It is widely acknowledged that on account of the growth of online and digital banking, consumers are less locally limited when shopping for banking services and less dependent on physical branches for their banking activities. The digitalization of banking has been accompanied by a net 10 percent decline in the number of bank branches nationwide since 2012.
Under the Community Reinvestment Act and its implementing regulations, covered depository institutions have been evaluated on their performance in meeting the banking and credit needs of the communities where they operate, in particular by adequately serving lower-income neighborhoods. For full-service retail banks, evaluations of CRA performance have focused on defined localities (such as counties and metropolitan areas) within which the bank operates physical branches, referred to as CRA assessment areas.
Due to the growth in online banking and the declining role of physical branches, some have argued that this branch-centric approach to CRA regulation has become obsolete. They contend that banks’ assessment areas should be broadened to include areas where they may have substantial online or mobile presence.[1] In 2023, the federal banking regulatory agencies adopted a new CRA rule that reflects this more expansive view of CRA assessment areas.[2]
This reasoning may have intuitive appeal. However, it does not hold up under scrutiny, as this note will demonstrate.
The argument for moving beyond branch-based assessment areas has two main components. First, because of branch closings, branch-based assessment areas are shrinking and are therefore less effective for ensuring that banks are meeting community needs. Second, because of online expansion, an increasing share of banks’ retail lending has moved outside of their traditional assessment areas. Therefore, banks’ CRA obligations should not be limited to where they have branches.
Neither of these assertions is consistent with the empirical facts described in this note. Assessment areas are not shrinking, and evidence does not support the view that branch closings have made the CRA less effective. Nor is there substantial evidence that bank lending has increasingly moved outside of their traditional assessment areas. Moreover, any kind of revised rule that pushes banks to focus more on the composition of their lending outside of their traditional assessment areas could prove counterproductive.
Banks’ Geographic Assessment Areas Have Not Been Contracting
Contrary to assertions that have been made about branches becoming obsolete and assessment areas shrinking due to branches being shuttered, reducing the effective geographic coverage of the CRA, branches are not at risk of becoming obsolete and assessment areas are not shrinking.
Predictions that bank branches have become obsolete and that branch closures will continue apace are not well founded. FDIC data show that the number of bank branches in the U.S. has stabilized with a smaller drop between 2021 and 2022 than any other one-year decline since 2015, and even a slight increase between 2022 and 2023.[3]
Banks’ incentives to provide branches remain strong for multiple reasons. Bank lending to small firms tends to be locally oriented because of informational efficiencies associated with a physical presence, as elaborated on later in this note. Branches are an effective way for banks to build trust and name recognition and strengthen local ties.
A 2018 research note from the Federal Reserve Board presents “evidence that the persistence of the large number of local bank branches across the country may be due to the fact that both depositors and small businesses continue to value local bank branches.”[4] A 2019 study from The Deloitte Center for Financial Services and a 2023 report from Forbes present similar findings based on international survey data.[5]
Moreover, there is no evidence to indicate that banks’ geographic assessment areas have been contracting via banks closing branches and exiting. If that were the case, then that would leave fewer lenders in local markets, leading to increases in market concentration measures.[6] However, as shown in Charts 1 and 2 below, using data from the Federal Financial Institutions Examination Council (FFIEC) on bank lending to small businesses (the FFIEC’s CRA data), concentration in local, small business lending has not been on the rise (if anything, it has been declining).
Chart 1 shows the annual (2004 through 2022) median Herfindahl-Hirschman index of market concentration across counties belonging to metropolitan or micropolitan statistical areas, measured alternatively with respect to number of loans and loan amounts. Chart 2 shows the analogous series for the mean HHI. Concentration has followed a declining trend since 2010 after rising because of a market contraction driven by the Great Recession.
As of 2020, concentration in terms of number of loans is about where it was in 2004, while in terms of dollar volume, it is substantially below where it was in 2004. Moreover, the observed levels of concentration indicate that local markets for small business lending in local banking markets are highly competitive.
Chart 1
Chart 2
CRA Effectiveness Has Not Been Diminished by Branch Closings
The decline in number of bank branches since 2012 has not been accompanied by a reduced focus on the needs of low- and moderate-income communities. In other words, there is little empirical support for the notion that the reduction in branches has been eroding the effectiveness of the CRA.
