Financial Institutions In New York City Discourage Prospective Small Business Applicants
Photo: ©kalafoto – stock.adobe.com
Bank staff were more cordial, affirming and helpful to White customers than to Black or Hispanic ones. White ‘mystery shoppers’ were more consistently treated with courtesy – and were more than twice as likely as Black participants to be offered one-on-one help completing an application. Bank employees encouraged White customers to apply for a small business loan more often than Black and Hispanic ones, by a statistically significant margin.
Bank staff in many cases appeared to violate federal anti-discrimination laws in their interactions with non-White small business owners. Out of 36 multi-layered tests our fair lending review revealed 51 different instances of discrimination in violation of the Equal Credit Opportunity Act.
Anneliese Lederer, Director, Fair Lending & Consumer Protection, NCRC
Sara Oros, Program Coordinator, Fair Housing/Fair Lending, NCRC
In Collaboration with
Dr. Sterling Bone, Professor of Marketing, Huntsman ProSales Director, Utah State University
Dr. Maura Scott, Persis E. Rockwood Professor of Marketing, College of Business, Florida State University
Dr. Martin Mende, Jim Moran Professor of Business Administration, College of Business, Florida State University
Dr. Glenn Christensen, Garrett Research Fellow, Associate Professor of Marketing, Brigham Young University
Three years after the COVID-19 pandemic forced people to stay at home, we are now understanding its economic effects. To prevent a complete economic collapse, governments and financial institutions introduced various policies. In April 2020, Congress established the Paycheck Protection Program (PPP) to help small businesses with emergency funding, which could become a forgivable grant if used for certain expenses. The Federal Reserve also maintained low interest rates to encourage lending.
Major banks, including JP Morgan Chase, Wells Fargo and Citi, suspended Home Equity Line of Credit (HELOC) programs to protect both themselves and consumers from the unknown economic effects of the pandemic. Before the crisis, HELOCs allowed business owners to access credit based on their home equity for business purposes.
The pandemic coincided with mass uprisings for racial justice across the United States. After George Floyd’s murder in 2020, many organizations, including financial institutions, pledged to change policies and promote racial equality.
In 2022, as Americans resumed pre-pandemic activities, the National Community Reinvestment Coalition (NCRC) studied the changing small business lending landscape. PPP lending ended. Banks resumed considering HELOC applications including those intended to finance business expansion. Americans launched 10.5 million new businesses after the pandemic began. Our goal was to understand who can access small business credit and what credit products are now available in this much-altered business lending landscape.
NCRC conducted multi-layered small business tests in New York City during the summer of 2022. This project drew upon years of experience with such methodologies: NCRC has been conducting “mystery shopper” testing across a variety of consumer-facing financial service sectors for more than 20 years.
We chose New York for its status as North America’s financial capital and for its progressive, multicultural reputation. With 121 financial institutions registered in New York State and a strong consumer protection agency, the region provided a suitable environment for this research.
As with our previous work, this study was designed to answer three related research questions. When minority and non-minority small business owners with similar economic and business profiles seek loan capital, do they (1) receive the same information, (2) face the same requirements for providing information and (3) experience the same levels of encouragement or discouragement?
NCRC, in collaboration with our academic partners, conducted pre-application multi-layered matched audit testing in New York City based on race and national origin. Multi-layered matched testing consists of testers visiting a physical bank branch to determine if there is a difference in treatment. We designed a 3-paired testing experiment with testers whose primary racial or ethnic identity was either Hispanic, Black or White. The multi-layered matched testing only occurred in the pre-application phase, with the tester requesting information about loan products. For this round of testing, we conducted 36 male multi-layered matched tests consisting of White, Black and Hispanic testers using profiles to represent small business owners. From July 17 – August 4, 2022, our testers contacted 36 bank branches representing 28 different financial institutions. We had a total of 108 interactions with bank branches.
The multi-layered mystery shoppers had nearly identical business profiles and strong credit histories to inquire about small business products. The profiles of all testers were sufficiently strong that, on paper, they would all qualify for credit. Furthermore, the Black and Hispanic testers’ profiles were intentionally designed to be slightly better than their White counterparts in income, assets and credit scores. This was done to make it a more conservative test for differential treatment. Immediately following the interaction, testers were asked to answer yes or no questions about whether bank employees made specific behaviors, queries and comments. They were also required to provide a narrative of the interaction. Each interaction was audio and video recorded.
