Annual Mortgage Lending Study 2025

By Druta Bhatt

Each year, our annual lending study serves as both a mirror and a map: it reflects the current state of lending in our communities while providing direction for action and advocacy. Our policy goals are grounded in a simple principle: transparent, equitable access to credit is a foundation for sustainable homeownership, community stability, and generational wealth building.

A Tool for Equitable Homeownership

At its core, this study examines lending trends that directly impact homeownership opportunities, particularly for historically underserved and marginalized communities. By highlighting gaps in mortgage lending, small business loans and community investment, our research supports policies that expand fair access to credit and dismantle systemic barriers.

Our objectives for the coming year remain clear:

• Increase Black and Minority homeownership rates by promoting inclusive lending practices and strong accountability for lenders.

• Strengthen investment in middle neighborhoods to prevent displacement and expand opportunities for first-time and working-class homebuyers.

• Advance community-driven solutions that leverage data to shape local, state, and federal policies.

The data in this lending study empowers neighborhood groups, community development organizations, and local governments to advocate for equitable lending commitments from financial institutions. We encourage stakeholders to use this study to build productive relationships with banks — not only to address lending gaps, but to co-create strategies that foster long-term financial security for residents.

Partnering with Banks for Stronger Communities

Banks are more than lenders — they are vital partners in community revitalization and mobilization of capital. This report offers a fact base that community groups can use to engage lenders in meaningful dialogue about local needs. Whether it’s improve access to first-time home buyer products, increasing investments in affordable housing, or support small businesses in main street corridors, this data helps communities hold banks accountable while identifying shared goals and building collaborative solutions.

We urge banks to view this study as a resource that highlights opportunities for deeper impact. Strong partnerships between lenders and communities can expand access to credit, grow homeownership, and strengthen local economies.

CRA Rollbacks: A Critical Crossroads

In July 2025, the Office of the Comptroller of the Currency (OCC), Federal Reserve, and The Federal Deposit Insurance Corporation (FDIC) issued a Notice of Proposed Rulemaking to rollback and rescind key parts of the 2023 Community Reinvestment Act (CRA) modernization. The 2023 rule had strengthened the CRA for the time in decades by updating how banks are evaluated in an era of digital banking, expanding assessment areas beyond physical branches, and increasing data transparency and community input.

The proposed rollback would reverse much of that progress:

• Narrowing assessment areas, which would let online, and national banks avoid responsibility for serving communities where they do business have no branches.

• Loosening performance standards, making it easier for banks to pass CRA exams with minimal reinvestment

• Reducing local community input, limiting how residents can weigh in on banks’ records.

• Halting improvements in lending data, making it harder to identify and address persistent gaps.

At a time when racial homeownership gaps remain wide and branch closures continue to hit underserved neighborhoods the hardest, these rollbacks threaten to weaken a critical tool for holding banks accountable. Our study helps communities push back—by providing clear evidence of where reinvestment is falling short and where local demands must be heard, even as federal oversight shifts.

Section 1071: New Data and New Threats

Another important piece of lending progress is also at risk: Section 1071 of the Dodd-Frank Act, which requires lenders to collect and report demographic data on small business credit applications. This new reporting would expose whether minority, women, and immigrant owned businesses are being fairly served — shining a light on gaps that have been hidden for decades.

Yet Section 1071 is under threat on two fronts:

• Legislatively, some members of Congress are pushing to delay, weaken, or overturn the rule.

• Regulatory, ongoing efforts to defund or dismantle the Consumer Financial Protection Bureau (CFPB) — which enforces Section 1071 — threaten the agency’s ability to implement and oversee this critical data collection.

Looking Ahead: Turning Data Into Action

As we release this lending study, our policy goals remain clear:

• Expand access to homeownership, especially for historically excluded communities.

• Strengthen investment in middle neighborhoods to prevent displacement and build neighborhood wealth.

• Protect fair lending gains at the federal level while pursuing local and state solutions that fill in the gaps.

• Equip communities with actionable data to negotiate and advocate for equitable lending.

Data alone is not change — but it is power. This study gives communities the facts they need to hold banks accountable, shape policy and build partnerships that turn lending into lasting opportunity.

