BAILEY ALLEN, PhD Student, Rawls College of Business, Texas Tech University
MIKE MAULDIN, Director of the Excellence in Banking Program, Texas Tech University
DREW WINTERS, Pickering Chair in Finance, Texas Tech University

The Dodd-Frank Wall Street Reform and Consumer Protection Act (hereafter, Dodd-Frank Act) was created—according to the first paragraph of the act itself: “To promote the financial stability of the United States by improving accountability and transparency in the financial system, end ‘too big to fail,’ protect the American taxpayer by ending bailouts, protect consumers from abusive financial services practices and for other purposes.” However, it is not achieving its objectives. For example, Allen, Cyree, Whitledge and Winters (2018) show that it did not end “Too Big to Fail.” ^[1] Additionally, Cyree (2016) finds an increase in regulatory burden for small banks following the passage of the act. ^[2]

Recently, the Bureau of Consumer Financial Protection (CFPB) proposed rules to implement section 1071 of the Dodd-Frank Act, “proposing to require covered financial institutions to collect and report to the bureau data on applications for credit for small businesses, including those owned by women or minorities.” ^[3] With the track record of failure and compliance burden from the Dodd-Frank Act, we believe it is important to examine the potential impact on banks from the implementation of section 1071 while the CFPB is still in the comment phase for the rule.

Section 1071 of the Dodd-Frank Act amends the Equal Credit Opportunity Act by creating section 704B titled “Small Business Loan Data Collection.” Here is the purpose of the change as stated in Dodd-Frank:

“This section is to facilitate enforcement of fair-lending laws and enable communities, governmental entities and creditors to identify business and community development needs and opportunities of women-owned, minority-owned and small businesses.”

To facilitate enforcement there must be information gathered. The statement on information gathering reads (in part):

“In the case of any application to a financial institution for credit for women-owned, minority-owned or small business, the financial institution shall 1) inquire whether the business is a women-owned, minority-owned or small business, and 2) maintain a record of the responses to such inquiry, separate from the application and accompanying information. Additionally, any applicant for credit may refuse to provide this information.”

The information compiled and maintained shall be itemized to clearly and conspicuously disclose:

• Number of the application and date the application is received;
• The type and purpose of the loan;
• The amount of credit or credit limit applied for and the amount or limit approved;
• The type of action taken with respect to the credit request and the date of the decision;
• The census tract of principal business location;
• Gross annual revenue of the business from the previous fiscal year;
• The race, sex and ethnicity of the principal owners; and
• Any additional information that the CFPB determines will aid in fulfilling the purpose of this section.

Finally, with the data collected, what are the rules for the data?

• Once compiled, the data is submitted to the CFPB annually;
• The data must be retained for not less than three years from the time of compilation;
• The data shall be made available to any member of the public, upon request; and
• The data shall be made available annually by the CFPB.

As a point of clarification, the details in this section come from section 1071 of the Dodd-Frank Act. The CFPB has a proposal for implementing section 1071. Click here for details on the implementation.

Cyree (2016) suggests that the Dodd-Frank Act increased the regulatory burden for small banks. Here, we examine the potential of section 1071 to increase the regulatory burden for small banks. Currently, it is against the law for banks to collect the race and gender of the principal owner(s) applying for a business loan, so we cannot address race and gender. However, bank call reports provide information on small-business loans and we will use this data in our analysis.
Call report (FFIEC 041 and 051) Schedule RC-C provides data on bank loans and leases for banks. Part II of Schedule RC-C provides data on loans to small businesses and small farms. Specifically, it provides the number and amounts outstanding of small-business loans. Part II of the schedule starts: “Report the number and amount currently outstanding as of the report date of business loans with ‘original amounts’ of $1 million or less and farm loans with ‘original amounts’ of $500,000 or less.”

The first step in our analysis is to consider small-business loans as a percent of total loans and leases by state. Figure 1 is a national map of states’ percentage of small-business loans. There are two main features in the results from the map. First, the percentages vary across states. Second, banks in the center of the country tend to have a larger percentage of their loan portfolio dedicated to small-business loans. These are the states with more small banks, which suggest—consistent with Cyree (2016)—the potential for more burden in small banks from the implementation of section 1071.

