if the loan application was started online. The public data released under HMDA does not provide detail on whether the application began in person, by phone or online. However, as mortgage companies, many of which lack a physical presence and rely heavily on fully online applications, have risen to prominence in the mortgage market, they have struggled to collect this data. While this suggests that applicants are choosing not to fill out this data, we also can observe that applicants have eagerly embraced the opportunity to identify themselves more precisely using the disaggregated data for Hispanic, Asian and Pacific Islander populations. In 2020, 61% chose to do so. This suggests that there is not an unwillingness of applicants to provide this data, but rather a problem in how this data is asked of them.

In addition to originating their mortgages, lenders can buy mortgages from other lenders. Depending on the business strategy and goals, this might be the primary business model for some lenders. From 2018 to 2020, these ‘purchased’ loans came to 12% of all records in HMDA. A small mortgage lender might not meet the requirements in size or volume to sell their loans directly to Fannie Mae so instead they sell them to a major bank. That bank, in turn, reports those loans under HMDA as ‘purchased’ loans.

HMDA allows lenders that purchase loans the option of removing demographic data from those records before submission to HMDA each year. Some of the country’s largest lenders appeared to exercise this option as they reported such data on none of their purchased loans, and in 2020 the percentage of purchased loans lacking this data jumped to 75%.

The increasing trend in lenders that do not report race and ethnicity data on loans that they purchase from other sources is particularly concerning. Among the 20 lenders that reported the most purchases in 2018 to 2020, 15 reported no data on 100% of those loans. This suggests they are exercising the option allowing them to strip data from loan records they purchase before reporting that data to HMDA. In other cases, lenders reported more loans without race and ethnicity data than they did in 2018. For example, Wells Fargo purchased 227,689 loans in 2020, and 44% of those lacked demographic data. This was a substantial increase from 2019, where just 24% of the purchases made by Wells Fargo lacked this data.

This increasing gap in HMDA is a threat to its use by both regulators and the public to monitor the mortgage market for signs of predatory activity, discrimination or unsafe lending. Two distinct problems need to be addressed. First, the CFPB should close the loophole that allows lenders to strip demographic data from loans that they purchase. Instead, lenders should be conscious of the fair lending implication of loans that they buy from other lenders. Originations that lack demographic data is a more complex issue, since the lender cannot force applicants to submit this data and if the loan began online the lender is under no obligation to fill in the information themselves. But the existence of lenders that do very well at collecting this data and the high degree of eligible borrowers that provide disaggregated racial data suggests that this may be an issue of how the lender is asking for the information. A working group of lenders and community-based organizations to share best practices might have a positive impact on data collection rates.

Disaggregated Race and Ethnicity Data

The addition of disaggregated data on Hispanic, Asian, Hawaiian and Pacific Islanders has opened up new dimensions for understanding discrimination and disparate impact on communities often overlooked in fair lending analysis.

What is clear from HMDA data is that applicants are not averse to reporting detailed race and ethnic data about themselves when they are asked for it in ways that encourage a response. In 2018, HMDA collected disaggregated data on Asian, Hawaiian and Pacific Islander, and Hispanic loan applicants for the first time. During the loan process, applicants are offered the opportunity to identify as one of the main categories or several more precise descriptions. Each applicant and co-applicant can select five different ethnicities and five races. For eligible applicants, this has been increasingly popular. In its inaugural year, 51% of applicants that identified as Asian, Hawaiian or Pacific Islander, or Hispanic also chose a disaggregated identifier. In 2020 that rose to over 61%.2 Future iterations of HMDA could seek to disaggregate the other racial groups to allow for more specificity in the data as lenders seem to have easily accommodated this data collection and borrowers have enthusiastically embraced it.

Mortgage Applications and Originations

Home Purchase Lending by Race 2007-2020

Due to changes in the reporting requirements for refinance and home improvement loans that took effect in 2018, it is difficult to accurately compare those loans to “legacy HMDA” data from 2007 to 2017. However, we can examine home purchase loans from both periods.

The most striking trend in home purchase lending between 2007 and 2020 is how little has changed in the distribution of loans by race and ethnicity.

Since 2007, there has been only slight variation in the share of home purchase loans going to BIPOC applicants. In 2007, 24.2% of home purchase originations were to BIPOC borrowers, and in 2020 that figure was just 26.6%, a 2.4% increase in 13 years. This closely mirrors the growth in BIPOC households, which have grown from 21.8% of all US households in 2010 to 24.2% in 2019. Most of this increase can be attributed to a growing share of loans made to Asian and Hispanic borrowers. At over $98,000 per year, the Asian median household income in 2019 far exceeded that of all other racial and ethnic groups.

Black homeownership rates peaked at 48.2% in 2005, and their home purchase market share was 7.5% by 2007. Following the housing crash and the start of the Great Recession, Black homeownership rates declined sharply, and the share of home purchase loans began to decrease as well. In the Great Recession, the wealth of US families overall was reduced by 28.5%, but for Black Americans, the decline was 47.6%. This disparate decline in Black wealth in the wake of the Great Recession was due to several things:

  1. Black households have nearly 50% of their assets concentrated in their principal residence,
  2. Higher ratios of debt to net worth among Black households (0.55) compared to Whites (0.15),
  3. A higher ratio of mortgage debt to home value among Black households (0.49) than White households (0.32).

All these factors, coupled with the overall economic insecurity created during the Great Recession, led to Black wealth being only 6% of White wealth from 2010 to 2019. Hispanic households faced similar issues and trends in the wake of the Great Recession. Since 2013, the share of home purchase loans to Black borrowers has slowly crept upward and, at 7.4%, is almost back to 2007 levels.