Together — residents, lenders, policy makers and advocates — we can ensure that access to credit and capital truly works for all communities, not just some.

Now in its 29th year, the Pittsburgh Community Reinvestment Group’s (PCRG) Annual Mortgage Lending Study remains a resource for understanding how home lending shapes communities across Southwestern Pennsylvania. This year’s report analyzes mortgage lending data from 2018 to 2023, offering a multi-year view of lending patterns and disparities across Allegheny, Armstrong, Beaver, Washington, and Westmoreland counties.

Drawing on data from the Home Mortgage Disclosure Act (HMDA), the study tracks lending trends by race, income, and geography, with a focus on how equitable — or inequitable — access to credit has been across the region. In addition to identifying regional trends, the report provides lender profiles that highlight how specific banks are serving — or underserving—local borrowers. It also offers neighborhood-level insights, connecting data to the lived experiences of residents in Pittsburgh’s diverse communities.

At its core, PCRG’s lending study aims to inform policy, guide advocacy, and promote greater transparency and accountability in the financial system. By documenting where lending is concentrated and where it is lacking, this study empowers community members, elected officials, and financial institutions to advance fair housing and reinvestment in all neighborhoods.

 Key Takeaways

1. Mortgage originations across all five counties dropped to their lowest levels since 2018, driven largely by high interest rates reducing refinancing activity. Home purchase originations and home improvements loans also dropped.

2. Banks hold the largest overall market share in the mortgage market, but mortgage companies dominate home purchase loans category, originating over 60% in 2023. The popularity of mortgage companies, which do not have equitable lending obligations under CRA, and which charge higher loan costs, is particularly high among minority races like Black, Asian and Native American groups.

3. Native American and African American borrowers continue to face significantly higher denial rates—nearly double that of white applicants in the five county region of Allegheny, Armstrong, Beaver, Washington and Westmoreland counties. In 2023, White applicants faced a denial rate of 18% and African American and Native American applicants faced interest rates of 29% and 36% respectively.

4. Home purchase loans have steadily declined since 2021, with upper-income borrowers receiving the majority of loans across Allegheny County and Washington County. While Armstrong, Beaver and Westmoreland Counties have majority moderate income borrowers originating home purchase loans owing to lower home values in these counties.

5. The share of home purchase loans to black borrowers has increased from 2022 to 2023 but remain at only 3.2% of all home purchase loans in Allegheny County with a population of 12.5% Blacks. The share of home purchase loans to low-moderate income (LMI) borrowers in Allegheny County have also increased slightly from 15% in 2022 to 17.33% in 2023.

6. Interest rates continue to remain high post 2021. Average total loan costs for home purchase loans also increased from 2018 to 2023, irrespective of interest rates. Mortgage companies have the highest average loan cost, 19% higher than banks, and credit unions have the lowest average loan costs. The average loan amount for home purchase also increases each year from 2018 to 2022 but slightly decreased for the first time from 2022 to 2023.

7. The share of same-sex home purchase originations increased slightly, from 2.33% of all mortgage originations in 2018 to 2.96% of all mortgage originations in 2023.

In each of the counties studied, the number of mortgage loans in 2023 reduced further from 2022 and dropped to the lowest numbers since 2018. The high interest rates affected the decrease in refinancing loans the most, followed by home purchase loans.

Allegheny County had 23,328 mortgage originations in 2023, a drop from 32,794 in 2022. Westmoreland County had 6,740 originations in 2023, Washington County had 4,308, Beaver County had 3,346, and Armstrong County had 991. Home purchase loans were highest in number in 2023, followed by refinancing and home improvement loans. The percentage of refinancing loans for each of the counties was around 19%-23% in 2023 and the share of home purchase loans was 45%- 50%. Following the pandemic, owing to low interest rates, the share of refinance loans was almost 50% in 2021.

The composition of loan actions, or the share of loan applications that were denied and originated has not changed much from 2018 to 2023. In 2023, Allegheny County had 41,424 mortgage applications, Armstrong County had 1,712, Beaver County had 5,931, Washington County had 7,684 and Westmoreland County had 11,770. The denial rate, or the percentage of loan applications that were denied, for each county was between 17% to 22% in 2023.