Data from the FFFIEC on bank lending to small businesses indicate virtually no change from year to year in the share originated in low- and moderate-income areas.[7] Home Mortgage Disclosure Act (HMDA) data collected by federal regulatory agencies from both bank and nonbank lenders show that banks have not lagged behind in providing home purchase loans to low- and moderate-income neighborhoods over this period.
In particular, a study from the Penn Institute for Urban Research finds that, separating by loan type (conventional conforming-size and FHA), the share of each category’s total loans that were originated in CRA-qualifying Census tracts rose during 2011 through 2017, with banks and nonbanks exhibiting parallel trends.[8] Chart 3 extends this analysis using HMDA data from 2017 through 2023 (the most recent available).[9] As seen in the chart, community banks (those with less than $10 billion in assets) and larger banks continued to keep pace with nonbanks in lending to low- and moderate-income areas.[10] The overall upward trend reflects a shift of housing demand toward neighborhoods with more affordable home values.
It is also important to consider that banks typically apply stricter underwriting standards for home purchase lending compared to nonbanks.[11] Thus, it is not simply the case that banks replicate the lending practices of nonbanks. Rather, they have a tighter credit box and face tighter regulatory constraints on risk-taking. Within this broader context, they have kept pace with nonbanks in providing credit to low- and moderate-income areas, notwithstanding their net loss of branches.
Chart 3
Nor have branch closings led to systematic loss of convenient access to bank branches for households that rely on the services offered at branches. A recent report from the Federal Reserve Bank of Cleveland documents little change in bank branch accessibility over the past two decades.[12] A similar analysis from the Bank Policy Institute finds that consumers’ average distance to the nearest bank branch has remained virtually unchanged since 2013.[13]
The same study from the Bank Policy Institute finds that generally, branches are closed where they are no longer in demand, due to population decline or to increased reliance on mobile banking in place of in-person branch services. Once such factors are controlled for, there is no evidence that branch closures have disproportionately occurred in lower-income neighborhoods. Thus, in general, banks continue to operate a sufficient number of branches to meet their CRA obligations in the community.
Likewise, the decade-long decline in number of bank branches has had no observable effect on household financial inclusion. On the contrary, the percentage of the population that is unbanked has also declined steadily and to a great degree since 2011, and this trend has been consistent across all demographic segments of the population.[14]
About 3.8 percent of the U.S. population live in “banking deserts,” according to a recent study from the Federal Reserve Bank of Philadelphia, based on applying a certain threshold distance to the nearest bank branch.[15] However, the identified banking deserts are disproportionately suburban middle- or upper-income areas, suggesting that the existence of banking deserts does not necessarily mean there is a financial exclusion problem.
Populations vulnerable to financial exclusion primarily are those facing severe economic disadvantages or geographic isolation. For instance, a very high proportion of the Native American population resides in banking deserts.[16] As such, financial exclusion and its impact on economically disadvantaged segments of the population represent longstanding challenges that must be overcome primarily through better integration of these communities into the economic and social mainstream. Further research and innovation are needed to improve economic integration and, along with utilizing existing tools like the CRA, overcome financial exclusion.
Has Bank Lending Been Moving Away From Traditional Assessment Areas?
Some have argued that due to branch closings, an increasing share of banks’ retail lending has moved outside of their traditional assessment areas. Therefore, it is reasoned, banks’ CRA obligations should not be limited to where they have branches.[17]
There is no systematic evidence supporting such assertions. On the contrary, as shown in Chart 4, according to regulatory data on the geographic distribution of small business loan originations, the percentage of small business loans within banks’ assessment areas, measured by loan counts, has followed an upward trend since 2012.[18]
Although in-area share in terms of dollars loaned exhibits a modest decline, this reflects a drop in the average size of loans within banks’ assessment areas, rather than an overall decrease in lending, as shown in Chart 5. The declining average loan size can be attributed to banks curtailing their risk exposure in response to high loan losses during the Great Recession and to regulatory constraints on loan volume, including increased regulatory capital requirements.
Chart 4
Chart 5
Increased CRA Emphasis on Banks’ Out-Of-Area Lending Could Prove Counterproductive
The consensus from multiple research studies is that the bank lending to small firms tends to be locally oriented because of informational efficiencies associated with a physical presence, person-to-person contact and pre-existing deposit account or other relationships with local businesses.[19] Knowledge of local economic and business conditions and access to checking account and other information from pre-existing relationships helps support lending decisions and mitigate credit risk, and financial guidance to small businesses can be provided through person-to-person contact.[20]
Such specialized or private knowledge and data often are referred to as “soft information,” as distinct from “hard information” that is more widely communicable to potential lenders, such as audited financial statements and credit record data.[21] These informational advantages have persisted despite advances in information technology and the declining number of physical bank branches.[22]
Because of the informational advantages associated with localized, small business lending by banks, a bank’s “community” is properly tied to areas in which it has a physical presence. These advantages also provide banks with an incentive to retain a branch presence in the communities in which they operate, consistent with the data presented above demonstrating the persistence of banks’ geographic footprint.
In other words, a bank’s physical presence in a community is important for implementing an effective CRA program. Pushing banks to expand their CRA programs out-of-area may impair the efficiency of these programs, resulting in higher costs of small business lending and ultimately constraining the amount of credit banks can provide.
Moreover, extending CRA out-of-area risks shifting the emphasis of banks’ small business loan programs toward larger, more established businesses whose creditworthiness can be verified based on hard information. Although small business lending tends to be based on soft information that is locally limited, this need not always be the case. Banks may also lend to larger or more well-established small businesses for which information is widely available at relatively low cost, or by providing loans backed by real estate for which reasonably accurate valuations are readily obtainable.[23]
Similar arguments may apply to community development lending and to certain categories of bank mortgage lending activity, such as loans originating through specialized, affordable home loan programs. In these cases, pushing banks to extend loans out-of-area again could prove counterproductive.
Conclusion
For full-service retail banks, evaluations of CRA performance have long focused on defined localities (such as counties and metropolitan areas) within which the bank operates physical branches, referred to as CRA assessment areas. But because of the growth in online banking and the declining role of physical branches, some have argued that this branch-centric approach to CRA regulation has become obsolete, and a new CRA rule adopted in 2023 applies a more expansive view of assessment areas.
The argument for moving beyond branch-based assessment areas has two main components. First, because of branch closings, branch-based assessment areas are shrinking and are therefore less effective for ensuring that banks are meeting community needs. Second, because of online expansion, an increasing share of banks’ retail lending has moved outside of their traditional assessment areas.
However, consideration of empirical evidence fails to support these arguments. Assessment areas are not shrinking; nor is there evidence that branch closings have made the CRA less effective. Bank lending is not migrating out from banks’ traditional assessment areas.
Any revised rule that pushes banks to focus more on the composition of their lending outside of their traditional assessment areas could also prove short-sighted. Pushing banks to expand their CRA programs out-of-area may impair the efficiency of these programs and could compel banks to recalibrate their small business loan programs toward larger, more established businesses, or to simply pull back from lending outside of their assessment areas.
[1] See pages 341-342 of the Federal Register Notice of the new rule for a summary of public comments received in support of broadening banks’ CRA assessment areas.
[2] In March 2024, a federal judge granted a preliminary injunction on the rule taking effect, as requested by several banking and business trade groups engaged in litigation challenging the rule. The federal banking agencies have appealed the decision to enjoin the rule to the Fifth Circuit Court of Appeals.
[3] After peaking at about 83,000 in 2012 after decades of expansion, the number of bank branches declined to 70,000 in 2022, according to FDIC data.
[4] See Elliot Anenberg, Andrew C. Chang, Serafin Grundl, Kevin B. Moore, and Richard Windle (2018). “The Branch Puzzle: Why Are there Still Bank Branches?” FEDS Notes, Board of Governors of the Federal Reserve System (link).
[5] See Val Srinivas and Richa Wadhwani (2019). “Recognizing the Value of Branches in a Digital World”. Deloitte Insights, The Deloitte Center for Financial Services (link). See also Michael Abbott (2023). “It’s 2023. Do Banks Still Need Branches?” Forbes (link).
[6] Banks exiting assessment areas via branch closings would generally raise average market concentration, unless simultaneously and systematically they are growing their presence in markets where they don’t have branches. Later in this note, we rule out the latter possibility by showing that out-of-area share of banks’ lending has not increased. The only other way bank exit could reduce concentration is if the exiting bank was a dominant lender in terms of market share. However, there is no indication that banks have been exiting markets where they are market-share dominant, and it seems counterintuitive that they would do so.
[7] See FFIEC (2023). “Findings from Analysis of Nationwide Summary Statistics for 2022 Community Reinvestment Act Data Fact Sheet” (link).
[8] See Paul Calem, Lauren Lambie-Hanson, and Susan Wachter (2019). “Is the CRA Still Relevant to Mortgage Lending?” Working Paper, Penn IUR Working Paper (link).
[9] Per the Penn IUR analysis, the chart shows the proportion of loans for the purchase of 1-4 unit, owner-occupied or investment properties that are originated in CRA-qualifying Census tracts, by origination year and lender type (first lien only). CRA qualifying tracts are those indicated to be low- or moderate income according to the HMDA classification and those identified by the FFIEC as CRA-eligible distressed or underserved rural areas. (link).
[10] Here, the distinction between community banks and larger banks is based on the threshold for being subject to the supervisory authority of the Consumer Financial Protection Bureau, which covers mortgage lending institutions with assets over $10 billion as of the HMDA submission date. This diverges from the size categories applied in the Penn IUR study, which are based on the contemporaneous asset size of the lending institution and also distinguishes between large banks (over $50 billion in assets) and regional banks ($10 billion to $50 billion).
[11] The average credit score is higher and average debt payment-to-income ratio is lower for bank-originated mortgages compared to those from nonbanks, in each of the two mortgage type categories. See The Urban Institute, “Housing Finance at a Glance” (November 2024), pages 18-19 (link).
[12] The report assesses accessibility on three dimensions: number of branches per banking institution, number of branches per 10,000 people, and average distance to the nearest branch as measured from Census tract centroid locations. See Kyle D. Fee and Erik Tiersten-Nyman (2021). “Has Bank Consolidation Changed People’s Access to a Full-Service Bank Branch?” Community Development Report, Federal Reserve Bank of Cleveland (link).
[13] See Paul S. Calem and Yasmeen Abdul-Razeq (2022). “What Drives Household Financial Inclusion? Analysis of Data Exposes Myths and Identifies Opportunities.” Research Note, Bank Policy Institute (link).
[14] Every other year, beginning in 2009, the FDIC Survey of Household Use of Banking and Financial Services (link) estimates the percentage of U.S. households that are unbanked (lack a checking or savings account at a bank or credit union). Based on these data, the percentage of unbanked households in the U.S. increased from 7.6 to 8.2 percent between 2009 and 2011, likely because of financial setbacks from the recession that ended in 2009. It fell to 7.7 percent in 2013 and then continued to decline steadily, with the most recent, 2023 FDIC survey (link) reporting the national unbanked rate at an historically low, 4.3 percent.
[15] See Alaina Barca and Harry Hou (2024), “U.S. Bank Branch Closures and Banking Deserts”, Report, Federal Reserve Bank of Philadelphia (link).
[16] See the reference cited in footnote 10.
[17] For example, 20 state attorneys general filed an amici brief in support of the banking agencies in the ongoing litigation, in which they argue that the decline in number of bank branches necessitates the change to geographic assessment areas stipulated in the new CRA rule.
[18] This study also finds that, when considering only larger loans (at least $100,000), the share originated out-of-area has been increasing, which contrasts with the trend observed above for overall lending. The study explains this finding by positing that progress in credit scoring technology has facilitated higher dollar value lending to well-established and more transparently creditworthy businesses on a wider geographic basis.
[19] See, for example, Sumit Agarwal and Robert Hauswald (2010). “Distance and Private Information in Lending.” Review of Financial Studies 23: 2757–88.
[20] For example, researchers from the Federal Reserve Bank of Philadelphia demonstrated a role of transaction account information in mitigating the credit risk of small business borrowers. See Loretta Mester and Leonard Nakamura (2007). “Transactions Accounts and Loan Monitoring.” The Review of Financial Studies 20 (3): 529–556 (link).
[21] One of the earlier papers to introduce this distinction was Mitchell A. Petersen and Raghuram G. Rajan (1994). “The Benefits of Firm-Creditor Relationships: Evidence from Small-Business Data”. Journal of Finance 49:3–37.
[22] See Kenneth Brevoort, John Holmes, and John Wolken (2010). “Distance Still Matters: The Information Revolution in Small Business Lending and the Persistent Role of Location, 1993-2003.” Finance and Economics Discussion Series 2010-08, Board of Governors of the Federal Reserve System (link). See also Robert M. Adams, Kenneth P. Brevoort, and John C. Driscoll (2020). “Is Lending Distance Really Changing? Distance Dynamics and Loan Composition in Small Business Lending.” Finance and Economics Discussion Series 2021-011, Board of Governors of the Federal Reserve System (link).
[23] Alternatively, the emphasis may shift to card-based lending that involves issuing credit cards to owners who have deemed creditworthy on the basis of their personal credit records.