Black homeownership rates, however, have not recovered since the Great Recession, having experienced the greatest declines in homeownership rates and reaching a point nearly equivalent to when discrimination was legal. This declining trend requires a concerted effort and intentional response to build Black homeownership to a historic high, to make gains at closing the country’s persistent racial wealth divide. Reaching a 60% Black homeownership rate is a substantive and necessary step to building wealth for the Black population, as would it be for Latinos and Native Americans. Further research is needed to understand how many more home loans would be needed to achieve these goals.

However, as the chart above shows, regardless of prevailing homeownership rates, the gap between Black and White homeownership has rarely wavered. Wealth is a relative measurement; one person’s fortune is another’s pocket change. But every month almost 75% of White families build wealth via homeownership while less than half of Black families do the same. Despite increases in Black homeownership (particularly following World War II), Black families have not substantially bridged the homeownership divide with White Americans. This homeownership disparity drives a massive wealth divide between White and Black families. Black families have a median net worth of $9,000, just 6% of the median net worth of White families at $160,000.

Since the inception of the mid-20th century Civil Rights era and the implementation of racially blind laws such as the Community Reinvestment Act, we can point to little progress in bridging wealth and homeownership disparities. At best, the work of consumer advocates, regulators and financial institutions to follow the law has maintained the status quo. And now it appears that an implicit goal of those laws, the increase in BIPOC wealth through homeownership, is further away than it was in 1968. Confronted with this failure, we should acknowledge that forty years of racially blind policies and regulatory efforts have failed to produce meaningful progress and as of 2020, the homeownership gap between Black and White families stands near its highest point in 120 years.

Originations by borrower and neighborhood characteristics 2018-2020

In 2020, just over 13 million originations were reported for forward loans on owner-occupied, 1-4 unit, site built homes.

Lending by borrower race and income

Several patterns in the distribution of loans to BIPOC communities are observed in the data. First, the percentage of all loans going to White non-Hispanic applicants declined steadily since 2018, falling from 67.3% of the loans originated that year to 64.9% in 2020, though still higher than their demographic representation of 60%. In 2018, loans lacking demographic data were 10.8% of the market, and by 2020 had increased by 2.7%. Loans to BIPOC applicants have not increased as a market share overall, though, actually falling slightly from 21.6% to 21.4%. Further research should determine how many of the loans lacking demographic data are likely to be BIPOC applicants.

While lending to BIPOC borrowers has remained relatively stable from 2018 to 2020, this belies a shift in intra-BIPOC lending market share. Asian applicants have increased their share of overall lending, from 5.4% of loans in 2018 to 6.5% in 2020, getting closer to their demographic representation of 7.2%. Most of these gains have come from refinance lending. Asian borrowers were 4% of refinances in 2018 and 7% in 2020, placing them just behind Hispanics at 7.8% of the market for refinances that year. However, when we look at the entire lending landscape, including home purchase, refinance and home equity lending, the figures are troubling. From 2018 through 2020, Black and Hispanic borrowers have seen their overall market positions erode from 6% to 5.2% and 9.8% to 9.2% of the market, respectively.

When looking specifically at home purchase loans, Black and Hispanic borrowers did see a slight increase in their market presence (Black 0.5% increase, Hispanic 1.3% increase), while Asian and non-Hispanic White homebuyers lost market share from 2018 to 2020. The Hispanic population in the US is expected to grow from 18.5% of the population to 24.5% by 2040, including a gain of over 20.6 million people over the age of 25,3 when we would expect them to become homeowners. Much has been said about the growing Hispanic homebuying population, and by 2040, 4.8 million of the 6.9 million new homeowners are projected to be Hispanic. Critically, we have yet to observe any massive shift in lending to Hispanic applicants, and Hispanic borrowers continue to see lower origination rates and higher loan costs than White non-Hispanic applicants. Despite being 16% of the US population over the age of 25, Hispanics received just 9% of loans in 2018-2020. Continued barriers to accessing credit and becoming a homeowner mean that despite increasing in number, Hispanic homeowners have not increased their homeownership rate to that of White families, and this will continue to drive a substantial wealth gap.

The experience of Asian borrowers has been much different. Overall, Asian borrowers increased their market share from 5.4% to 6.5% of all loans from 2018 through 2020. In the refinance space, this was even more pronounced. In 2018, Asians accounted for 21% of all refinance loans made to a BIPOC applicant. In 2020, that figure grew to 35%.

Among the three groups that allow disaggregation, the emergence of Asian Indian borrowers as a powerful force in the mortgage market is clear. In 2018, 29% of Asian applicants were Asian Indian, in 2020 that rose to 36% though only about 24% of the Asian American population. In that time period, almost 654,000 loan applications were recorded with Asian Indian applicants. Not inconsequentially, Asian Indian households in the US earned a median income of $119,000, about $20,000 more than the Asian median income for that year.

While the distribution of home purchase loans among Asian subgroups remained the same from 2018 to 2020, the rise of Asian Indians as a share of Asian applicants was responsible for Asian gains in refinance lending in 2020. In 2018, just 24% of Asian refinances had an Asian Indian applicant, second to Filipino applicants at 25%. In 2020, Asian Indian applicants accounted for 38% of all Asian refinance loans, while Filipinos fell to just 14%. When removing the Asian Indian and Chinese borrowers from the data, the remaining BIPOC borrowers, including several Asian groups, barely saw any growth in the home purchase market share. They lost market share in refinance and home equity lending. This disparity within the BIPOC communities highlights the need for more disaggregated data in HMDA to better identify trends and differences in lending to different communities. In future evolutions of HMDA, the expansion of this data should be considered. For example, among people of African descent there is evidence of disparate treatment based on place of birth and culture that is not captured by the “Black or African American” label.