The different lending institutions are divided into three categories – bank or affiliate, mortgage company and credit union. While banks and credit unions offer similar services, credit unions, unlike banks, are not for-profit and membership to certain groups is a pre-requisite for accessing the services of a credit union. Mortgage companies on the other hand, don’t provide the full range of financial services that banks provide but focus only on mortgage lending. Of these three types of lenders, only banks are covered under the federal Community Reinvestment Act (CRA) which obligates them to lend equitably to all demographic groups in their service areas.

In 2023, 53.3% of the all-mortgage originations in the five-county region of Allegheny, Armstrong, Beaver, Washington and Westmoreland counties were by banks or bank affiliated financial institutions. Second to banks, mortgage companies originated 36.54% of mortgage loans that year and credit unions originated the rest. However, when we observe the market shared by loan purpose, it is a different story. Mortgage companies dominate when it comes to home purchase loans in this region, with 60.26% home purchase loans originated by mortgage companies in 2023. The gap between the share of home purchase loans by banks and mortgage companies has been widening since 2019. This gap was highest in 2021 and has closed only a little since then. The share of home purchase loans by mortgage companies increases further for Hispanic, Asian and Native American borrowers. Home improvement loans are highly dominated by banks throughout 2018 to 2023 with banks taking up 79.16% of market share in 2023. Banks and affiliate lenders also maintain a majority share in refinancing loans for white and native American borrowers throughout the period studied. For Asian, African American, and Hispanic borrowers, mortgage companies have a higher market share for all loan purposes combined.

Denial rate is the percentage of loan applications that are denied. The graphs show denial rates of different races and ethnicity and denial rates of borrowers with different income categories. Denial rates for all groups have been increasing from 2021.

 In 2023, Native Americans faced the highest denial rate of 36.05%, followed by African Americans at 29.33%. Asians have the least denial rate at 16.05%, followed by whites at 18.28%. From 2018 to 2023, each year Asians have had the least denial rate, followed by whites, and Hispanics. Native Americans and blacks have faced the highest denial rates, double that of whites. The denial rate is also vastly different depending on applicant’s income. Low-income applicants have their loans most often denied as they tend to have riskier profiles that might demonstrate poor ability to pay back mortgages.

The number of home purchase loans continues to decrease after 2021, in 2022 and 2023 in each of the five counties - Allegheny, Armstrong, Beaver, Washington and Westmoreland Counties. In Allegheny County the share of home purchase loans across applicants of different income groups remains similar, where upper income borrowers having the maximum number of loans and low-income borrowers having the minimum number of loans. Beaver County, Washington County and Westmoreland Counties have a similar case. Low-Moderate income borrowers together make 38.8% of total yearly originations in 2022 and 2023 in Allegheny County. In Armstrong County however, moderate income borrowers have highest number of loans, followed by middle income, low income and upper income.

In Allegheny County, for home purchase loans, white borrowers, the majority demographic, are maximum. In 2023, white borrowers were followed by African American borrowers. Over the years Asians and Black borrowers have alternated being the second highest borrowers after whites in Allegheny County. Year-on-year we the share of Black borrowers change mainly because the total mortgage loans fluctuate but black borrowers don’t as much.

Home purchase interest rates continue to rise post 2021. In 2023, the average interest rate charged for home purchase loans was 6.62%. For different racial and income groups, interest rate was between 6.5% to 6.8% in 2023. Average total cost of mortgage loans has been rising, irrespective of interest rates. In 2018, the average total loan cost was $4,733 in the five-county region of Allegheny, Armstrong, Beaver, Westmoreland and Washington Counties. This rose to $6,730 in 2023, a 42% increase since 2018. Loan costs have both fixed costs and costs that increases with increase in loan amount. This is also why loan costs are slightly lower for lower income applicants. The average loan costs for mortgage companies are higher than those of banks and credit unions for all years from 2018 to 2023. Credit unions, which do not operate for profit, have the lowest average loan costs. In 2023, average total loan cost for home purchase loans originated by mortgage companies was $7,177 , 19% more than the loan cost associated with originations by bank or bank affiliate financial institutions.