We audited the marketplace by randomly selecting from all small business lending institutions to represent a broad cross-section of the small business lending marketplace. The banks selected ranged from lenders with assets over $10 billion to community banks. Tests were conducted in person by all of the testers visiting the same bank branch. The purpose of the research was to determine the baseline customer service level that male testers of different races and national origins received when seeking information about small business loans to help stay afloat as we exited the COVID-19 pandemic.
Testing, a critical tool for fair lending enforcement, is used to assess equal access to credit. Federal agencies also conduct similar testing to investigate when they have been alerted to suspicious behavior. Upon implementation of Section 1071 of the Dodd-Frank Wall Street Reform and Consumer Protection Act, matched pair testing will be the only method to measure pre-application small business discrimination and discouragement since banks will not be required to collect demographic data on loan inquiries before an application is submitted. The Department of Housing and Urban Development (HUD) partners with fair housing groups to do testing under their Fair Housing Initiatives Program (FHIP). The Department of Justice (DOJ) Civil Rights Division has taken on many cases developed from its testing program. Additionally, the Consumer Financial Protection Bureau (CFPB) took enforcement action against BancorpSouth Bank after implementing testing that revealed multiple threats to consumers’ fair and equal access to mortgages. The US Supreme Court has upheld testing as a crucial private enforcement tool in fighting against civil rights violations.
The Equal Credit Opportunity Act (ECOA) makes it illegal for a creditor to discriminate at any point during the credit process. There are three ways that discrimination in the credit process is expressed: overt statements, disparate treatment and disparate impact.
Overt statements discrimination occurs when a creditor’s employee makes specific discriminatory or stereotyping statements about a member of a protected class. Such a statement can occur both verbally towards the customer or in passing to a coworker. It can also take place via email.
Disparate treatment discrimination occurs when two equally qualified applicants or prospective applicants receive different rates or information. The only difference between these applicants or prospective applicants is their status as a member of a protected class.
Disparate impact discrimination, meanwhile,is a neutral business policy or practice that has a disproportionate negative effect upon a protected class.
We analyzed our testing data using two approaches: a fair lending review and a market analysis using chi-square statistical difference tests. For the chi-square difference, we applied statistical analysis to evaluate whether differences in the interactions between White testers, Black testers and Hispanic testers were significant across all banks visited. The chi-square test for independence is a particularly robust way for social scientists to evaluate whether or not there are substantial differences in outcomes between groups. Testers were asked dichotomous (“yes or no”) questions about their interaction with bank personnel in a survey.
For the fair lending review, each video, transcript and narrative of the interaction was qualitatively analyzed independently by two fair lending experts as matched-pair sets to determine a difference in treatment under fair lending standards based on a difference in treatment on race and/or national origin.
Who We Tested:
We tested 28 different financial institutions, which is 23% of the institutions within the state of New York. All of these institutions are bound not just by the anti-discrimination law of ECOA but also by the Community Reinvestment Act (CRA).
The CRA’s purpose is to stop redlining through an affirmative requirement to invest in the communities that financial institution’s branches are located in. Every three years, these institutions are evaluated by regulators to see how well they are implementing the CRA. The CRA exam is broken into three different components, with some variability depending on the size of the bank: lending (such as single family mortgages and small business loans), investments (as in multifamily housing equity investments) and services. The scope of these tests is subject to revision in the CRA rulemaking pending at the time of this paper’s drafting. CRA testing examines how well these institutions are making loans to low- to moderate-income (LMI) communities, derived from tract-level Census data. CRA is not explicitly race-based, rather it examines lending in LMI areas which incorporate all races. In some areas LMI can be a proxy for race or national origin due to a specific LMI area also having a disproportionate amount of Black or Hispanic population living there.
Financial institutions are graded on lending in three areas as noted above, including: mortgage loans, philanthropy and small business loans under 1 million dollars made within the bank’s CRA assessment area. The CRA examination data is publicly available for anyone to view. Of the 28 financial institutions that we tested, 13 institutions have a rating of Outstanding on their most recent CRA exam, 14 institutions received a rating of Satisfactory on their most recent exam and 1 institution had a rating of Needs Improvement. An outstanding and satisfactory score innately is intended to indicate that these institutions are serving the community’s needs and making loans available to all borrowers, though CRA exams have been criticized for widespread grade inflation.
Many of the financial institutions we tested have not only received strong CRA ratings but have also made public pledges to support racial equity by meeting the needs of minority business owners. Such institutions would be expected to perform well in testing such as ours: A high CRA grade and/or voluntarily issued public commitment to racial equity should correspond to staff-level behavior, fostering the expectation that all of our minority testers would be treated as well if not better than their White counterparts.
Unfortunately our results suggest the opposite. White testers were more consistently treated with courtesy – and were more than twice as likely as Black testers to be actively encouraged to apply for credit or to be offered one-on-one help completing an application.
Favoring the Control Group Over the Protected Group
We conducted 36 multi-layered tests of 28 institutions. Multi-layered testing results in us sending in a White, Black and Hispanic Male testers into the same branch of that financial institution. Our White, Black and Hispanic Male testers inquired about credit products in the pre-application arena. Overall we found the following results:
Our testers experienced discrimination in 37 of the 108 total interactions they had with banks during the probe. Our qualitative analysis revealed 51 instances of discrimination in those 37 discriminatory interactions, broken down into the following behaviors: discouragement, difference in products offered, difference in information provided and difference in information required or requested.
Analysis of the Multi-layered Tests
In part because of the strong anti-discrimination provisions of ECOA, we do not observe many overt statements of discrimination towards specific protected classes. However, during this round of testing we observed a bank employee making antisemitic comments. Although the focus of our testing was race and national origin, we identified comments that were discriminatory under the protected class of religion. While speaking with a Hispanic tester, a bank agent made a comment about Jews, specifically that Jews are smart with money. While this bank branch is located in Brooklyn, it is not located in a majority Jewish section of Brooklyn. There has been an increased rise in antisemitism in the US. In 2022, there were “3,697 antisemitic incidents throughout the United States. This is a 36% increase” from the prior year.
There was no reason to bring up the Jewish community at all during this interaction.
The discrimination that our testing revealed through the fair lending review is how nearly identical applicants are treated differently based upon being a member of a protected group.
Reg B, the implementing regulation of ECOA, contains a provision that discouragement of an applicant or prospective applicant in applying or pursuing credit due to their protected status is discriminatory. Match-pair and multi-layered testing provides insight into how discouragement manifests in the pre-application arena before data collection laws are triggered. Discouragement stops applicants and prospective applicants from entering into the funding funnel.
Data analysis across the marketplace shows statistically significant behaviors at the beginning and the end of the interaction which discourages potential applicants from applying for credit. An applicant’s initial experience with a bank employee sets the tone of the whole interaction. The data shows that both Black and Hispanic testers were more likely to be treated differently relative to White testers.
In response to the below questions, Black and Hispanic testers experienced these behaviors by bank officers at a lower statistically significant rate than the White testers.
These actions make a potential customer more comfortable and demonstrate to the potential customer the bank employee’s willingness to engage in this relationship. Discouragement occurs when these encouraging actions are not present during the interaction for protected prospective applicants. The “Offered you a seat” metric was not statistically significant for the Black v White or the Black v Hispanic comparison.
We also identified discriminatory conduct during the concluding portion of a potential consumer’s meeting with a loan officer. Like the tone-setting initial interaction, behavior in the concluding phase of a first conversation is especially important: It is the last impression that a potential customer remembers.
This stage is where the loan officer may offer to continue following up with the potential client or request that the prospective business owner reach out if they need any additional assistance. In this round, White testers experienced less discouragement and more help in the conclusion of the conversation. These behaviors were measured through whether or not testers were: encouraged to apply, offered to help to complete an application and thanked for coming into the branch. For example, testers were asked the question: “Did the representative encourage you to apply?”. White testers were encouraged 81.1% of the time, far more than Hispanic testers (51.50%) and more than twice as often as Black testers (36.1%).
The fair lending review of the multi-layered testing revealed the following examples of discouragement.
Bank 1: The Hispanic tester was told that they need to be a customer of the bank for at least a year before they could get a loan. The White tester was told that they will first need to be a customer of the bank and then they could get a loan in about two weeks. The loan officer told the White tester about another client who had a previous relationship with X bank but then left for four years. That person was unsuccessful at getting a loan with their current bank so they came back to X Bank and got a loan immediately.
Bank 2: At this financial institution, the White tester was able to meet with someone in person. Both protected testers were unable to speak with someone in person and were provided different information about the next steps to establishing a relationship with the financial institution that were not as warm and welcoming as the experience the White tester received. Specifically, the Black tester was told to make an appointment online and the Hispanic tester was told they would need to establish a business relationship in order to get more detailed information about products.
The type of information requested of potential applicants in the pre-approval arena impacts the type of applications they submit. The more robust the application, the more likely a bank is to quickly accept it. Small businesses are harmed when they do not receive all the information about required documentation at one time because they have less time and resources to allocate in seeking financial products. It is more likely that the person seeking funding to expand the business is also critical to the day-to-day functions as well.
This round of testing revealed three different instances where bank employees required different information from White and non-White testers: differences in information asked about the business, in whether they were asked about loan collateral and difference in follow-up regarding which documents are required.
Information Asked About the Business
There were statistically significant differences in the information asked by the bank employee to the tester about the business depending if the employee was speaking to the White tester or the non-White testers. We measured if testers were asked for six types of information:
- annual business sales
- average core balance
- company profits
- existing loans
- income statement
- balance sheet
Our testing revealed that the White tester was asked at statistically significantly higher rates about their business’s average core balances, existing loans, income statement and balance sheet compared to the aggregate of the Hispanic and Black testers. Asking this type of information demonstrates engagement in the pre-application arena and a desire by the employee to engage deeper with the prospective White applicant.
Difference in Asking About Loan Collateral
The statistical analysis of the loan information bank employees sought from White testers versus what they asked of Hispanic and Black testers reveals a significant difference in being asked about collateral available to use with the loan. This inquiry could be a factor in discouraging business owners.
Bank staff told non-White testers about collateral requirements roughly twice as often as they did when speaking to White testers. Staff mentioned collateral to White testers 16.2% of the time. Aggregating the experiences of Black and Hispanic testers who represent the legally protected classes under ECOA, staff mentioned collateral requirements to Protected Class testers 31.9% of the time(chi-square = 3.042, p = .081).
Follow-up Documents Requested
Follow-up interactions usually reinforce the types of documents required when applying. The fair lending review of the follow-up interactions revealed a difference in treatment. At one financial institution, both the control and Black protected tester received email follow up from the same loan officer. However, the loan officer provided a different list of required documents to each tester.
Both testers were told that they would need to submit three years of personal tax returns, a financial statement form and internally-prepared income statements and balance sheets as of June 20, 2022.
But the White control tester was also told to submit these documents:
- Personal financial statement
- Corporate documents (articles of organization, filing receipt, bylaws)
- Business information sheet and provide a copy of your driver’s license
- Personal information sheet
The Black tester who was not told about those four additional documentation categories was further told he needed to submit three years of business tax returns as well, while the White tester was told to only provide personal returns.
The difference in the more robust information request highlights the ease with which the control tester would be able to access a loan compared to the Black protected tester even though on paper, the Black protected tester is better off financially.
Potential applicants who receive different information are unable to make the same informed choice. Our fair lending review found that at one financial institution, staff gave all testers different rate information. The White tester received a rate range of 6.5-14.5%. The Black tester was told “7.25 on the low side to a high of 12%”. The Hispanic tester was told “prime plus 7.99 up to prime plus 17.99”.
The higher an interest rate a borrower has, the less disposable income they have to both invest in their business and in themselves. Again, on paper, both protected testers were more qualified than the White tester who was quoted rates that start at 6.5% – 0.75 percentage points lower than the starting rate for the Black tester and 1.49 percentage points lower than the Hispanic tester.
This round we returned to in-person testing for our small business lending research. The market has changed significantly because of the pandemic. We observed these changes through multiple additional physical barriers testers experienced (social distancing, clear barriers, masks) and potential consequences of those barriers. Testers generally spent less time speaking with loan officers this round when compared to pre-pandemic in-person testing. In addition, testers received fewer follow up emails compared to previous rounds as well, which could be the result of these shorter conversations.
The fair lending review continues to reveal concerning patterns in the small business lending arena. To combat these differences in treatment based upon race and national origin, we propose the following actions:
Implement Compliance Management Systems (CMS)
Banks should maintain a continuous review of CMS to ensure that models are using less discriminatory alternatives in their predictions. Furthermore, CMSs need to focus on discouragement. Despite the Townstone Financial decision, discouraging consumers from applying for loans remains a violation of ECOA.
Conduct matched-pair testing and fair lending reviews
Financial institutions need to use matched-pair testing in their fair lending compliance processes to ensure that they are not violating fair lending laws and not discouraging applicants from applying for loans. The testing and the results can be used to improve employee training.
Include implicit bias training in fair lending training
Financial institutions need implicit bias training. This type of training will help employees realize their own biases and their effects on consumer behavior.
Monitor customer follow up attempts
In our previous round of testing in 2021 we highlighted inconsistencies with the distribution of material over email to prospective applicants. The same inconsistent follow-up practices continued in the new tests. This is a potential fair-lending issue as well as a general compliance issue.
In addition to these recommendations we also urge regulators to continue to increase enforcement efforts.