The share of loans going to LMI borrowers can vary greatly between White and BIPOC households. Since 2018 as the number of loans has skyrocketed, the share going to LMI households has declined, mainly from 2019 to 2020. In 2018, 27.5% of refinance loans went to an LMI homeowner, and in 2020 that figure fell to 19%, a decline of nearly one third. The overall share of refinance loans going to BIPOC borrowers remained relatively stable during this period, primarily due to the rise of Asian Indian borrowers. For all other BIPOC communities, there was a drop in market share over the same period.

What does this mean? As homeowners refinance, they normally accelerate the pace at which they accrue wealth. Over time, homeowners who entered 2020 in a position to refinance, with enough equity in the home, sufficient credit, stable income, etc., will build far more wealth than homeowners who were not in that position. The trends of a stark increase of Asian Indians refinancing, the highest income earners among Asians, coupled with the decline of refinance loans going to LMI and BIPOC homeowners, is likely to help maintain the racial inequality we see in homeownership and wealth. Those who are already high earners, such as Asian Indians or White borrowers, take up a larger portion of the refinancing market, allowing them to save more money, or extract equity, while their lower-income counterparts are stuck with higher costs, thus stratifying the groups further. Further research is required to better understand if there will be a disproportionate impact on LMI and BIPOC homeowners over time. We also need to better understand how this decreased share of BIPOC homeowners getting refinanced might impact the long-term trends of the racial homeownership and wealth gaps.

Lending in BIPOC and LMI tracts

Lending to borrowers on owner-occupied single-family homes in LMI or majority-minority census tracts as a share of all lending declined moderately from 2018 through 2020. Majority-minority census tracts account for 31% of all neighborhoods in the US, and 30% of Americans live in one.4 Despite this, originations to majority-minority neighborhoods fell from 19.1% of all loans in 2018 to 18.5% in 2020. Home purchase lending during this period saw little change, though, with most of the losses concentrated in refinance and home improvement lending. LMI neighborhoods include 30% of all neighborhoods, with 27% of Americans living in one. Yet, loans in LMI neighborhoods declined even more, from 16.5% in 2018 to 13.7% in 2020. This shift is almost entirely due to the rapid decline in the share of refinance loans going to LMI neighborhoods, dropping from 16% in 2018 to just 12.2% in 2020. While the overall number of loans to LMI neighborhoods increased along with all lending, the share of those loans made on homes in LMI neighborhoods declined considerably. Further research into the lending across LMI and majority-minority neighborhoods is needed to assess how much of this lending can be attributed to middle- and upper-income or White non-Hispanic borrowers and how those patterns vary across different regions and cities. Gentrification and displacement are critical threats to LMI and BIPOC communities in some of the largest metros in the country. In conjunction with other tools, HMDA can be used to measure the influx of upper-income home buyers in these communities as a sign of possible displacement.

Older adults and mortgage lending

In NCRC’s 2020 report on “Mortgages And Older Adults After COVID-19,” which analyzed 2018 HMDA data and its implications in a pandemic year, it was found that older adults (55 years or older) see many of the same systemic and structural barriers in the housing market as younger borrowers, with Black and Hispanic applicants reporting lower origination rates and higher denial rates than White and Asian borrowers. In addition, older adults are often the target of predatory lending and were frequently victimized by mortgage lenders during the 2005-2009 home lending frenzy. The report revealed that 30% of all mortgages went to an older adult and they were overwhelmingly refinance loans.

In 2020, it was clear that the influx of younger refinance borrowers has eroded the dominance of older adults in this space. In 2018, over 45% of refinance loans were to older adults, and in 2020, that figure fell to 31.5%. However, as interest rates rebound, it is very likely that older adults will increase their share of the mortgage market. This hypothesis is based on the growing number of older adults as well as the lack of low rates pushing younger homeowners to refinance.

Over the next two decades, the growing population in the oldest age groups will lift the share of all US households age 65 and over from 26 percent in 2018 to 34 percent in 2038. The Joint Center for Housing Studies projects that the number of households aged 75–79 will increase 49 percent in 2018–2028, to 8.9 million, and by another 20 percent in 2028–2038, to 10.7 million.
— Harvard Joint Center for Housing Studies Housing Americas Older Adults 2019

The older adult mortgage market differs in many ways from the rest of HMDA. Older adults are more likely to have equity from previous homeownership or other investments. Older adults hold less of their wealth in their homes and they are more likely to have pensions or other investment accounts. Older adults also need to be able to access the equity in their home to repair or modify it as they age. In many cases, special accommodations are needed for older adults to allow them to age in place.

Reverse mortgages

Reverse mortgages are a small but important market within HMDA. Most of these loans are Home Equity Conversion Mortgage (HECM) loans guaranteed by HUD. In both 2018 and 2019, the first years in which reverse mortgages were added to the dataset, there were just 61,000 applications reported, and in 2020 this number spiked to 70,114. The origination rate on those applications increased as well, from 54% in 2018 to 61% in 2020. The upward trend in reverse mortgages mirrored the demographic changes in the number of American households entering retirement age with mortgages.