The average loan amount for home purchase loans in the five-county region is increasing year-on-year since 2018, except in 2023. In 2023, the average home purchase loan amount was $248,233, slightly down or flat from $249,442 in 2022. For low-income individuals, average loan amount in 2023 was $117,494 and $169,770 for moderate income home buyers. For African Americans, the average loan amount in 2023 was $192,318 and for white borrowers was $243,805. Asian borrowers had the highest average loan amount at $349,265 in 2023, and Black borrowers had the lowest. Average loan amounts are lower for FHA and VA insured loans, compared to conventional loans.

This section of home purchase loans explores home purchase by applicant and co-applicant sex. In 2023, there were 482 home purchase loans taken out by borrowers where the applicant and co-applicant had the same sex. The composition of loans based on applicants’ sex is not different in 2018 and 2023. In 2023, 37.1% of all home purchase loans were taken out by male-female applicant partners, 35.79% by single male applicants, 24.14% by single female applicants and the rest 2.96% by same-sex co-applicants.

Bank Profiles

Pittsburgh Neighborhood Profiles

Data and Methodology

Definitions

1. CRA bank size
   As per the 2025 CRA standard that this report uses,
   Large bank is a bank that, as of December 31 of either of the prior two calendar years, had assets of more than $1.609 billion. The intermediate small bank is bank with assets of at least $402 million as of December 31 of both prior two calendar years and less than $1.609 billion as of December 31 of either of the prior two calendar years. A Small bank is a bank with assets less than $402 million as of December 31 of both prior two calendar years. For this report, we have grouped together small banks and intermediate small banks while analyzing relative performance of banks

2. Income Groups
   Low Income
   Individual income that is less than 50 percent of the area median income, or a median family income that is less than 50 percent, in the case of a geography.
   Moderate Income
   Individual income that is at least 50 percent and less than 80 percent of the area median income, or a median family income that is at least 50 percent and less than 80 percent, in the case of a geography.
   Middle Income
   Individual income that is at least 80 percent and less than 120 percent of the area median income, or a median family income that is at least 80 percent and less than 120 percent, in the case of a geography.
   Upper Income
   Individual income that is 120 percent or more of the area median income, or a median family income that is 120 percent or more, in the case of a geography.
   Low- and Moderate-Income (LMI)
   For analysis, low income and moderate income is grouped together to create low-and-moderate income group.

3. Metropolitan Statistical Area (MSA)
   An MSA is a core area containing at least one urbanized area of 50,000 or more inhabitants, together with adjacent communities having a high degree of economic and social integration with that core. Pittsburgh MSA consists of Allegheny, Armstrong, Beaver, Butler, Fayette, Lawrence, Washington and Westmoreland counties.

Data Sources

The following are the list of data sources for this study:

• Housing Mortgage Disclosure Act (HMDA)
  HMDA 2018 to HMDA 2023 Loan/Application Register (LAR) and Transmittal Sheet (TS) data is used for this study. The data is filtered to only include loans originating within the Pennsylvanian counties of Allegheny, Armstrong, Beaver, Westmoreland, and Washington.

• Federal Deposit Insurance Corporation (FDIC) market share report
  FDIC report on market share of deposits in Pittsburgh MSA 2024 helped to create a list of depository institutions within Pittsburgh MSA. This list of financial institutions was used to create lending profiles of each bank with at least one branch in Pittsburgh MSA.

• Allegheny County Department of Human Services (DHS) Tract to Pittsburgh Neighborhood tables
  Tract to Pittsburgh Neighborhood (2010) and Tract to Pittsburgh Neighborhood (2020) are look-up tables where each relevant GeoID is grouped under a Pittsburgh neighborhood. HMDA 2020 and HMDA 2021 follow 2010 census boundaries and thus the 2010 lookup table was used to add a field for Pittsburgh neighborhoods to the HMDA loans data. HMDA 2022 follows the 2020 census boundaries and thus the 2020 look-up table helped there.

• NCRC crosswalk of financial institutions
  This is a dataset connecting LEI (Legal Identifier of Financial Organizations) and name of financial institution with the parent company and separates the institutions as “Bank or Affiliate”, “Mortgage company” or Credit Union”. This dataset from NCRC was used to categorize the lenders present Greater Pittsburgh as per the three criteria mentioned.