Figure 1

To access the potential burden of section 1071 on small banks more directly, we examine data from Texas banks. We define a Texas bank as any bank that uses a Texas address on its June 30, 2021, call report. We use data from Texas banks without foreign locations. We report results in Table 1 for our sample and various measures of potential burden. ^[4]

Table 1: Texas Banks Loans and Employees

Panel A:  Total Assets and Total Loans

Group	Size range for group	Number	Average total assets	Average total loans
1	<= $100 million	60	$67.4 million	$26.7 million
2	$100 million < size <= $500 million	186	$248.7 million	$132.1 million
3	$500 million < size <= $1 billion	71	$696.9 million	$420.7 million
4	$1 billion < size <= $10 billion	73	$2.6 billion	$1.6 billion
5	$10 billion <	7	$24.8 billion	$13.3 billion

Panel B:  Employees

Group	Size range for group	Number	Total full-time employees	Average full-time employees
1	<= $100 million	60	835	13.9
2	$100 million < size <= $500 million	186	7,665	41.2
3	$500 million < size <= $1 billion	71	7,734	108.9
4	$1 billion < size <= $10 billion	73	28,839	395.1
5	$10 billion <	7	18,265	2,609.3

Panel C:  Small-business Loan Percentages

Group	Non-farm	Farm	Total
1	24 percent	14.9 percent	38.9 percent
2	23.3 percent	7.2 percent	30.5 percent
3	20.5 percent	2.9 percent	23.4 percent
4	15.4 percent	1.5% percent	16.9 percent
5	11.1 percent	0.6 percent	11.7 percent

Panel D:  Small-Business Loans per Employee

Group	Number of loans	Median number of  small-business loans per employee
1	11,256	11
2	89,079	10.9
3	310,222	8.8
4	212,332	6.9
5	122,985	5.7

 

Appendix: U.S. Banks Loans and Employees (Domestic Offices Only)

Panel A:  Total Assets and Total Loans

Group	Size range for group	Number	Average total assets	Average total loans
1	<= $100 million	850	$61.6 million	$32.3 million
2	$100 million < size <= $500 million	2,332	$251.1 million	$149.1 million
3	$500 million < size <= $1 billion	750	$698.4 million	$440.9 million
4	$1 billion < size <= $10 billion	799	$2.6 billion	$1.7 billion
5	$10 billion <	107	$26.5 billion	$16.7 billion

Panel B:  Employees

Group	Size range for group	Number	Total full-time employees	Average full-time employees
1	<= $100 million	850	10,752	12.6
2	$100 million < size <= $500 million	2,332	104,288	44.7
3	$500 million < size <= $1 billion	750	86,867	115.8
4	$1 billion < size <= $10 billion	799	270,815	338.9
5	$10 billion <	107	246,477	2,303.5

Panel C:  Small-business loan percentages

Group	Non-farm	Farm	Total
1	19.1 percent	17.5 percent	36.6 percent
2	22 percent	8.4 percent	30.3 percent
3	20.1 percent	3.7 percent	23.8 percent
4	15.7 percent	1.8 percent	17.5 percent
5	9.6 percent	0.4 percent	10 percent

Panel D:  Small-business Loans per Employee

Group	Number of loans	Median number of small-business loans per employee
1	150,916	12.8
2	1,253,294	11.5
3	1,701,204	8.7
4	2,553,736	7.3
5	3,730,134	5.7

 

The purpose of Panel A of Table 1 is a basic sample description with definitions for our size groupings following our premise that section 1071 is likely more of a burden to small banks. Panel B further describes the sample with the total and average number of full-time equivalent employees. Group 1 (smallest banks) of the sample has an average of about 14 employees, while group 5 (largest banks) has an average of about 2,600 employees. This suggests that any additional reporting required on banks will pose a greater burden on the smallest banks.

Section 1071 requires reporting on small-business loans. Panel C of Table 1 provides the percentage of a bank’s total loan portfolio held in small-business loans. We provide the percentage of non-farm and farm small-business loans. There are a number of interesting insights from Panel C. First, the percentage of non-farm small-business loans is about 24 percent for banks with $500 million or less in total assets. Then the percentage decreases by about 5 percent of the total loan portfolio as we move up bank sizes across the three largest groups. Second, small business farm loans also decline as bank size increases. Additionally, banks above $500 million in total assets are not significant players in farm lending. Finally, total small-business loans as a percentage of total loans decline as bank size increases. For banks with $100 million or less in total assets, small-business loans comprise about 40 percent of their total loan portfolio while for banks with more than $10 billion in total assets, small-business loans are only about 10 percent of their portfolios. The bottom line from Panel C is that smaller banks are the ones in the business of making small-business loans. ^[5]

Panel D is an attempt to proxy for the burden of section 1071 on small banks. Group 1 banks have 11,256 small-business loans, which is 11 small-business loans per employee. Group 3 banks have 310,222 small-business loans, which is about nine small-business loans per employee. Group 5 banks have 122,985 small-business loans, which is about six loans per employee. Thus, the average employee in the smallest banks will have to do two times the reporting for section 1071 than the average employee in the largest bank. The largest banks with an average of 2,600 employees can have an employee dedicated to section 1071 reporting while one of (on average) 14 full-time employees of the smallest banks will have section 1071 reporting added to their list of tasks. Additionally, small banks are unlikely to have the IT infrastructure needed to automate the data collection and the cost of acquiring technology for this purpose will be more of a burden on small banks.

 

Implementing any rule creates costs on banks and our analysis shows that the burden is likely to be biggest for the smallest banks. If we are going to increase the burden on small banks, we should clearly understand the public policy benefit of the rule.

The CFPB states: “…proposed rule would create the first comprehensive database of small-business credit applications in the United States. This would include critical information about women-owned and minority-owned small businesses to help regulators and the public identify and address fair lending concerns. The database would also enable a range of stakeholders to better identify business and community development needs and opportunities for small businesses, including women-owned and minority-owned small businesses. Just as the bureau works in other ways to help foster fairness and opportunity in consumer financial services markets for all consumers, the proposed 1071 rule is structured to realize these same goals for the small-business market—for all small businesses within the scope of the rule, including those that are owned by women and minorities.” ^[6]

This suggests that the CFPB wants the ability to analyze small-business lending discrimination in a manner similar to mortgage loan analysis on HMDA data. However, this is unnecessary and likely of little economic benefit.

Bank regulators already examine banks for compliance with the Community Reinvestment Act (CRA). The CRA encourages banks to make credit available to communities in which they do business, including low- and moderate-income neighborhoods. The regulators evaluate and rate the banks for their performance under the CRA. Cyree and Winters (2021) ^[7] show that more than 90 percent of banks are rated satisfactory for compliance with CRA for a sample from 2007 to 2016 with about 8 percent of the sample banks rated outstanding. These ratings suggest little, if any, lending discrimination into low- and moderate-income neighborhoods. However, if lending discrimination is suspected, the regulators have direct access to bank loan applications and files for analysis, which would eliminate the reporting proposed under section 1071.

HMDA requires reporting the race and gender of mortgage loan applicants. Cyree and Winters (2021) find that—after a reasonable sets of controls—banks rated outstanding for CRA compliance are less likely to grant a mortgage loan to racial minorities than to whites. This result suggests statistical discrimination. However, with about 240,000 applicants, the sample size makes small differences between groups statistically significant. Additionally, their model has very low explanatory power (adjusted r-square of about 6 percent), which suggests an omitted variable problem. Delis and Papadopoulos (2019) ^[8] explain that problem as the lack of crucial underlying factors and information on borrowers’ creditworthiness, including, but not limited to, credit score variables (e.g., FICO score), LTV ratio, DTI ratio and down-payment size. These criteria are privately observed by the lending institutions and play a crucial role when a bank assesses an applicant’s riskiness and decides whether or not to originate the loan. In other words, the banks have private information about loan applicants that are not in the HMDA data. The bottom line from the HMDA data is that statistical analysis of a large and incomplete dataset is likely to lead to statistically significant results that lack economic value.

The data requirements for reporting under section 1071 create an omitted variable problem similar to what occurs with the HMDA data. Specifically, the information in the application needed to do a careful credit analysis is stripped from the gender and race data. The proposed data only allow analysis if different groups have a different percentage of loans granted. The data will not allow for an analysis of creditworthiness. Accordingly, the proposed data cannot assist with the stated purpose of section 1071 (“identify business and community development needs and opportunities of women-owned, minority-owned and small businesses”), but will create an additional burden, especially for small banks, of increased overhead and nuisance lawsuits.

Our discussion on omitted variables in the proposed data collection under section 1071 should not be read as a need for more data collection under the proposed rule. Small-business financial reports and small-business lending documentation are not standardized, unlike mortgage lending, and the lack of standardization creates statistical issues similar to the omitted variable problem. Additionally, relationships in small-business lending are difficult to quantify for data collection. So, expanding the data collection will not fix the statistical issues for the analysis of the proposed data.

Additionally, the burden from section 1071 on community banks will be important for local economies. Though community banks tend to be relatively small, their small-business lending far exceeds their aggregate lending share. Nationally, community banks represent 15 percent of the industry’s total loans, but 30 percent of its CRE loans, 36 percent of small-business loans and 70 percent of agricultural loans. ^[9] Perhaps a more accurate and recent indicator of the importance of the community banking sector in funding small-business enterprises is their robust participation in the Paycheck Protection Program (PPP). Data from the Small Business Administration (SBA) on this critical lifeline shows the importance of smaller banks in supporting the borrowing needs of small businesses and, by extrapolation, would indicate that the reporting burden of section 1071 would fall disproportionately. The SBA recently reflected in their PPP approvals through May 31, 2021, that $799.83 billion were made in PPP loans. ^[10] Banks less than $10 billion in size accounted for $335.28 billion or 41.9 percent of PPP loans. ^[11] This is important when you consider that FDIC data indicates that banks $10 billion and under represent approximately 12 percent of the industry assets. ^[12] Regardless of how we examine community bank lending, it is abundantly clear that community banks play an outsized role in the small-business lending space and make a disproportionately large percentage of these loans in a variety of categories.

 

Our analysis suggests that the implementation of Section 1071 will be an unintended attack on relationship banking that occurs every day in every region of our country. It will create a barrier for credit for the truly small businesses that are less sophisticated, but essential to the community. Specifically, truly small businesses have limited financial data and thus rely on relationship lending from community banks. The limited data will not allow the community (small) banks to protect themselves against lawsuits using section 1071, nor conform to the data requirements of non-community bank (large) lenders.

Another unintended consequence of this one-size-fits-all approach—which is punitive to smaller community banks—to data collection will be the cost of compliance and risk of non-compliance to implement section 1071. A cost-benefit analysis will lead many community banks to cease making small-businesses loans. We find that Texas banks in our sample that are $10 billion and under in total assets have about $32 billion of small-business loans outstanding on their books. A disruption to the small-business loan process, especially as the state is still working through issues that were created by COVID-19, would be devastating to the overall Texas economy and to many underserved rural communities.

Chapter 4 of the 2020 FDIC Community Banking Study supports the findings of this research and further strengthens the argument against the implementation of Section 1071 of the Dodd-Frank Act. ^[13]

^[1] Kyle D. Allen, Ken B. Cyree, Matthew D. Whitledge and Drew B. Winters, 2018, “An Event Study Analysis of Too-Big-To-Fail After the Dodd-Frank Act: Who is Too Big to Fail?” Journal of Economics and Business, No. 98, pages 19–31.

^[2] Ken B. Cyree, 2016, “The Effects of Regulatory Compliance for Small Banks Around Crisis-Based Regulation,” Journal of Financial Research, No. 39, pages 215–245.

^[3] Consumer Financial Protection Bureau, “Small Business Lending Data Collection Under the Equal Credit Opportunity Act” (see summary section).

^[4] Our discussion focuses on Texas banks. For reference, we provide an appendix that replicates Table 1 for all U.S. banks (without foreign locations, for direct comparison to Table 1). Excluding foreign operations excludes the largest banks, but does not change our results on the burden of section 1071 on small banks.

^[5] Large banks lend larger dollar amounts in small-business loans, but these dollar amounts are a much smaller percentage of their loan portfolio.

^[6] Consumer Financial Protection Bureau, “Small Business Lending Data Collection Under the Equal Credit Opportunity Act” (see page 4).

^[7] Ken B. Cyree and Drew B. Winters, 2021, “Investigating Bank Lending Discrimination in the U.S. Using CRA-rated Banks’ HMDA Loan Data,” working paper.

^[8] Manthos D. Delis and Panagiotis Papadopoulos, 2019, “Mortgage Lending Discrimination Across the U.S.: New Methodology and New Evidence,” Journal of Financial Services Research, No. 56, pages 341–368.

^[9] Federal Deposit Insurance Corp., FDIC Community Banking Study 2020.

^[10] U.S. Small Business Administration, Paycheck Protection Program Report: Approvals Through May 31, 2021.

^[11] U.S. Small Business Administration, Paycheck Protection Program Weekly Reports 2021; Paycheck Protection Program: Approvals Through August 8, 2020.

^[12] Federal Deposit Insurance Corp., FDIC Community Banking Study 2020.

^[13]  Federal Deposit Insurance Corp., FDIC Community Banking Study 2020, Chapter 4.