Racially, the reverse mortgage market looked very similar to the forward loan market. In 2018 and 2019, about 17% of reverse mortgages went to BIPOC borrowers. This figure dropped to 15% in 2020, suggesting that the increase in reverse mortgage lending, like refinance lending, may have been due to White homeowners taking advantage of the low rates and rapid increase in home values.

Additional research on reverse mortgage lending could focus on this pattern of pandemic lending. Older adults have faced high levels of economic insecurity during the pandemic; did this hardship result in more reverse loans? Did rising home values offer an opportunity for homeowners who did not have enough equity for a reverse mortgage? How many of these mortgages were driven either directly or indirectly by COVID-19?

Gender and Same-Sex Status

At this time, HMDA does not allow borrowers to identify themselves by sexual orientation. However, the co-applicant data can be used to identify loans made to couples of the same gender.

In 2020, NCRC released a report on same-sex mortgage lending using 2018 HMDA data. That report found that same-sex couples were more likely to be BIPOC and have lower incomes than different-sex couples. They were also more likely to have their loans denied and generally paid more in closing costs and interest rates when they did get a loan than different-sex couples.

Very little is understood about the way age, gender, sexual orientation and race intersect with the mortgage market. Each of these variables, along with the location of the property, brings a complex history of privilege and exclusion that is deserving of additional research. As HMDA expands to allow a more precise way to collect data on borrowers, we can begin to better understand the impact of structural barriers to home lending.

Since 2018, there has been little change in the share of same-sex lending in HMDA across all loan purposes, and less than 2% of loans in the HMDA dataset in 2020 had a same-sex pair of borrowers. As of 2017, 4.5% of Americans identified as LGBTQ+. HMDA data should be improved with the addition of sexual orientation/sexual identity data to allow a more precise analysis of how LGBTQ+ borrowers encounter the mortgage market. Any future research would be greatly improved by the inclusion of optional reporting of sexual orientation as a part of the demographic data collected by the lender. Research topics could include what the potential for discrimination is for LGBTQ+ loan applicants? What is the mechanism by which this discrimination plays out? How can regulators and advocates detect evidence of this kind of discrimination?

Biracial Lending

The Census Bureau has noted that a growing number of Americans identify as multi- or bi-racial. As this identity becomes more common, it is important to use the additional fields of demographic data in HMDA to better understand how this population experiences the mortgage market.

While still a small share of all of the records in HMDA, over 380,000 applicants in 2020 identified as biracial on their loan applications. As NCRC explored in our work with UnidosUS on Afro Latino lending patterns, these communities are likely to cluster in large, highly diverse metro areas. However, the percentage of Americans that identify as biracial or as being of ‘some other race’ has dramatically expanded compared to previous census figures. Future research could focus on the ways in which biracial and multiracial people encounter the mortgage market and do they tend to be similar to different races in their experience?

There were just over 213,000 originations in the 2020 HMDA data on owner-occupied, site-built, 1-4 units homes where the borrower identified as having either Hispanic and some race other than White, or any combination of races. When we looked closely at the primary race/ethnic selection made, we found that 46% of these applicants identified as Hispanic, 28% Asian, 12% Native American, 8% Black and 4% Hawaiian or Pacific Islander.

Origination rates by race

The most common outcome of a loan application is an ‘origination,’ where the loan closes. Loans can also be denied by the lender, creating a ‘denial.’ Several other outcomes are not lender denials but do not result in an origination; for example, if the borrower chooses another lender and abandons the application. These other outcomes are known broadly as ‘fallout.’ To ascertain whether specific groups are seeing a disparate impact from the loan process, it is important to understand this distinction and look at the origination and denial rates for BIPOC or LMI borrowers.

Overall, origination rates in 2020 were higher both for BIPOC and White loan applicants. For White borrowers, overall origination rates rose from 64.3% in 2018 to 68.9% in 2020, while BIPOC applicants went from 53.7% to 59.4%.

This increase was almost entirely due to refinance lending, where origination rates increased sharply from just 51% in 2018 to 64% in 2020. This was enough to offset a small decline in the home purchase origination rate. Overall, BIPOC refinance origination rates jumped from 43.1% in 2018 to 57.8% in 2020. As noted above,  the increase in BIPOC refinancing was driven by Asian homeowners. NCRC’s 2020 HMDA Preliminary Analysis explored the decrease in Black and Hispanic market share in refinance lending between 2018 and 2020. Further research on how extreme shifts in interest rates impact homeownership opportunities for BIPOC and LMI families, Black and Hispanic households particularly, could help us understand how broader fiscal policies can reduce or enlarge the racial wealth gap.

S&P Dow Jones Indices LLC, S&P/Case-Shiller U.S. National Home Price Index [CSUSHPINSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/CSUSHPINSA, October 24, 2021.

Freddie Mac, 30-Year Fixed Rate Mortgage Average in the United States [MORTGAGE30US], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/MORTGAGE30US, October 24, 2021.

These origination rates are just a part of the story. There were clear variations in home loan closing costs and interest rates between different racial groups. The disaggregated racial data on Hispanic, Hawaiian and Pacific Islander, and Asian borrowers revealed stark differences in origination rates, closing costs and interest rates.

While the difference in interest rates requires attention, the overall low rate environment prevented them from having a substantial negative impact on borrowers. For example, in the chart above, the lowest interest rate is 2.92% (Asian Indian) and the highest is 3.26% (Black). On a $100,000 loan over 30 years, the higher interest rate will translate to an additional cost of $6,600 in interest paid, a difference of just $18 each month for the homeowner. However, as rates rebound this may change. As rates move further away from zero, there exists more opportunity for lender arbitrage, where wealthier borrowers with better credit can qualify for steeply discounted rates that will not be an option for LMI or BIPOC applicants.