HMDA Data Dictionary

Applicant Information

Race and Ethnicity
Within the dataset, the Home Mortgage Disclosure Act (HMDA) requires lenders to provide demographic information related to age, gender, race and ethnicity, and income status of the applicants. Applicants can voluntarily supply this information, or financial institutions can record it on a visual or surname confirmation basis. Since 2018, the FFIEC has increased the level of specificity in recording race and ethnicity. An example would be that an applicant of Hispanic or Latino ethnicity could voluntarily choose to identify as Mexican, Puerto Rican, Cuban, or other Hispanic or Latino origins.

 For the purposes of this study, PCRG has reconstructed the broader race and ethnicity categories to simplify analysis. Because there are multiple spaces to select identity for race or ethnicity, if someone identifies as a broad category and a selective category, they are designated as the broadest category. An example would be an applicant who identifies as Asian (a value of 2 in the HMDA dataset) and again as Japanese (a value of 24).

Furthermore, PCRG also included both race and ethnicity into a single demographic characteristic, using the following criteria:

• Race/Ethnicity Not Available – the applicant (and co-applicant if applicable) chooses not to identify as any race or ethnicity

• White – the applicant (and co-applicant if applicable) identifies as both white and of non-Hispanic or Latino ethnicity

• Black – if single applicant identifies as Black or African American and as non-Hispanic or Latino ethnicity. If co-applicants, both identify as non-Hispanic or Latino and at least one applicant/co-applicant identifies as white and the other identifies as Black or African American. If both applicant/co-applicants identify as one or more minority races, PCRG gives precedence to applicant race over co-applicant race. This logic also applies to American Indian or Alaska Native applicants, Asian applicants, and Native Hawaiian or Other Pacific Islander Applicants. Native Hawaiian or Other Pacific Islander is shortened as HoPI in this study.

• If an applicant and or co-applicant identifies as Hispanic or Latino, they are designated as a Hispanic or Latino applicant.
  

Age and Gender

HMDA datasets provide information on the age and gender of the applicants. Gender is either self-identified as either male, female, both male and female, or not applicable. It may also be listed as not provided by mail, internet, or telephone application. This also applies for the co-applicants if they are included. Similarly, age is bucketed into groups and includes an additional flag for whether the applicant is over the age of 62.

 Income Status

HMDA reporting also provides the applicant’s reported gross annual income and the ratio, as a percentage, of monthly debt to monthly income. Additionally, HMDA also includes the applicant’s credit score type, but not the applicant’s actual credit score.

For the purposes of this analysis, PCRG calculated the applicant’s income group using the income thresholds for area median income. This process uses the applicant’s reported income relative to the FFIEC provided Pittsburgh Metro Area median family income as per the ratios mentioned in the income definitions above.

Additionally, PCRG treats applicants without a reported income as a separate category of ‘Unknown Income.’  While loans reported without an income are typically associated with loans purchased from another institution, there is a significant number of primary originations that do not have reported applicant income levels. PCRG kept these records for analysis as ‘Unknown Income’ to preserve as many records as possible, and to allow for additional analysis of purchased loans in the future.

HMDA records applicants’ debt-to-income ratios, grouped in various sized bins, which can be used to clarify denial reasons.

Loan Information

Loan Actions

1. Applications
   PCRG is treating loan applications as loan actions in HMDA where the ‘action_taken’ field in the dataset is less than or equal to 5, meaning it does not include purchased loans or preapproval actions taken.

2. Denial Rate
   The denial rate for any given group is equal to the number of loans denied divided by total applications.

3. Denial Reasons
   HMDA reporting requires mortgage lenders to submit the reasons for a denied loan application in the HMDA dataset and allows lenders to list up to four discrete denial reasons. Because lenders can deny loans for multiple reasons, aggregations of denial reasons may differ over reporting periods and not equal 100 percent. Reasons for loan denial include Debt-to-income ratio, Employment history, Insufficient collateral, Credit history, Insufficient cash, Unverifiable information, Incomplete credit application, Mortgage insurance denied, and Other reasons.

