NCRC’s Comment on the CFPB’s Proposed Rule to Strip Section 1071 Data Collection

CFPB Would Obscure Lending to Small Farms

Just as with lending to businesses of various sizes, the CFPB would obscure lending to small farms by its proposal to exclude agricultural lenders altogether from the reporting requirements. The agency’s rationale is that lending to small farms employs different underwriting and other considerations in part because agricultural lending uses different collateral. The CFPB states, “Agricultural loans are often secured by biological-based assets such as crops or livestock, which are subject to variables and risk from weather and disease. These characteristics create unique underwriting challenges that make such loans difficult to compare to those in other industries.”[27]

However, the agency could identify loans readily as made by agricultural lenders or for the purpose of financing farms. This would alleviate concerns about comparing agricultural and non-agricultural lending since analysts could then assess farm lending trends separately from non-farm trends. Furthermore, excluding small farm lending from the Section 1071 database frustrates the community development and fair lending purposes of the statutes. Last year, CRA reporting banks issued about 197,000 small farm loans.[28] When considering lending from other institutions, the CFPB probably is obscuring more than 200,000 loans annually, impeding efforts to understand whether small farms or minority and women-owned farms can access credit.

In its 2023 rule, the CFPB included agricultural loans in Section 1071 data, suggesting that lending to farms can pose fair lending risks since it is conducted on a decentralized basis and involves several discretionary elements. In its proposed 2023 rule, the CFPB stated, “The share of minority representation in farming, particularly that of Black farmers, has declined sharply over the last 100 years. Based on the disposition of numerous lawsuits alleging discrimination against minority farmers, the Bureau believes that credit discrimination may play a role in this decline.”^[29]

The history of discrimination against African American farmers also illustrates the importance of collecting data regarding whether the loans were issued by a public sector lender or had a federal guarantee. Under the largest discrimination settlement in United States history, the Department of Agriculture has made more than $1 billion of payments to African American farmers that experienced discrimination. The Department of Agriculture lending programs were operated by biased county organizations whose elected committee members were virtually all white (1% were African American).^[30]

Given this history, Section 1071 data must include agricultural lending and needs to make sure that public sector lending programs and support systems are held accountable. The CFPB proposes to exempt Farm Credit System (FCS) lenders because they are smaller lenders and hence may have difficulties submitting high quality data.[31] The CFPB, however, also states that the Farm Credit Administration (FCA), the regulator of the FCS, requires FCS lenders to submit data.[32] Intuitively, then, the FCS lenders would have had several years’ experience of submitting data. Moreover, the FCA and CFPB could coordinate their data requirements and eliminate duplication as much as possible. Finally, the CFPB discounts the sophistication of FCS lenders. These are customer-owned cooperatives that support over 600,000 farms nationwide, issuing more than 250,000 loans in 2024.[33] They offer a range of products and services to farms, suggesting that they are more than seat-of-the-pants operations.[34]

In addition to critical fair lending issues, NCRC believes that better understanding credit needs of small farmers is vital to helping ensure the survival of a vibrant sector of the economy. The CFPB stated that 98 percent of the farms in this country are family farms and 90 percent of family farms were on the smaller side, bringing in $350,000 or less in revenue on an annual basis.^[35] In order to promote adaptive and sustainable agriculture, our country cannot rely on big agribusiness but will need a viable sector of smaller farmers also. Section 1071 data becomes all that more important in assessing the extent to which the credit needs of small farmers are being met.

Deletion of Reporting by Merchant Cash Advance (MCA) Lenders Distorts the Picture of the Market

Although merchant cash advance (MCA) lenders are high volume lenders usually offering more than 1,000 loans according to the CFPB, the CFPB nevertheless proposes to exempt them from data reporting because of their unusual lending model.[36] In its proposal, the CFPB asserts that uncertainty exists as to whether MCAs are a traditional form of credit.[37] Under the MCA model, a MCA lender extends funds to a small business which agrees to repay the MCA a regular percentage of its sales. This differs from a regular loan in that interest and fees are considered differently. However, the CFPB in its 2023 rule considered that MCAs are like a traditional loan in that repayment is regular over predictable periods of time.[38] Furthermore, a magistrate judge in the Southern District of Florida in Revenue Based Finance Coalition v. CFPB upheld the CFPB’s conclusion that MCAs can be considered a form of credit.[39]

The proposed rule nevertheless maintains that MCAs are not a form of credit and thus inclusion of MCAs in the Section 1071 database would not further the purposes of the Dodd-Frank Act. Regardless, inclusion furthers the fair lending purposes of Section 1071. As the magistrate judge noted in Revenue Based Finance Coalition v. CFPB, MCAs are relied upon by traditionally underserved small businesses since the CFPB had found that approval rates for MCAs are considerably higher than for other forms of credit.[40] The CFPB also revealed that minority-owned firms rely on MCAs to a greater extent than white-owned firms.[41] This form of relatively easy credit might expose traditionally underserved businesses to predatory practices.

A study by Opportunity Fund found that MCAs had an average Annual Percentage Rate (APR) of 94 percent, almost three times higher than some of the laxest state usury limits.^[42] In some of the most egregious cases, MCAs had rates that exceeded 1,000 percent.^[43] The CFPB, in developing its 2023 rule, cited high default rates, due in part to large percentages of revenues or receipts that MCA providers took from the small businesses.^[44] In addition, MCA providers often require that small business borrowers sign “Confessions of Judgment” (COJs) agreements in their lending contracts. As the Congressional Research Service documents, COJs waive a borrower’s procedural due process rights, including the right to mount legal defenses against the claims of a MCA, and forces them to accept liability and automatically lose any legal disputes with lenders. ^[45] Bloomberg found the number of cases ending with a confession in favor of a merchant cash advance company in New York state rose from 14 cases in 2014 to over 3,500 cases in 2018. These COJ cases are estimated to have won the MCA industry an estimated $500 million.^[46]

The CFPB’s proposed rule errs in elevating concerns about precise definitions of credit over pressing fair lending concerns. Without comprehensive data on MCAs, the Section 1071 database cannot adequately investigate whether abusive and discriminatory lending is occurring. The CFPB disingenuously asserts that states regulate MCAs, so it is best to leave enforcement at the state level.[47] Yet, most states have minimal consumer protections for MCAs.[48]

Deletion of Data Points on Characteristics of Small Businesses Thwarts Community Development and Fair Lending Purposes of Section 1071

The CFPB is proposing to delete important discretionary variables related to the characteristics of small business loan applicants. The agency reasons that although these data points have value, the cost to lending institutions and the complexity associated with collecting these data points outweigh the value of these points. The CFPB states:

Accordingly, if the Bureau now believes that the relative utility of the data is not strong enough to justify the additional operational complexity for financial institutions, that is sufficient reason to propose removing the discretionary data point, even if the discretionary data point would otherwise advance the purposes of the statute.[49]

The CFPB preliminarily concludes that initiating the data collection with an expansive rule that covered more data points would tend to make the initial collections more complicated and result in lesser data quality and integrity.[50]

The CFPB applies this logic to several of the data points of the 2023 final rule.

Number of Workers More Important than CFPB’s Portrayal

Regarding the number of workers of a small business, the CFPB asserts that it would be difficult for small business loan applicants to identify how many workers are employed because small businesses employ seasonal or other part time employees. Yet, this difficulty amounts to the exception disproving the rule. As the CFPB admits, most small businesses do not have employees.[51] For most small business loan applicants then, estimating the number of workers would not be an issue.

For those that do have employees, CFPB guidance could clarify how to report the number of workers. The proposed rule acknowledges clarifying guidance of this nature.[52] In addition, if the CFPB was still concerned about ambiguity, it could issue more instructions such as specifying that employees who are seasonal should not be counted or that those who work less than a specified time period (such as four or six months over a year) should not be counted. The lender could further be instructed to rely on the judgment of the small business applicant. While this could introduce some inconsistencies in the data, tolerance for inconsistencies is justified if the data has utility. Moreover, in the majority of instances in which a small business has none or just one or two employees, little ambiguity exists.

The CFPB suggests that the operational difficulty overwhelms the utility of the data point concerning number of workers since this data point only meets the statutory purpose of community development and not the purpose of facilitating fair lending enforcement.[53] Disqualifying a data point because it only meets one of two statutory purposes is an odd criterion for deletion, especially if the variable has significant community development implications. In fact, previous CFPB research indicates that women-owned and minority-owned small businesses often do not have employees. Perhaps lending could be used in a manner that would effectively support the expansion of these businesses and their hiring of employees. Stakeholders could not begin to assess these possibilities if the data is not collected.

Women-owned businesses with employees account for about 20% of all employer businesses, number approximately 1.1 million and generate $1.8 trillion in revenue annually.^[54] The CFPB estimated that an additional 10 million women-owned firms do not have employees.^[55] These statistics suggest that if access to credit meaningfully increased the percentage of women-owned business with employees, then revenues and output of women-owned businesses could rise substantially, benefiting the entire economy.

Similarly, research found that although 12.6 percent of the U.S. population was Black, only 2.1 percent of small businesses with employees were Black-owned. ^[56] Again, our nation could likely be missing out on community development opportunities if data is not collected on number of workers employed by small businesses applying for and receiving loans. Finally, the CFPB prematurely is dismissing the fair lending utility of the data point involving number of workers. Econometric analysis identifying racial or gender disparities in receiving loans or in denials seeks to control for as many variables regarding small business characteristics as possible. The capacity to hire and retain workers would be an important variable to control in assessing lending disparities.

NCRC appreciates that the CFPB is retaining some important variables on the characteristics of small businesses. These variables will make it possible for analysts to determine if businesses by sector or age are receiving loans without too much difficulty, and if they are not, how to overcome barriers to lending and community development. In particular, the CFPB states that “it is not proposing to remove NAICS code, time in business, and number of principal owners because those discretionary data points are generally integral to collection and understanding of statutorily required data points and the Bureau did not receive evidence during the implementation period of logistical challenges not previously considered.”[57]

However, as important as it is to retain the NAICs code (determining the sector of small businesses), time in business, and number of principal owners, what distinguishes these variables from the number of workers is that they did not receive any adverse comments during or after public comment periods. It is indeed an extraordinarily high standard for a data point to clear; it will only survive if no lender or trade association issued a comment that disparaged it. The standard for deletion must be strenuous and must rest on the strength of evidence of logistical or operational difficulties. The evidence concerning number of workers is weak and does not meet this standard.

Deleting Small Busineses with More than $1 Million in Revenue a Huge Loss to Explanatory Power of Database

 In deleting reporting on small business loan applicants with revenues between $1 to $5 million, the CFPB proposes a drastic cut in the size of the Section 1071 database by upwards of 50 percent as described above. A reduction of this magnitude must be undertaken carefully and justified by showing that the observations to be deleted do not involve the experiences of small businesses. The CFPB not only fails to do so, but its explanation of burden reduction is unconvincing.

The CFPB offers three reasons for deleting data collection for loans to businesses between $1 million to $5 million in revenue. It states that CRA focusses on lending to small businesses with revenues less than $1 million and that the Equal Credit Opportunity Act (ECOA) requires lending institutions to provide notices of adverse actions (denials of credit) to small businesses with less than $1 million in revenue. The third reason is that lenders generally define small businesses as those with less than $1 million in revenue.[58]

None of these three reasons suggest that it is difficult to record the experiences of small businesses with revenues above $1 million. Instead, the rationale hints that a regulatory focus on small businesses with revenues under $1 million has created a habit or custom for lenders to consider businesses with these revenues as small.

The statutory purposes of Section 1071 compel a consideration as to whether businesses with revenues above $1 million experience similar opportunities and constraints as their smaller counterparts. They might have modest advantages due to their larger size, but would comparative analysis nevertheless be useful in identifying similarities and differences in their lending experiences to smaller enterprises with under $1 million in revenue so that barriers facing businesses of various revenue sizes might be better identified and overcome.

Research suggests that businesses with revenues between $1 to $5 million have similarities to their smaller peers and that inclusion in the Section 1071 database would advance the statutory purposes of Section 1071. A Federal Reserve survey found that 67 percent of businesses with revenues between $1 to $10 million have up to 20 employees while 98 percent of those with revenues between $100,000 to $1 million have that number employees.[59] Since the revenue category of up to $10 million in the Fed survey has a revenue limit twice as high as the $5 million limit in the 2023 rule, it is likely that difference in the number of employees among businesses with $1 million and $5 million in revenue is smaller. Thus, comparisons between businesses with revenues under $1 million and $5 million would be useful since they are not too dissimilar in size and number of employees.[60]

While businesses less than $10 million in revenue are more likely to be classified as low credit risk than their smaller revenue peers, the difference of 15 percentage points creates a useful contrast for analysis but is not so dissimilar as to render comparisons not meaningful.[61] Further, this difference is exaggerated since the revenue size categories are broader in the Federal Reserve survey than those used in the 2023 rule.

Again, while businesses with under $5 million in revenue likely fare better in the lending marketplace, the experience is nevertheless similar enough to be a useful yardstick for measuring the experiences of the smallest businesses. In a report conducted for the Appalachian Regional Commission (ARC), NCRC obtained Pepperdine survey data that revealed approval rates for businesses of various revenue sizes. As revealed by survey data for the first quarter of 2012, just 18 percent of the small businesses with revenues less than $500,000 who sought loans obtained them. In contrast, 35 percent of the businesses with revenues between $500,000 and $1 million and 55 percent of the businesses with revenues between $1 million and $5 million received loans.^[62] While approval rates are higher for businesses with more revenue, those with under $5 million in revenue still confronted a high denial rate of 45 percent. It would be invaluable to use the Section 1071 data containing information on businesses with up to $5 million in revenue size to assess whether these disparities in approval or denial rates narrow over time, and if not, what policy or programmatic steps could be undertaken to reduce the disparities.

Finally, the CFPB errs in its assessment of CRA’s use of business revenue size. CRA requires banks to report data on lending to businesses of all revenue sizes, not just those with less than $1 million in revenue. CRA requires more expansive data collection than even the 2023 final CFPB rule because an important performance measure in CRA exams is the percentage of loans issued to small businesses with revenues under $1 million. The denominator for the performance measure is all lending to small businesses of all revenue sizes.

Since banks have been collecting this data for decades after the 1995 regulatory update to CRA, the 2023 rule that trimmed the requirement to small businesses with revenue under $5 million is a reasonable compromise. Any further reductions are unnecessary considering the CRA data collection experience. Lastly, econometric analysis for fair lending purposes is more effective if there are more observations over a wider range of small businesses with various revenue sizes.

The CFPB Must Preserve the Racial and Ethnic Categories of the 2023 Rule as Well as LGTBQ Status

The CFPB indicates in its proposed rule that it has not made final decisions regarding disaggregated racial or ethnic categories.[63] NCRC urges the agency to retain the disaggregated categories in order to further realize the fair lending purpose of Section 1071. The final 2023 rule establishes subcategories for Hispanics, Asians, African Americans, and Native Hawaiian or other Pacific Islanders. Experience with HMDA data submissions reveals that the subcategories are popular with loan applicants and are not onerous for lenders. A 2018 revision to HMDA allowed Latino, Asian, Hawaiian or Pacific Islander applicants to report more precisely on their country or place of origin. Among eligible 2021 applicants, over 60 percent utilized these optional fields to specify their identity.^[64]

Moreover, NCRC research with HMDA data exposes a variety of differences in the lending marketplace using disaggregated race and ethnicity data. For example, NCRC found that Chinese borrowers had substantially more equity immediately after the approval of their home purchase loans than other categories of Asian borrowers and other race and ethnicities.[65] If the subcategories were deleted, the power and precision of fair lending analysis would suffer significantly, in contradiction to the fair lending purpose of Section 1071. Differences within the Asian population would be obscured in home lending, for example, making the data much less useful for identifying any possible discrimination against subcategories within the Asian population.

Deletion of Data Regarding LGBTQI+ Will Impede Fair Lending Enforcement

The CFPB proposes to eliminate data collection and reporting relating to whether a small business is owned by LGBTQI+ individuals. The CFPB cites a Presidential executive order requiring agencies “seeking information not to discuss gender identity and to refer to sex using a binary of male/female.”[66] Distributing a data collection form to lenders that would have consisted of an open-ended question to loan applicants from the 2023 rule regarding gender identity does not amount to “discussing” gender identity. The agency would distribute a form and not discuss gender identify with any applicant. Moreover, applicants would volunteer this information if they wished. The voluntary nature of the data collection would not infringe on anyone’s privacy despite the CFPB’s assertion to the contrary.[67]

A failure to collect gender identity would thwart the fair lending purpose of Section 1071. To deny discrimination against the LGBTQI+ population simply ignores history. Discrimination is illegal and impedes economic growth and opportunity. In its final rule of 2023, the CFPB cited a survey revealing that 46 percent of LGBTQI+ businesses reported not receiving credit requested compared with 35 percent of non-LGBTQI+ businesses.^[68] Continued monitoring of the lending marketplace with Section 1071 data is imperative to identify and root out these disparities.

Deleting Data Revealing Experiences of Applying for Credit Impedes the Community Development and Fair lending Purposes of Section 1071

The CFPB proposes deleting three discretionary data points – application method, application recipient and denial reasons – that are critical for comparing the experiences businesses with different characteristics have in the lending marketplace. These three data points are application method, application recipient, and denial reasons.

Application method refers to whether the application was made in-person, over the telephone, online or via mail. In its 2023 final rule, the CFPB reasoned that an important community development purpose was whether applications via the internet, telephone, or mail would compensate for serving small businesses in areas lacking branches. In addition, the CFPB suggested that application method could reveal which application methods were effective for collecting demographic data. If any of them were found to be lacking, efforts to improve collection could be targeted at the method with data collection weaknesses.[69] Deleting application method impedes the public’s ability to determine if women- or minority-owned businesses face barriers due to application method such as whether a scarcity of branches leads to the use of online or telephone applications that are more likely to result in denials.

Application recipient refers to whether the application was submitted “directly to the financial institution or its affiliate, or whether the applicant submitted the covered application indirectly to the financial institution via a third party.”^[70] This is similar to application method in that it would enable stakeholders to determine whether wholesale or retail channels seem to be better at facilitating access to credit and which channel therefore is better at bolstering community development. In addition, in the years before the financial crisis of 2008 when high cost and abusive mortgage lending surged, third-party mortgage brokers contributed to risky lending practices.[71] The application recipient data point is therefore necessary to track developments in the retail and wholesale channels, detect early warning signs, and devise any remedial programmatic or policy changes.

If a small borrower successfully negotiates the application channel and submits a loan application only to be denied, the denial reasons data point would have provided important information about why the denial occurred. A lender could have indicated up to four reasons for denial.^[72] The denial reasons include credit characteristics of the business, credit characteristics of the principal owners, use of credit proceeds (some lenders do not finance certain uses of loans), insufficient collateral, government criteria (applies to loans that would have received government guarantees), cashflow, time in business, aggregate exposure (too much debt), unverifiable information or other.^[73]

The denial reasons data point would have promoted the statutory purposes of community development and fair lending. Analyzing the reasons for denial is important for increasing access to credit for traditionally underserved businesses. For example, if lack of collateral was a highly cited reason for denying businesses owned by people of color, then efforts to increase the collateral of these businesses would be an effective way to bolster their access to credit.

The CFPB cites operational complexity and reputational risk for reasons to delete the data points related to the application process. For example, the agency says that application method is complicated because as an application process unfolds, a small business could experience contact via more than one medium.[74] Yet, reasonable guidance could be developed to alleviate uncertainty such as what method was the predominant method, or which method was the initial contact.

Also, the CFPB, upon reconsideration, believes that denial reasons could be “harmful or sensitive for applicants.”[75] This supposes that the public would be able to identify individual applicants and their denial reasons. In the half century of HMDA’s existence, NCRC is not aware of a single applicant that has been identified and shamed through HMDA data (Also, the CFPB has not provided any example of identification in HMDA data).

The validity of the CFPB’s reasoning is further diminished when considering that lenders must provide small businesses with adverse notices specifying reasons for denial of their loan applications.[76] Again, NCRC is not aware of any public shaming as a result of identification of any applicant via the adverse notices. Important fair lending and community development reasons for variables related to the application process are trumped by suspect rationale concerning complexity and sensitivity.

Removal of Pricing Information Would Perpetuate a Lack of Transparency Regarding Affordability and Potential Discrimination

The CFPB intends to delete all pricing information in its proposed rule, but the removal of pricing information makes the Section 1071 data quite limited in its usefulness to achieving the community development and fair lending purposes of the Dodd-Frank Act. The 2023 rule required the collection and dissemination of broker fees; the total amount of all non-interest charges that are scheduled to be imposed over the first annual period; for a merchant cash advance or other sales-based financing transaction, the difference between the amount advanced and the amount to be repaid; and information about any applicable prepayment penalties.[77]

Without pricing data, the Section 1071 data would inform the public if traditionally underserved businesses were able to access loans but would be unable to identify whether the loans were affordable, sustainable, and non-discriminatory. As the CFPB was gathering information for the 2023 rule, lending institutions as well as community groups opined that information on interest rates and fees was critical to determining the ability of small businesses to afford their loans and use their loans to grow and prosper.[78]

The experience with HMDA data illustrates the importance of pricing data for Section 1071 data. Over the last several years, the HMDA pricing data have been critical to identifying disparate pricing among protected classes, including higher prices charged to same-sex borrowers and disaggregated categories of Asian borrowers and Latino borrowers.^[79] While some pricing data were required by the HMDA amendments in Dodd-Frank, other data points were added under the Bureau’s discretionary authority. 

The challenges that women-owned businesses and businesses owned by people of color have when seeking to access credit are well-documented, and regular analyses of pricing disparities in the small business market must be included in efforts to address these challenges. The CFPB cited research conducted by the Federal Reserve and others that documented higher priced loans for minority-owned firms than white-owned ones for several products including credit cards and those offered by online lenders.^[80]

As the CFPB stated during the 2023 rulemaking, “Heightened risks to fair lending and small business development may arise from different pricing for the same products and the selective marketing of higher-priced or even predatory and unsustainable products. Because price-setting is integral to the functioning of any market, any analysis of the small business lending market—including to enforce fair lending laws or identify community and business development opportunities— would be less meaningful without this information.”^[81]

The CFPB continued by stating that without pricing data, the Section 1071 data, “might show that a particular market segment is expanding and apparently filling an important need, but this could actually be an area with predatory conduct. Pricing information would allow the Bureau and others to understand the situation more accurately. Data collection without pricing information could have the unintended consequence of incentivizing irresponsible lending, as providers seeking to increase representation of underserved groups could be encouraged to adopt high-cost models of lending.”^[82]

In its current proposed rule, the CFPB undercuts its previous rational by raising concerns that the pricing data might “reveal confidential business information or lead to incorrect inferences about discrimination.”[83] The experience with HMDA pricing data does not indicate any concerns about lenders’ confidential strategies being revealed or any blunting of innovation. Moreover, users of HMDA data understand that it cannot prove discrimination definitively because of missing variables.[84] Yet, it increases efficiency by pointing stakeholders, the CFPB, and lenders to potential fair lending trouble spots, enabling resolutions before enforcement or litigation is needed to resolve any discriminatory practices. In contrast, false allegations of discrimination have not been rampant or needlessly interfering with institutions’ home lending activity.

Even the proposed rule acknowledges that removing pricing information could introduce inefficiencies for lenders and the CFPB. It states that “Pricing is one dimension by which a lender could potentially discriminate against a credit applicant. Removing this information could reduce the efficiency of fair lending examinations or transparency that would have resulted from its inclusion, relative to the baseline.”[85]

Nonetheless, in its proposed rule the CFPB states that “In adopting the pricing data point, the Bureau assumed (in 2023) that community groups would use data responsibly but did not address how other members of the public with access to the data might use it.”[86] While NCRC appreciates the indirect acknowledgement in the proposed rule that community groups have been responsible users of the data, concerns over the general public’s hypothetical and sensational use of the data should not outweigh the fair lending and community development imperatives of the pricing data. There is little or no evidence that NCRC is aware of regarding unfair smearing of lenders’ reputation by the general public.

In the years of the subprime lending surge, the agencies and the public were kept in the dark about the widespread rise of high cost and abusive lending due to the absence of pricing and loan term and conditions in the home loan data. On the surface, the increase in lending in communities of color appeared to be beneficial but once delinquencies and defaults surged, interventions were too little and too late. Publicly available data would have served as an early warning system and could have helped curb the multi-year rise in abusive lending by spurring earlier interventions.

Remarking on the importance of data in stirring action and collaboration to curb abusive lending, the late former Federal Reserve Governor Ed Gramlich remarked:

Lenders can gain an increased awareness of the lending and pricing practices of their organizations, and of their competitors, through analysis of HMDA data. As a result, lenders may take opportunities to compete in areas where the data show concentrations of high-priced lending. Competitive pressures in such markets should increase efficiency in pricing, ensuring that prices for mortgages are commensurate with risk and do not just reflect an absence of competition.

New and strengthened collaborations should emerge among lenders, community groups, realtors, and other participants in the real estate market. As lenders seek to enter new markets, or to better serve their existing markets, the groups that have an intimate knowledge of their communities will be viewed as valuable partners.

One of the most effective ways of ensuring competition is to make data available.[87]

Recognizing the drawbacks of its proposed rule, the agency stated “The Bureau also expected that creditors would use the data to more effectively understand small business credit market conditions and that communities would use the data to identify gaps in credit access for small business owners. In each of these cases, the Bureau expects that creditors, communities, and governmental entities would experience costs in the form of a reduction in these benefits relative to the baseline.”[88]

Although the agency sees the validity of Governor Gramlich’s points regarding economic efficiency and data, it proceeded with its drastic proposal to eliminate all the pricing data from the 2023 rule. The absence of robust home loan data on pricing and loan terms and conditions contributed to the worst global recession since the Great Depression. The pricing data would likewise be vital in small business lending which is experiencing worrisome trends among MCA and other types of un- and under-regulated credit.

CFPB Cost-Benefit Analysis Underestimates Value of the Section 1071 Data

Throughout its proposed rule, the CFPB asserts that complexity, uncertainty, and reputational risk drive the high costs of the Section 1071 data points. Too many data points also would cause data quality to deteriorate according to the CFPB. Yet, the agency’s quantification of costs remains unconvincing. Using the same methodology for estimating costs as it did when promulgating the 2023 rule, the agency estimates modest costs just as it did in 2023.

The CFPB estimates that savings per individual lender under its proposed rule would range from a modest $751 to $14,000 annually. The savings vary due to an institution’s loan volume and whether the proposed rule would exempt it from data reporting. Summing these savings across the depository and non-depository institutions that would still be required to report yields savings of about $20 million for depository institutions and $400,000 for non-depository institutions annually. Again, summing the dollars saved across hundreds of institutions obscures the relatively modest savings ranging from several hundred to a few thousand dollars for each lender.

For the lenders that would no longer report data, the annual savings would be upwards of $100 million and $44 million for depository and non-depository institutions, respectively.[89] Considering that these savings are spread over more than 2,000 lenders, the per institution saving is modest.[90]

The insignificant to modest savings for lenders is reinforced when considering costs per loan. For lenders that would be exempt from reporting under the proposed rule, the savings per loan application would range from $8 to $34. For those that would still be reporting the reduced data, the savings per loan application would range from $1 to $7.[91] The per loan application savings is minimal regardless of whether a lender is exempt or still required to report data.

In contrast, the benefits of the 2023 rule requiring robust data disclosure could include motivating institutions to significantly increase their lending levels. In addition, the pricing data could motivate institutions to enhance the affordability of their loans. The total benefits could be thousands or tens of thousands more affordable loans. CRA small business lending in low-income tracts totaled 371,000 loans in 2024.[92] A one percent increase to women-owned or minority-owned small businesses in these tracts due to enhanced data disclosure would be about 3,700 loans while a 10 percent increase would be 37,000 loans.

This is likely a conservative estimate considering that home lending surged to traditionally underserved populations after Congress in 1989 required demographic borrower information in the HMDA data.[93] NCRC analyzed the years of lending shortly after the new HMDA data became available, obtaining the data from the Federal Financial Institutions Examination Council. We found that conventional home purchase lending to African Americans surged from 81,322 loans to 138,034 for an increase of 41.4 percent from 1993 to 1995. Similarly, as the table below shows, the percentage surge for Hispanics and low- and moderate-income borrowers (incomes up to 80 percent of area median income) was about 30 and 18 percent respectively.

Another way to consider the benefits of Section 1071 data is assessing the value of CRA data and exams. Economists from the Federal Reserve System have shown the positive impact of CRA through data analysis. Their econometric analysis demonstrated that, holding all other factors constant, when a census tract is no longer eligible for CRA consideration because it is no longer a low- and moderate-income (LMI) tract, small business lending declined by about 6 percent. By increasing public accountability, CRA motivates banks to work harder to find creditworthy borrowers.[94] CRA is implemented by federal agency exams and by publicly available data that is used on the exams. If the current CRA data became unavailable for some reason, it is likely that lending would decrease in LMI tracts. In contrast, if the publicly available data included robust information on the gender and race of the small business owners, it would likely be even more effective in increasing access to traditionally underserved small businesses.

Put another way, dramatically reducing the 2023 rule would be costly in terms of potentially reducing lending by significant amounts, particularly to small businesses with revenues between $1 to $5 million, whose lending experiences would no longer be revealed in the data. These businesses confront barriers to lending just like their smaller peers as discussed above.

Ruling in a lawsuit filed by the Texas Bankers Association in United States District Court for the Southern District of Texas, Judge Randy Crane ruled in favor of the CFPB’s 2023 rule accepting the agency’s analysis concerning costs and benefits. The judge declared that the “administrative record is voluminous, and its breakdown of the Bureau’s decision-making is comprehensive; moreover, the agency has reasonably assessed the effects of the Final Rule, including its anticipated costs versus benefits. The Court therefore “do[es] not find the agency’s action outside the realm of reasonableness.”[95] Judge Crane agreed with the CFPB’s 2023 analysis that costs are modest.

The CFPB’s qualitative analysis of its proposed rule also is unconvincing concerning savings for lenders and benefits for the public. In the opening pages of its proposed rule, the CFPB explains that HMDA data was expanded gradually and incrementally over its fifty year history, starting with mainly census tract location information in 1970s, adding demographic borrower information in 1990, and introducing additional variables concerning loan terms and conditions in the early to mid-2000s.[96] According to the CFPB, HMDA data quality was maintained by this slow and incremental approach. In contrast, the CFPB asserts that a Section 1071 database with several variables at the inception would experience serious data quality issues as lenders would have difficulties understanding how to disclose the data.

The analogy is not applicable. HMDA data disclosure commenced in the 1970s and 1980s when computer technology was less sophisticated than today and costs were likely higher for lenders. In contrast, lenders now have had experience with enhanced demographic and loan term and conditions data points in HMDA data for several years.[97] This experience would have facilitated lender compliance and understanding of the Section 1071 data points in the 2023 rule. Finally, the CFPB had developed the variables in the 2023 rule over several years starting in 2017 with input from lending institutions and the general public.[98] Lending institutions therefore had ample opportunity to influence the development of the data points and anticipate them.

If the CFPB was concerned about data quality, it would not propose deleting regulatory requirements and guidance concerning anti-discouragement. Anti-discouragement refers to lenders discouraging applicants from providing demographic information on their applications. The 2023 rule provided reasonable and flexible advice against discouragement to lenders regarding when they could request demographic data and other information required under Section 1071. The CFPB advised in the 2023 rule that requesting demographic information earlier in the application process is usually more effective in generating a response.^[99] The CFPB would have monitored for low response rates to requests for demographic information including comparing lending institutions to their peers based on their size or geographical area.^[100]

Now the CFPB proposes to delete most of the provisions against anti-discouragement, claiming that they were inefficient and complex.[101] The proposed rule appears to accept the views of an industry trade association at face value, and does not further scrutinize this claim with its own analysis including evaluating lender experience complying with similar provisions when collecting HMDA data. A cost-benefit analysis that affirms assertions of a few stakeholders without further investigation is not a vigorous and reasonable analysis.

Another serious flaw in the CFPB’s qualitative analysis of costs is that the 2023 CFPB rule and the 2023 regulatory update to CRA stipulated that the Section 1071 data would replace the CRA data on small businesses and farms. Under this change, banks and thrifts would report only the Section 1071 data and would dispense with reporting the CRA data. The Section 1071 data would have been used for CRA exams. This would have saved considerable resources.

In contrast, under this proposed rule and the proposed rescission of the 2023 CRA update, several banks and thrifts will be required to report two databases, CRA and Section 1071, on small business and farm data annually. This duplicative data reporting requirement would likely eliminate the savings the CFPB had calculated for banks and thrifts that would remain Section 1071 reporters under its proposed rule. Moreover, the savings for banks and thrifts no longer required to report Section 1071 data would be significantly diminished since many of them would continue to be CRA reporters.

In addition, all banks and thrifts will continue reporting Consolidated Report of Condition and Income data, informally known as Call Report data. They will still be tabulating data on the number and dollar amount of their business lending activity. Some of them may also continue to maintain internal data on the characteristics of their small business loan applicants so they can assess how well they are faring in the small business market and to document successes in small business lending to their CRA examiners. Overall, the savings projected by the CFPB for its proposed rule are significantly overestimated considering the continuing small business data collection and reporting activity even if the CFPB implements its rule as proposed.  

Conclusion – CFPB Must Withdraw its Proposed Rule

The CFPB must withdraw its proposed rule. The proposal will thwart the statutory purposes of Section 1071 to facilitate fair lending and community development. By exempting thousands of lending institutions from data reporting requirements, deleting critical data points on the characteristics of small businesses, and removing loan term and condition data, the agency renders the database unable to achieve the statutory purposes of Section 1071 since the data will under-report lending in large swaths of the country including metropolitan areas and rural counties. Additionally, the data will be unable to reveal whether the lending is affordable and sustainable or, alternatively, abusive and potentially discriminatory.

Significantly, deletion of reporting data for lending to businesses with revenues between $1 million to $5 million dramatically reduces the size of the database by up to one half which, in turn, significantly reduces its explanatory power. The experiences of businesses with revenues under $5 million in the lending marketplace might be modestly more favorable than their smaller peers but those experiences would not be too dissimilar and would provide insights into how to increase lending and remove barriers for all small businesses.

When signing the Dodd-Frank Act, President Obama remarked:

It will finally bring transparency to the kinds of complex and risky transactions that helped trigger the financial crisis. We all win when shareholders have more power and more information…We all win when consumers are protected against abuse.[102]

The proposed rule would counter this mission by reducing data requirements and hence transparency. It will limit the information that stakeholders will have to hold institutions accountable and better protect themselves. 

The CFPB’s cost-benefit analysis supporting its proposal is unconvincing. It reveals modest benefits in terms of reducing costs for lenders, while at the same time ignoring admitted costs such as reductions in the data’s ability to achieve statutory fair lending and community development objectives. The CFPB’s assertion that gradual and incremental enhancements to Section 1071 data would best preserve data quality is also not convincing considering the years of preparation lenders and the agency undertook to develop the 2023 rule and the similarities with the enhanced HMDA data that lenders have been reporting for several years.

The agency appears to be overly concerned about reputational harm to lenders associated with the data required under the 2023 rule. Yet, the HMDA data has not been used by the public to unfairly castigate lenders. Instead, it has been used, when necessary, to help identify lenders engaged in abusive and discriminatory practices. Moreover, a tremendously beneficial use of the data would be to identify win-win opportunities – overlooked and profitable opportunities for lenders and wealth building for communities. The capacity of the Section 1071 data to identify these win-win opportunities will be greatly diminished by the proposed rule.

The public should not have to wait until 2028 for the data to be collected considering that Congress passed and President Obama signed the Dodd Frank Act containing Section 1071 in 2010. Congress did not anticipate that the urgent imperatives of fair lending and community development facilitated by the data would have to wait almost two decades. The public should also not have to wait several more months or years for the CFPB to decide whether the public should have access to the data, even though full dissemination of the data to the public will be most useful and meaningful. The data works best when it sheds sunshine on the wide variety of lending needs and opportunities, not when it is so limited that it cloaks lending in a veil of secrecy which enables hidden practices like redlining or predatory lending.

In the days preceding Congressional passage of the Dodd-Frank Act, former Representative James R. Langevin remarked in the Congressional Record that the CFPB was created “to protect families and small businesses by ensuring that bank loans, mortgages, credit cards and other financial products are fair, affordable and transparent.”[103] This rule takes a significant step backwards from fairness and transparency.

Thank you for considering our views on this serious matter. If you have any questions, please feel free to contact me on jvantol@ncrc.org or Josh Silver, Senior Fellow, on endredline77@gmail.com.

Sincerely,

Jesse Van Tol
President and CEO

Member Sign-Ons:

African American Alliance of CDFI CEOs

Association for Neighborhood and Housing Development

Carolina Farm Stewardship Association

CEIBA

Dearfield Fund for Black Wealth

Fair Housing Center of Central Indiana

Fair Housing Center for Rights & Research

Fair Housing Center of West Michigan

Freedom Equity Inc.

HEAL (Health, Environment, Agriculture, Labor) Food Alliance

Homes on the Hill, CDC

Housing Opportunities Made Equal of Greater Cincinnati

Housing Oregon

ICON CDC

New Jersey Citizen Action

NHS Brooklyn CDC, Inc

Opportunity Resource Fund

Philadelphia Association of Community Development Associations

Prosperity Indiana

R.A.A. – Ready, Aim, Advocate

REBOUND, Inc

Rockland Housing Action Coalition

San Francisco Housing Development Corporation

Scale Link

South Dallas Fair Park Inner-city Community Development Corporation

Southwest Economic Solutions DBA MiSide Wealth

St. Louis Equal Housing and Community Reinvestment Alliance

Tribal Homeownership Coalition of the Southwest

United South Broadway Corporation

Urban Economic Development Association of Wisconsin (UEDA)

Urban Land Conservancy