Of more immediate concern were the steep differences in closing costs for loans that were, ostensibly, identical. Closing fees include a number of charges for a variety of different items, but primarily they are an origination fee, mortgage insurance charges, and other services related to underwriting of the loan. For home purchase loans, there is an enormous difference in the costs incurred by different applicants. Loan size, credit score, physical location and borrower experience can all play a role in determining the cost to close on a mortgage. But the impact was clearly different across different racial groups, with Asian Indian home buyers having paid $2,896 in closing costs compared with $4,026 paid by non-Hispanic Whites. However, as the chart above shows, Black, Hispanic, Filipino and Hawaiian or Pacific Islander buyers often paid far more. CFPB analysis of non-public HMDA data found that the median credit scores differ greatly between these groups as well, shaping the data in ways that are more complex than they first appear.

More research would be needed to account for differences in local housing markets, loan amounts and property values. Regardless of the reason for these differences, closing fees and down payments remained a major barrier to homeownership for most borrowersAssistance for first-generation homebuyers to overcome these barriers would likely have a substantial positive impact on Black and Hispanic homeownership.

Differences between racial groups raised a number of questions that future research could explore. Asian Indians emerged as a growing segment of mortgage consumers, but their origination rate remained low compared with other borrowers. Was this due to denials, or were Asian Indian borrowers more likely to shop for a mortgage, starting applications with several lenders before settling on one of them?

In particular, the variations in loan interest rates and closing fees need to be better understood. For instance, while Filipinos make up the third highest Asian subgroup of borrowers, they experienced higher average closing costs ($4,655) than Asians overall ($3,899) and Asian Indians ($3,294), as well as higher than the average closing costs of Black borrowers ($3,818). These closing costs were similar to Hispanic home purchase borrowers’ closing costs ($4,294). Hawaiian or Pacific Islanders experienced the highest average interest rate at 4.1%; within this group, Native Hawaiians experienced the highest average closing costs of all demographics at $4,772. These differences could be the result of structural, regional, political or cultural barriers that need to be better understood so regulators, advocates and the lending community can address them.

In their report on AAPI lending, the CFPB noted substantial differences between AAPI groups, including income, age and credit score variations that hint at why some AAPI groups appear to see different outcomes in the mortgage market.

Denial rates and reasons

Income is often the core reason for a mortgage being denied. The lack of money for closing costs or sufficient income to support the monthly payment accounted for about 41.6% of denials to LMI borrowers in 2018. In 2020, that figure grew to 45.5%.

The rise in origination rates for refinance loans accompanied a commensurate drop in denial rates for those loans. Denial rates across most BIPOC communities were almost halved between 2018 and 2020 for refinance and cash-out refinance lending (fallout loan rates remained essentially the same).

Part of the reason for the fall-off in denials can be inferred from the changes in the frequency of denial reason codes reported by lenders. The debt-to-income ratio remained the most common reason for denial, and this code indicates that the borrower does not earn enough to afford the loan they are attempting to obtain. The share of applications denied for debt-to-income declined modestly since 2018, from 29.4% to 27.9%. A much larger decrease was seen in the percentage of loans denied for poor credit, from 29.7% to 24.8%. One reason for the decline of those two leading denial reasons was that borrowers that did not finish their loan application jumped from 9.5% in 2018 to 16.3% of all denials in 2020. Meanwhile, denials for insufficient collateral dropped from 16% to 12.3%. Taken as a whole, along with the lower movement in interest rates and rising home prices, the changes to denial rates suggest several trends that cannot be directly observed in HMDA data.

  • Income in comparison to housing cost continues to be a major stumbling block for many borrowers, and though it has declined as a share of all denials, the debt-to-income ratio remains the most common reason for denial.
  • Credit history has declined substantially as a share of all denials. This does not indicate that credit requirements are easing, though. According to the Urban Institute, credit scores reported by the government sponsored enterprises (GSEs) have been at their highest points since at least 2014, with bank-reported credit scores on mortgage borrowers peaking at over 770 FICO.
  • Case-Shiller reports that home values have risen sharply during the pandemic, reducing the chance a property will not appraise for the requested amount. This is also likely to reduce the number of applications declined due to insufficient collateral value.
  • More hypothetically, the rapid increase of loans denied for incomplete credit applications, jumping 6.8% from 2018 to 2020, could indicate that borrowers are more likely to shop around for the best rate as lenders scrambled to process as many refinance loans as they could while rates remained low.

Loan pricing

Loan interest rates and closing costs

Since 2018, interest rates have fallen broadly across all mortgage lending types. This can best be seen in a historical context by observing the 30-year fixed rate loan interest rate over time.

Freddie Mac, 30-Year Fixed Rate Mortgage Average in the United States [MORTGAGE30US], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/MORTGAGE30US, October 13, 2021.

This means that lenders have been able to offer historically low interest rates for home mortgage lending. As discussed elsewhere in this report, low interest rates have driven an explosion in refinance borrowing in 2020.

As interest rates declined, the loan amounts increased across all loans, growing from $232,349 in 2018 to $286,888 in 2020. Much of this increase was driven by refinancing lending, where the 2018 average loan amount was $225,454. In 2020, that increased by almost $68,000 to $293,428. This was more than twice the increase in home purchase loan amounts.