Loan Purpose

The HMDA dataset includes three major categories of loan purposes: home purchase loans, home improvement loans, and refinancing loans. After 2018, HMDA further distinguished refinancing loans by dividing standard refinancing and cash out refinancing into separate categories. Additional loan purposes include other nonstandard loans and a rarely used non-applicable category. For the purposes of this analysis, PCRG has aggregated refinanced loans into a single category. Much like HMDA reported loans, the focus of report mainly concerns home purchase, home improvement, and refinancing loans – but we did not exclude other and non-applicable loans from aggregate totals.

 Loan Types

There are multiple categories for loan types in the HMDA dataset. A conventional loan is a mortgage loan for a one-to-four housing unit property that is not insured or guaranteed by the federal government, and therefore riskier to the lender. Financial institutions often make these loans to applicants with higher income levels, well-established and high credit scores, and larger down payments. The other loan types are loans that are either insured or guaranteed by the federal government through either the Federal Housing Administration, the Department of Veterans Affairs, or the Department of Agriculture’s Rural Housing Service or Farm Service Agency. These loan products are typically available to a wide variety of applicants, often with modest incomes and credit scores, and lower down payment requirements – that may require mortgage interest to be paid by the borrower.

 Property Types

HMDA breaks down housing units by unit size and construction type. PCRG’s primary focus of residential lending is on 1–4-unit housing, separated from multifamily housing (housing with more than 5 units). Additionally, the construction type focuses on housing units built on site or manufactured (prefabricated housing units and/or mobile homes).

 Other Loan Factors

• Occupancy Type
  HMDA tracks whether loans are the primary residence of the applicant (also referred to as owner-occupied housing units), or whether the unit is a secondary residence or investment property.

• Exempt Reporting Flags
  Small and intermediate-small banks, as well as mortgage firms and credit unions are not required to report certain characteristics of a loan application such as: whether the loan is a reverse mortgage, whether the loan is an open-end line of credit, or whether the loan is for a business or commercial purpose.

 Location Information

Loan applications receive a series of location-based codes and references to accurately locate the property of the application down to the census tract level. This includes:

• MSA-MA: Metropolitan Statistical Area or Metropolitan Division, an area that has a population greater than 50,000 residents and at least one urbanized core area. For loans in the seven county Pittsburgh, PA MSA, this code is 38300.

• State Code: the two-letter abbreviation for the state - PA for Pennsylvania

• County Code: the five-digit FIPS code which gives the state and county of the loan application. For Allegheny County, this would be 42003, where 42 represents the state (Pennsylvania) and 003 represents the county (Allegheny)

• Census Tract: the eleven-digit FIPS code, which gives the state, county, and census tract location. An example would be the Central Business District of Pittsburgh (also known as the Golden Triangle) would be 42003020100. 020100 would be the census tract within Allegheny County (42003). Census tracts for Pittsburgh, as well as all other townships, municipalities, and locales in Allegheny County are all listed with the 42003 census tracts. A census tract is typically an area of between 2,000 and 8,000 residents – and can conform to (but not exactly with) neighborhood, town, and other boundaries. There are some census tracts within the City of Pittsburgh with little to no population, often representing city parks, graveyards and cemeteries, and fully commercial districts.

 Income Information

Along with the location information, the FFIEC reports tract level information that is useful for analysis. This includes the tract level median family income as a percentage of the metropolitan statistical area median income (annually adjusted by the FFIEC).

 For the 2023 HMDA data, this uses the 2020 census boundary lines. The 2022 HMDA data is the first to use the 2020 census tract boundaries, and the 2020 ACS 5-year estimates for tract level data, with income adjusted to 2022 dollars. FFIEC area median incomes are like HUD area median incomes and assume an average family size of four. Using the calculations discussed previously, PCRG was able to determine each census tracts’ income level as either low-, moderate-, middle-, or upper-income. Additionally, some tracts are given a designation by the FFIEC as ‘unknown income.’ This is generally reserved for areas that do not have significant households, families, or traditional housing units – such as the tracts that represent city parks, cemeteries and graveyards, and fully commercial areas. This also includes areas like parts of the Oakland neighborhood in Pittsburgh that is predominately student housing and other university space, as well as Marshall-Shadeland which housed State Correctional Institution – Pittsburgh until it’s closure in 2017.