Loan costs are reported in several ways in HMDA. The origination fee collected by the lender at closing is included with other fees, including the mortgage insurance upfront premium on FHA loans, under either “total loan costs” or “total points and fees.” A separate charge is recorded if the borrower paid any discount points or received a lender credit.

Banks versus Mortgage Companies

In 2020, mortgage company lenders expanded their share of the mortgage market, making 58% of all originations, compared with 44% in 2018. In 2020, mortgage companies made 62% of all home purchase loans and 61% of all refinance loans.

Mortgage companies moved much more aggressively into the refinance market as rates declined. In 2018, bank lenders reported 1.5 million refinance applications, compared with almost 2 million reported by mortgage companies. In 2020, banks took 3.7 million applications for refinances. Mortgage companies greatly surpassed this with just over 7.8 million refinance applications that year.

Mortgage companies, banks and credit unions serve different communities and often offer different products. As mortgage companies assert themselves as the primary source of loans for many BIPOC and LMI borrowers, future research should focus on the extent to which this impacts their applicants and what impact, if any, it has on the racial wealth gap.

Government Lending


Government-backed loan programs include the Federal Housing Administration (FHA) program, which is administered by the Department of Housing and Urban Development (HUD), the Department of Veterans Affairs (VA) program, and the Rural Housing Service (RHS) program managed by the Department of Agriculture. These programs do not make loans, but provide additional insurance to the investor. Ideally, this allows lenders to make mortgage loans to borrowers that might not otherwise be able to qualify for a loan. Government-backed loans fluctuate as a percent of all lending. In 2005, FHA lending dipped to just under 2% of the mortgage market, and by 2009 it expanded to almost 18%, as conventional lending became extremely difficult for borrowers to obtain. In 2020, the FHA share of the market fell for the fourth consecutive year to 8.8%. What has not fluctuated is that government-backed lending remains critical for home purchase credit for most BIPOC borrowers. In 2020, 62% of Black homebuyers and 61% of Black refinance borrowers used FHA. This is another area where we see substantial variation in the Asian subgroups, with 36% of Filipinos using government loans to buy a home, compared with just 5% of Asian Indian and Chinese buyers.

Costs of FHA/VA Versus Conventional Lending

Though a critical pillar of support, government-backed lending has a cost borne mainly by the borrowers themselves. The most noticeable way this occurs is via the mortgage insurance upfront premium that borrowers pay when they close on their FHA loan. In HMDA, this fee is not singled out, but is a part of the costs included in the total loan costs or total points and fees reported by the lender. In the table below, they are included in the “other charges to borrower” figure.

Conventional loans averaged $1,705 in other charges, while FHA’s average was $5,874. While HMDA cannot definitively tell us if this is due to the upfront mortgage insurance premium, it is the most likely answer for why the cost was so much higher for government-backed loans.

Banks and Mortgage companies in the FHA/VA market

In 2020, banks continued to avoid government-backed loan programs, as they have since the end of the Great Recession.

Government-backed loans that wind up in default can be problematic for banks. In cases where there was an error in underwriting, the original lender can be forced to repurchase a defaulted loan from the investor. This requirement led to large losses by banks following the Great Recession. As a result, banks rarely make government loans at all.

“The real question for me is should we be in the FHA business at all,”
Jamie Dimon, CEO of JPMorgan Chase

In 2020, home purchase loans made by banks were conventional 81.3% of the time, compared to just 59.3% of loans made by mortgage companies.

Among the top 20 lenders, this disparity in the use of FHA and VA lending was apparent. None of the nine banks in the top twenty loan originators used government lending for more than 20% of their lending. The FHA program is intended as a countercyclical insurance policy to ensure the flow of mortgage loans does not freeze in a market downturn. In the years prior to the Great Recession, the market share of FHA lending fell to 1.9% of all loans. Following the housing crash that expanded to 17.9%. As of July 2021, it had fallen back to 7.99%. Except that in 2020, it remains the primary source of home purchase lending for Black (58%), Hawaiian or Pacific Islander (47%), Native American (49%) and Hispanic (46%) borrowers. This leads to a permanent disparity in the cost burden for the FHA program, shouldering those borrowers least able to pay with the majority of the cost. Future research should address the exit of banks from the FHA/VA market and whether this has a disparate impact on the ability of BIPOC borrowers to access the market.

The VA market also showed troubling signs of disparate impact. VA lending was concentrated among a small subset of HMDA reporters, meaning that these lenders have an outsized impact on the entire VA market. Closing costs for VA loans closely mimicked FHA lending, yet VA loans require no Mortgage Insurance Premium, so the source of those additional fees was unclear from HMDA. In a report issued by Representative Katie Porter’s office, researchers raised possible issues with excessive loan ‘churning’ in the VA market. This is a process where a lender originates a loan at an artificially high interest rate to an unsuspecting borrower. Then, a short time later they offer to refinance the loan for that borrower to a lower rate, collecting fees at both closings at a cost to the borrower. Lacking a universal loan identifier, public HMDA data offer no way for advocates or the public to detect this kind of predatory behavior.

Loan Details by Individual Race

Looking at key statistics about how each racial and ethnic group fares in the mortgage market reveals key differences in the cost and wealth being built by individual communities.

Hispanic borrowers (Mexican in particular) and Black borrowers were the most likely to buy a home in an LMI neighborhood. In 2020, of the 142,810 Mexican homebuyers, 29% were buying in an LMI neighborhood, just slightly higher than Black (28%) and Hispanics that didn’t choose a subgroup (28%). Disparities of LMI neighborhood concentration can also be seen within Asian and Hawaiian or Pacific Islander groups. Only 8% of Asian Indian home purchases were in LMI neighborhoods, whereas 23% of Vietnamese were. Biracial borrowers bought homes in LMI neighborhoods 21% of the time.

Twenty-three percent of total borrowers were LMI themselves. Trends mirrored the LMI neighborhood rates, as 31% of Black and Hispanic borrowers were LMI, whereas 22% of White and 19% of the total Asian demographic were LMI. Vietnamese borrowers specifically had higher rates of being LMI (31%), as did those identifying as “other Asian” (29%).

Home equity, calculated by subtracting the property value from the loan amount, gives us a snapshot of the wealth a homeowner has in the home at the time of closing. This equity has implications on wealth accumulation and perpetuating the racial wealth divide.

Looking at some of the detailed lending data from HMDA offered a glimpse of how race relates to the ability to build wealth via homeownership. Black homebuyers had the least equity in their home at purchase. The 2020 average loan amount to a Black home purchase borrower was $259,220 and the average property value $285,118. This resulted in just $25,546 in equity at closing. Puerto Rican borrowers reported the lowest loan amount, borrowing an average amount of $235,926 to finance their home purchase. The property they bought was worth an average of $262,683, giving them $26,708 in equity at closing.

At the other end of the spectrum, Chinese home buyers in 2020 borrowed $517,494 on average on homes worth $761,631, giving them $232,525 in equity. Asian Indians weren’t far behind, borrowing $497,313 to buy homes worth $640,323, a $141,781 difference. White homebuyers fall between these extremes, averaging $72,655 at closing, almost three times the equity held by Black homeowners. The long term impact not only of the homeownership gap but also this equity gap would be a topic for future research.

Breaking down the equity owned by borrowers at closing by race and age revealed complex patterns that offer compelling evidence that higher income and higher wealth households of color build more wealth at a quicker rate than lower income BIPOC communities. In some cases, as with Chinese homebuyers, this may reflect international investment or the presence of banks dedicated to servicing the Chinese American community. However, the impact of regional economies and clusters of specific groups should be considered when assessing disparate impact. When a specific racial or ethnic group is concentrated in a high cost market (e.g., Hawaiians and Hawaii) we can expect them to report larger loan and property values. Meanwhile, African Americans in the US are bifurcated between urban markets with high home costs and rural southern states where home values are much lower. How this imparts wealth or creates additional barriers to homeownership for those groups should be better understood.

It is very difficult to accurately estimate the wealth of individual groups due to the various ways in which families store wealth; property, cash, retirement, investment or tangible goods are just some of the ways in which families attempt to store and build wealth. Much of the data on these strategies lacks demographic data, making it difficult to estimate the racial wealth gap at the level of detail offered by HMDA. Other products such as the Survey of Consumer Finances (SCF) and the Survey of Income and Program Participation (SIPP) are too small or infrequent to offer the detail that HMDA provides. Homeownership is the primary way in which most families start to build wealth, and 27% of the racial wealth gap is explained by the number of years of homeownership that a family has. HMDA offers one of the best ways to view the exact amount of home equity borrowers “owned” at the time they closed on their mortgage.

However, the home equity figures likely mask a greater division due to what we do know about how families build assets over time. For older adults aged 55 and up, only about 23% of their wealth is stored in their principal residence. For homeowners under 35, their home equity is 57% of the wealth they have. So, using the chart above as a guide, 25-34 year olds had an average of $50,786 in equity at closing on their home purchase. If this is 57% of their total wealth, their total wealth would be $89,098. For 65-74 year olds that reported $112,410 in equity, we would estimate their total wealth to be $488,739.

These trends reiterate previous research findings that Black and Hispanic home purchasers have higher rates of mortgage debt relative to their property value. As the saying goes, it’s expensive to be poor. Black and Hispanic borrowers are not just buying less valuable homes than White (and some Asian) borrowers, they are paying more to do so. That excess wealth can be reinvested over time in other ways to diversify the equity a family has at its disposal, an option that is not possible for communities at the lower end of the equity ladder.

Detailed Data on Trends of Chinese Borrowers

Making use of the disaggregated racial and ethnic data, there appears to be a very different pattern to loans with a Chinese applicant in comparison to all other races and ethnic groups. During the 2018 to 2020 time period, there were just over 104,000 originations of home purchase loans to borrowers that identified as Chinese. The vast majority of these were conventional loans, 75% of which were located in 19 metros, and 50% of which were made by just 18 different lenders. A substantial difference in the amount of equity at closing, or down payment, for Chinese borrowers was clear. This was true especially at the younger end of the age range. Borrowers under the age of 25 that were not Chinese had an average amount of equity at closing of $26,519 from 2018 to 2020. Chinese borrowers in this age group averaged $157,988.

Borrower income does not appear to explain this gap. Nationwide, the average non-Chinese home purchase conventional loan borrower had an income just 24% lower than the average Chinese borrower. Yet, they reported 67% less equity at closing. This indicates that Chinese borrowers, at every age level but especially among younger borrowers, are coming to the closing table with down payments that appear to be far in excess of what their income would suggest when compared with every other racial and ethnic category.

There is a thriving industry facilitating the investment in US housing markets by Chinese citizens. They are the largest group of foreign investors in US real estate. Many of these loans are also in communities with vibrant Chinese-American communities, making it difficult to determine if this is investor activity at all. However, the wealth gap that this data suggests may result in localized disinvestment in other BIPOC communities.

Future research could look at several aspects of this trend. Is there evidence of investor activity in HMDA that distorts the data or harms LMI or BIPOC households in the communities where they cluster? Is there a relationship between this relatively small share of all mortgages and gentrification activity that is clustered in many of the same cities?

Top 50 Lenders

Banks, mortgage companies and credit unions are very different institutions, with different consumers and goals based on their role in the market. Little is known at this point about how changes to the market, including recessions, interest rate changes and other activity impact how lenders service their consumers. One way of thinking about this is to look at how much of the lending by each type of lender is going to people of color or LMI borrowers, or in LMI or majority-minority neighborhoods.


Home purchase lending from 2018 through 2020 didn’t show much of a change. Banks, mortgage companies and credit unions all made minor reductions in lending to LMI and majority-minority neighborhoods as a share of all of the loans they made. But there were offsetting increases in the share of loans going to LMI and BIPOC applicants. However, as noted elsewhere in this report, when Asian Indian and Chinese borrowers are removed from the data, BIPOC borrowers saw no change in their share of the market from 2018 to 2020.

A review of the top lenders by number of originations shows stark disparities in how different lenders and different kinds of lenders lend to LMI borrowers.

In 2020, LMI borrowers accounted for 23% of all loans on owner-occupied homes, a decrease from 27% in 2018. This decline varies greatly between different loan purposes and lender types. Lending to LMI borrowers declined for all lenders from 2018 to 2020, but mortgage companies in particular saw their lending to this segment fail to keep pace. Similar but less severe declines are seen in lending in LMI and majority-minority census tracts. More research is needed to understand better how low interest rates drive this shift in lender behavior and whether expanding CRA to cover these institutions, as has happened in MassachusettsIllinois and most recently in New York, would have mitigated this loss.

Top 50 Metros

As with race and ethnicity, where a home loan is made has enormous implications on the home equity and wealth that the family owning the home will have.

The dominance of the Bay Area was clear, with San Jose reporting the largest average loan amounts ($948,436) and home values ($1,358,368). This means that at closing the average owner-occupant homebuyer has $408,739 in equity on their home. By contrast, 15 of the top 50 metros offered home buyers less than $50,000 in equity at closing. Future research might examine the role of location and how it intersects with race to create wealth via processes like buying a home.


Working with HMDA data

In order to better understand lending patterns, the 2018-2020 HMDA data was used to conduct a descriptive analysis. Working with HMDA data presents several challenges for researchers. The current iteration of HMDA debuted in 2018 after the CFPB took over data collection. The CFPB added several additional data points related to borrower demographics, loan prices and features, down payment, property value and income. Changes were also made in the collection and reporting requirements. All of this makes it difficult to directly compare legacy HMDA (pre-2018) with new HMDA data. For that reason, this report limits most of its analysis to 2018-2020.

In order to complete our analysis of the 2018–2020 HMDA data, NCRC downloaded the national snapshot from the CFPB website when it was released. We combined this data with data taken from the reporter panel and transmittal sheet to build a table of lender information. We then combined this with data from the FHFA lender file for each year. This allowed us to verify the lender type variable as a bank/affiliate, mortgage company, or credit union.

HMDA data can be filtered in many ways. In most cases, unless otherwise stated, we have confined this analysis to forward loans on owner-occupied, site-built, 1-4 unit properties. For more detailed methods, please contact Jason Richardson.

For the section on biracial lending, we have compared data across all five fields relating to ethnicity and also the five fields relating to race in HMDA. Applicants that identified as any of the following combinations were considered to be biracial. Co-applicants were not included.

  • Hispanic and either non-Hispanic, or any race other than White.
  • Any combination of two races.

Standard definitions used in this report

How to determine borrower race, ethnicity, gender, same-sex status and age in HMDA

This report follows standard NCRC methods of determining race and ethnicity as well as loan pricing. These definitions are highly subjective, however. The race and ethnicity data in HMDA is rich and many respondents choose several different combinations. Our methodology favors identifying those applicants that choose disaggregated race data if they are Hispanic, Asian, or Hawaiian or Pacific Islander. Depending on your research focus, this might not be an appropriate method for your analysis. Please contact NCRC research for help with determining how to calculate race and ethnicity.


In this report, we have at times removed records that had data on interest rates or rate spreads that were over twice the standard deviation from the mean in 2019. The number of records impacted by this was extremely small as a percent of all records.

Print Friendly, PDF & Email
  1. In this report, we will review the trends since the implementation of new HMDA data on disaggregated race, loan costs, and features. Given the importance of HMDA to documenting and addressing the racial wealth gap, we will look extensively at the race, ethnicity, and income of borrowers and how lenders treat them in comparison to other applicants. Unless otherwise noted, this report focuses on forward mortgages made on owner-occupied, 1-4 unit site-built properties. In cases where loans do not identify whether they are forward loans due to the exemptions afforded to smaller lenders, we have counted them as forward loans. All data is from the CFPB/FFIEC from 2018 to 2020. We have excluded some outliers from analyses on loan costs. Please see our methods section for more documentation on this process.
  2. For more on how NCRC identifies applicant race, see our white paper on the subject. https://ncrc.org/ncrcs-hmda-2018-methodology-how-to-calculate-race-and-ethnicity/
  3. Author’s calculations https://www.census.gov/data/datasets/2017/demo/popproj/2017-popproj.html
  4. https://www.ffiec.gov/censusapp.htm
Scroll to Top

NCRC 2020 Home Mortgage Report: Examining Shifts During COVID-19

Complete the form to download
the full report: