Berger 2023 Let U.S. Put Our Moneys Together Minority Owned Banks And Resilience To Crises
Summary
Working Paper WP 23-13 published in June 2023 by the Federal Reserve Bank of Philadelphia, titled "Let Us Put Our Moneys Together": Minority-Owned Banks and Resilience to Crises, by Allen N. Berger, Maryann P. Feldman, W. Scott Langford and Raluca A. Roman. The study analyzes individual banks in local markets from 2006 to 2020 using propensity score matching and instrumental variables across 28,129 observations. It finds that minority-owned banks improved community economic resilience during the global financial crisis and COVID-19 through small business and household lending. It estimates that similar behavior across all U.S. banks during the global financial crisis would have maintained at least 1.9 million minority jobs and $50 billion in annual small-business credit.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Berger 2023 Let U.S. Put Our Moneys Together Minority Owned Banks And Resilience To Crises.”
Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/berger-2023-let-u-s-put-our-moneys-together-minority-owned-banks-and-r-e3dafab345/
, SHA-256 957dfedc93b1895c83f48cc6b078a6953ef07c0b757caa6dcff3f1f839036a99.Full text
Working Papers
RESEARCH DEPARTMENT
“Let Us Put Our Moneys
Together”
Minority-Owned Banks and
Resilience to Crises
WP 23-13
Allen N. Berger
University of South Carolina
Maryann P. Feldman PUBLISHED
Arizona State University June 2023
W. Scott Langford
Arizona State University
Raluca A. Roman
Federal Reserve Bank of Philadelphia
ISSN: 1962-5361
Disclaimer: This Philadelphia Fed working paper represents preliminary research that is being
circulated for discussion purposes. The views expressed in these papers are solely those of
the authors and do not necessarily reflect the views of the Federal Reserve Bank of Philadelphia
or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
Philadelphia Fed working papers are free to download at: https://philadelphiafed.org/research- DOI: https://doi.org/
and-data/publications/working-papers. 10.21799/frbp.wp.2023.13
“Let Us Put Our Moneys Together”:
Minority-Owned Banks and Resilience to Crises*
Allen N. Berger Maryann P. Feldman
University of South Carolina Arizona State University
W. Scott Langford Raluca A. Roman
Arizona State University Federal Reserve Bank of Philadelphia
June 2023
Abstract
Minority-owned banks have a mission to promote economic well-being in their communities. In
particular, specialization in lending based on a central mechanism of shared-minority identity can
yield an advantage in serving community needs through times of financial and economic crises.
To test this proposition, we analyze individual banks in their local market context from 2006 to
2020. Results suggest minority-owned banks improve economic resilience in their communities
during the global financial crisis (GFC) and the COVID-19 crisis through increased small business
and household lending, but fewer benefits are found during other phases of the business cycle. Our
results are robust and stand up to treatments of identification concerns, including propensity score
matching (PSM) and instrumental variables (IV). Our results imply that if all U.S. banks behaved
in a manner consistent with minority-owned banks through the GFC, at least 1.9 million more
minority jobs would have been maintained and at least $50 billion more in credit would have been
available to small businesses on an annual basis. These findings are consistent with predictions of
the economic resilience literature but not those of the finance-growth nexus literature.
Keywords: minority-owned banks, minority employment, community banking, crises, economic
resilience, small business credit, household credit, shared-minority identity
JEL Codes: G01, G21, O12, J15, J21
*
Berger: University of South Carolina; email: aberger@moore.sc.edu. Feldman: Arizona State University; email:
maryann.feldman@asu.edu. Langford: Arizona State University; email: scott.langford@asu.edu. Roman: Federal
Reserve Bank of Philadelphia; email: raluca.roman@phil.frb.org. The views expressed are those of the authors and
do not necessarily reflect those of the Federal Reserve Bank of Philadelphia or the Federal Reserve System. The
authors thank Mitchell Berlin for his insightful comments. The authors also gratefully acknowledge financial support
from the Kenan Institute of Private Enterprise at UNC-Chapel Hill.
“Let us put our moneys together… …let us put our money out at usury among ourselves and reap the benefit ourselves …”
— Maggie Lena Walker (1901)
“A resilient society will enjoy stronger growth over the long run because it will better absorb shocks. For that reason, a
resilient society is better equipped to take risks. And risk-taking is an essential driver of growth.”
— Markus Brunnermeier (2021), The Resilient Society
1. Introduction
Local economies fare differently during times of crisis based on their resiliency. Minority
individuals are particularly vulnerable through economic crises — minority employment was
more sensitive to economic conditions through both the global financial crisis (GFC) (e.g.,
Orrenius and Zavodny, 2010; Hoynes, Miller, and Schaller, 2012; Couch, Fairlie, and Xu,
2018) and the COVID-19 crisis (e.g., Brodeur, Gray, Islam, and Bhuiyan, 2021; Montenovo,
Jiang, Lozano-Rojas, Schmutte, Simon, Weinberg, and Wing, 2021). To provide
opportunities and address resiliency, minority communities founded their own financial
institutions. Minority-owned banks — depository institutions with majority stock owned by
individuals who identify themselves as minority (Black, Asian, Hispanic, or Native American)
— have a long, though fraught history in the U.S. (Ammons, 1996; Baradaran, 2017). The quote
by Maggie Lena Walker (1901), the first Black female bank president in the U.S., captures the
key motivation of minority-owned banks — to promote wealth, end the cycle of poverty,
and strengthen economic resilience in their communities. Markus Brunnermeier’s quote
highlights that economic resilience is key for absorbing the shocks from economic
downturns and other adverse events such as COVID-19. Here, we empirically test the
extent to which minority-owned banks strengthen economic resilience within and beyond
their own communities.
Within the finance literature, one known mechanism for alleviating economic distress is
relationship lending (Kysucky and Norden, 2016; Bolton, Freixas, Gambacorta, and Mistrulli,
2016; Berger, Bouwman, Norden, Roman, Udell, and Wang, 2022a, 2022b). In that context,
lenders use private information to assess borrower risk and to make lending decisions. In
particular, lenders use soft information, which reflects intangible borrower
characteristics (e.g., dependability) collected through long-term relationships.
In this paper, we propose a novel mechanism through which information is
transmitted, not yet considered in the literature: shared minority identity. Here, the lender
and borrower share an identity via a relevant characteristic (e.g., race, ethnicity),
which enables more effective information transmission. Through more effective verbal
and nonverbal communications, lenders are able to more efficiently acquire and process soft
information from borrowers who share their identity. Distinct from relationship lending, a
long-term relationship is not required. These mechanisms are not mutually exclusive; lenders may
use both relationship lending and shared minority identity. The shared minority identity may be
particularly important for individuals who may not always be best served by the banking industry
(e.g., racial and ethnic minorities).
Members of the same ethnic or minority group share a common native language, have
shared traditions and culture, and other difficult-to-measure-but-differentiating characteristics. We
acknowledge that the concept of shared minority identity is broad and may reflect many
characteristics (e.g., race, ethnicity, gender). The shared minority identity may only function
within minority groups, or it may function in a more diffused way between groups. We regard
being a minority of any race or ethnicity as potentially powerful. There are also few minority-
owned banks, raising statistical power problems. Thus, to construct the minority category, we
aggregate several racial and ethnic minorities into one classification. Here, we demonstrate how a
shared minority identity results in increased local resiliency.
Minority-owned banks, loan officers, and other bank employees can establish higher levels
of mutual trust and reduce transaction costs when dealing with borrowers in their own racial or
ethnic groups because of more effective communication (e.g., communicate in a borrower’s native
language or dialect) that would provide more nuanced evaluation of risk and reduce taste-based
unfavorable racial discrimination. Minority borrowers might also be less likely to strategically
default on loans from minority-owned banks to avoid loss of reputation or community capital (i.e.,
defaulting on peers in their own community). Here, we consider the extent to which minority-
owned banks impact economic growth and resilience in their communities and the mechanisms
through which they achieve this. Specifically, we examine this by considering how minority-
owned banks impact employment growth across the business cycle, as well as the underlying
mechanism, the credit supply.
Our results indicate that the presence of minority-owned banks helps mitigate employment
losses for minorities through both the GFC and the COVID-19 crisis. We find evidence consistent
with a shared-identity transmission of information, where minority-owned banks maintain small
business and residential real estate credit supplies through each crisis. Results stand in contrast to
the extant literature. Across a range of outcomes, Berger, Bouwman, Norden, Roman, Udell, and
Wang (2022a) find relationship lenders take advantage of firms through the COVID-19 crisis. We
2
also find that during times of crisis, minority-owned banks mitigate employment losses for local
nonminority individuals as well. In sum, we find minority-owned banks enhance the economic
strength of their communities through economic resilience.
This study builds on a series of literature streams to contribute to our understanding of how
bank-ownership characteristics impact local economies, and the role of shared identity in
facilitating the transfer of information. The finance-growth nexus literature established that more
robust financial systems lead to faster economic growth (King and Levine, 1993; Berger, Hasan,
and Klapper, 2004). This stream highlights the potential for minority-owned banks to strengthen
economic growth through a range of business cycle conditions. The economic resilience literature
focuses on the institutions or policies that mitigate the impact of financial and economic crises
(e.g., Martin, 2012), in particular local finance (Petach and Weiler, 2019; Levine, Lin, and Wensi,
2020; Petach, Weiler, and Conroy, 2021; Langford and Feldman, 2021). This stream highlights
the potential for minority-owned banks to strengthen economic resilience through economic and
financial crises. The role of minority-owned banks in economic development and resilience has
not been previously investigated.
In terms of mechanisms, the idea that minority-owned banks better serve their own
communities reflects the importance of efficient collection and processing of information (Stein
2002). This improves financing for small businesses and enables greater overall bank credit flows.
The final body of literature — small business finance — also speaks to the underlying mechanisms
by considering how the various sources of small business finance shift over both the firm and
business cycles (e.g., Berger and Udell, 1998, Berger and Roman, 2018). This body of literature
considers how types of finance firms’ access depends on informational opacity, or the capacity of
firms to signal their quality. Here, we expect minority-owned banks, through more effective
communication, to have a greater capacity to assess firm quality. In sum, our contribution is to
demonstrate that minority-owned banks enhance economic resilience, consistent with a novel
lending mechanism, shared minority identity.
We investigate the impact of minority-owned banks through three conditions: regular
business cycle times, the GFC, and the COVID-19 crisis. We construct a county-year level panel
data set (2006‒2020) and regress market outcomes on the presence of minority-owned banks,
controlling for other banking, socioeconomic, and demographic factors. We find that as the
county-level market share of minority-owned banks increases, overall, minority, and White
3
employment growth reductions are attenuated during the GFC (2007‒2009) and in most cases also
during the COVID-19 crisis (2020), though fewer positive effects are observed during other
portions of the business cycle.
We also construct a series of tests to address identification concerns and enhance our
confidence that our results are causal. To mitigate selection bias concerns, we use a propensity
score matching (PSM) technique to predict the probability of minority-owned bank presence and
limit the samples to those counties with common support. We also limit our samples to counties
that either contain or are adjacent to those that contain minority-owned banks. To mitigate reverse-
causality concerns, we use an instrumental variable (IV) technique. Here, we use minority ancestry
to predict minority-owned bank market share. Finally, to increase confidence in our results, we
analyze a series of alternative specifications. Our main results hold, increasing confidence in the
robustness of these results. Notably, our results have significant economic impacts. If all banks in
the U.S. behaved consistent with minority-owned banks through the GFC, at least 1.9 million
minority jobs would have been maintained and $50 billion in credit would have been available on
an annual basis to local firms.
In terms of expected mechanisms, broadly, we expect employees of minority-owned banks
to have more effective lines of communication with minority-owned firms and households. We
examine underlying mechanisms by testing whether minority-owned banks impact the local
economy through credit supply to small businesses and households. We estimate the effect of
minority-ownership status on small business and residential mortgage lending through each crisis
and the remainder of the business cycle. We find minority-owned banks reduce small business and
residential mortgage lending less through the GFC, and sometimes also during the COVID-19
crisis. These analyses indicate minority-owned banks mitigated the effects of the GFC through
maintaining small business and household lending. These findings are consistent with a shared
minority identity enabling minority-owned banks to have superior skills in collecting and
processing information, but it is not possible to directly test this with the available data. Through
noncrisis times, our findings are nuanced; small business origination growth is lower in minority-
owned banks, while residential mortgage lending rates are higher.
The remainder of this paper is organized as follows. Section 2 discusses prior research on
relations between local finance and economic resilience, as well as research on the GFC and the
COVID-19 crisis and outlines our empirical approach. Section 3 provides our main county-level
4
estimation results for overall, minority, and White employment, while Section 4 provides evidence
on bank-level small business lending as a channel for the main effects. Section 5 offers additional
small business and household lending results to further interrogate the channels. Section 6 provides
conclusions, policy implications, and recommendations for future research.
2. Minority-Owned Banks and Economic Resilience
2.1 – Local Finance and Economic Growth
The finance-growth nexus literature demonstrates that robust financial systems cause faster
economic growth across business and economic conditions (e.g., King and Levine, 1993).
Consistent with this point, as the market share of community banks increase, GDP growth
increases in both developed and developing nations via several mechanisms (e.g., small business
lending, overall credit flows), consistent with relationship lending (Berger, Hasan, and Klapper,
2004). To the authors’ knowledge, this is the first study to investigate the extent to which minority-
owned banks promote local economic growth overall and for minorities in their communities
across economic and financial environments.
2.2 – Local Finance and Economic Resilience
Martin and Sunley (2015) define economic resilience as the capacity of a geographic region to
withstand and recover from shocks to its growth trajectory. Economic resilience represents the
capacity to adapt (e.g., Christopherson, Michie, and Tyler, 2010; Clark, Huang, and Walsh, 2010;
Pike, Dawley, and Tomaney, 2010; Simmie and Martin, 2010). Our central argument is that
minority-owned banks strengthen the economic resilience of their communities by lessening the
impact of financial and economic crisis. Empirical studies demonstrate income inequality (Lewin,
Watson, and Brown, 2017; Rahe, Weber, Wu, and Fisher, 2019), state-level fiscal policies (Gjerde,
Prescott, and Rice, 2019), and local finance affect economic resilience (Petach and Weiler, 2019;
Levine, Lin, and Wensi, 2020; Petach, Weiler, and Conroy, 2021; Langford and Feldman, 2021).
The GFC and the current COVID-19 crisis provide an opportunity to assess the differential
effects of minority-owned banks on economic downturns. These downturns are particularly
important for minority workers; studies demonstrate minority employment is more sensitive to
economic conditions, both through the GFC (e.g., Orrenius and Zavodny, 2010; Hoynes, Miller,
and Schaller, 2012; Couch, Fairlie, and Xu, 2018) and the current COVID-19 crisis (e.g., Brodeur,
Gray, Islam and Bhuiyan, 2021; Montenovo, Jiang, Lozano-Rojas, Schmutte, Simon, Weinberg,
5
and Wing, 2021). Thus, to the extent that minority-owned banks achieve their mission by
promoting economic resilience, greater presence of minority-owned banks would be associated
with improved economic outcomes for minorities in their communities during economic and
financial crises, when such aid may be most urgently needed.
2.3 – Minority-Owned Banks and Discrimination
Defined as depository institutions with 51 percent or more of the voting stock owned by individuals
who identify themselves as minority (e.g., Black, Asian, Hispanic, Native American), minority-
owned banks typically serve a specific community. In particular, Black-owned banks have a long
history in the U.S.; Black-owned bank openings peaked between Reconstruction and the Great
Depression (Gerena, 2007). However, in recent years, non-Black, minority-owned banks have
gained prominence. Figure 1 provides the number of minority-owned banks in the U.S. over the
time period of interest (2006‒2020). Here, we see the number of minority-owned banks began to
increase after 2014, a trend driven by an increase in Asian and Hispanic-owned banks. Notably,
the number of Black-owned banks declined from 31 in 2006 to 18 in 2021 with $5.35 billion in
total assets (authors’ calculations). In contrast, Hispanic and Asian-owned banks increased in
number over the same time period (Hispanic, 2006: 19; 2020: 30; Asian, 2006: 39; 2020: 67). This
indicates minority-owned banks represent an increasingly strong force within the banking industry,
although the composition is changing.2
Figure 2 presents the geographic distribution of minority-owned banks across the
contiguous U.S. in 2020. We show the distribution using an indicator variable for the presence of
a bank branch of a minority-owned bank. Figure 3 presents the geographic distribution of the
minority population across the contiguous U.S. We show the distribution as of 2020, using the
proportion of the total population classified as a racial or ethnic minority. These figures indicate
minority-owned banks exist across the U.S., with pockets in population centers (e.g., California,
New York, Florida) and regions with high-minority populations (e.g., Texas, Oklahoma). The
largest concentration of minority-owned bank branches is along the Texas‒Mexico border, as well
as central and southern Florida. Notably, the Great Plains register no minority-owned banks.
2
Ideally, we would differentiate among types of minorities. However, because of the small number of minority-owned
banks, and the associated statistical analysis issues, our sample includes minority-owned banks as a whole.
6
Minority-owned banks differ from nonminority-owned banks in terms of behavior (Kwast
and Black, 1983), portfolio composition (Bates and Bradford, 1980), performance (Gardner, 1982;
Ziorklui, 1994; Lawrence, 1997), efficiency (Elyasiani and Mehdian, 1992; Henderson, 1999;
Iqbal, Ramaswamy, and Akhigbe, 1999; Barth and Xu, 2020), reaction to deregulation (Holdren
and Heyliger, 1993), lending practices (Black, Collins, and Cyree, 1997; Black, Robinson, and
Schweitzer, 2001) and bank failures (Kashian and Drago, 2016. We extend this stream of research
by examining the interaction between minority-owned banks and local economies.
We expect minority-owned banks to have a particularly strong impact on the credit supply
for minority-owned firms and households, an effect driven by shared minority identity and better
in-house information expertise to evaluate the credit quality of such borrowers. Drexler and Schoar
(2014) show that as relationship strength between loan officers and borrowers decreases, the credit
supply is reduced and default likelihood increases. Fisman, Paravisini, and Vig (2017) show lender
and borrower cultural proximity increases credit supply and reduces default using a loan officer
rotation policy, which introduces exogenous variation in religion and caste. These studies show
soft information is a key factor in credit access; this paper extends consideration to shared-identity
information and economic impacts.
Another mechanism is discrimination. Taste-based discrimination is a preference for a
specific group, independent of expected profit (Becker, 1957). In contrast, statistical
discrimination reflects profit maximization differences in loan outcomes. Although the mechanism
differs, the outcomes remain consistent, with minorities often being disadvantaged by majority-
owned lenders. There is evidence of discriminatory behavior in consumer credit markets toward
minorities (Beck, Behr, and Madestam, 2018; Begley and Purnanandam, 2021; Bhutta, Hizmo,
and Ringo, 2021; Giacoletti, Heimer, and Yu, 2021; Ambrose, Conklin, and Lopez, 2021; Bartlett,
Morse, Stanton, and Wallace, 2022). In particular, minorities face great difficulty in obtaining
credit (Begley and Purnanandam, 2021; Bhutta, Hizmo, and Ringo, 2021; Giacoletti, Heimer, and
Yu, 2021; Ambrose, Conklin, and Lopez, 2021). For example, Munnell, Tootell, Browne, and
McEneaney(1996) demonstrate that, conditional on personal and property characteristics, White
applicants have lower rejection rates than minorities. Similarly, Bartlett, Morse, Stanton, and
Wallace (2022) provide evidence that approximately 1 million minority applicants were rejected
between 2009 and 2015 due to discrimination. Chu, Ma, and Zhang (2022) demonstrate that when
7
banks transition from private to public ownership, mortgage denial rates for Blacks are reduced.
This points to ownership structures impacting taste-based discrimination.
A key mechanism commonly identified in mitigating economic distress is relationship
lending. Relationship lending occurs when a bank repeatedly interacts with a borrower, and
through this, obtains private information (Boot, 2000; Beck, Degryse, De Haas, and Van Horen,
2018). This soft information can reflect creditworthiness, thus enabling banks to adapt their
lending behavior accordingly (Rajan, 1992; von Thadden, 1995; Beck, Degryse, De Haas, and Van
Horen, 2018). The paradigm has been that large banks conduct primarily transaction lending, while
small banks conduct primarily relationship lending (e.g., Cole, Goldberg, and White, 2004; Berger,
Miller, Petersen, Rajan, and Stein, 2005; Berger, Bouwman, and Kim, 2017). Minority-owned
banks are smaller relative to other banks, so we expect minority-owned banks to also have a higher
proportion of relationship loans. We also expect minority-owned banks to more effectively collect
and process both soft and hard information from their borrowers via shared minority identity.
Specifically, loan officers and other employees of minority-owned banks may be better able to
screen and monitor minority-owned businesses and households because of better comprehension
and expression of verbal and nonverbal communication, higher mutual trust, and/or greater taste
for loans to minorities. These improve access to additional information that would increase the
statistical evaluation of credit access to minority firms and households. Additionally, minority
borrowers may also behave more responsibly and default less when borrowing from minority-
owned banks to maintain a good reputation in their community and avoid costly social sanctions.
Frame, Huang, Mayer, and Sunderam, (2021) show that although minority borrowers have lower
mortgage application completion, approval, and take-up rates, these differences decrease when
minority borrowers work with minority loan officers. This result is consistent with our expected
mechanism, shared minority identity. Here, we consider a specific circumstance, economic
distress. Jiang, Lee, and Liu (2022) also provide evidence that minority loan officers are less likely
to reject minority mortgage borrowers. Vatsa (2021) demonstrates that the loss of a minority-
owned bank decreases the minority credit supply at the census-tract level. In sum, minority-owned
banks as an antidiscriminatory force are expected to positively impact economic resilience during
times of financial and economic crises.
8
2.4 – Small Business Finance
The small business finance body of literature characterizes the sources of small business finance
and how they shift over both the firm and business cycles (e.g., Berger and Udell, 1998; Berger
and Roman, 2018). Firms may access several forms of finance; private debt (e.g., bank loans),
public debt (e.g., mortgage-backed securities), private equity (e.g., venture capital), and public
equity (e.g., stock markets). The source of finance depends on the capacity of the firm to effectively
communicate its quality to potential lenders. At the extremes, public firms have extensive reporting
requirements that enable public stock offerings, while small, young firms have few financial
records and weak or nonexistent reputations, requiring these firms to obtain financing from
individuals and banks. In this study, we focus on bank financing, as it is one of the most important
sources of finance for small businesses (e.g., Berger and Udell, 1998; Robb and Robinson, 2014;
Berger and Roman, 2018). To the authors’ knowledge, this is the first study that considers how
bank-ownership characteristics impact small business access to finance across the business cycle.
Here, we consider two components of this. First, how firms access private debt, or finance,
from banks through both regular business cycle times and crises. Second, how individuals access
public debt in the form of home mortgages, from which these individuals may gain access to equity
for their businesses. We expect minority-owned banks to have stronger ties with their borrowers
via a shared identity, thus reducing information opacity. Thus, to the extent that minority-owned
banks form and strengthen relationships with their borrowers, they would achieve their mission of
both increasing economic growth and resilience via more efficiently selecting their borrowers.
2.5 – Global Financial Crisis
The GFC provides an adverse event to observe how minority-owned banks impact economic
resilience. During the GFC, the four largest banks (Bank of America, Citigroup, JPMorgan Chase,
and Wells Fargo) reduced their small business lending more than other institutions (Chen, Hanson,
and Stein, 2017). DeYoung, Gron, Torna, and Winton, (2015) find that banks that rely more heavily
on relationships continued lending during the GFC. Cotugnoa, Monferrà, and Sampagnaro (2013)
show that credit rationing is reduced through the GFC as relationship strength increases. Bolton,
Freixas, Gambacorta, and Mistrulli (2016) find although relational banks charge higher rates
during normal times, they offer more favorable terms during crises. Beck, Degryse, De Haas, and
Van Horen (2018) provide evidence consistent with relationship-oriented banks, alleviating credit
9
constraints through economic downturns; the observed effect increases for small firms and through
worsening economic conditions. The recent literature consistently demonstrates that community
banks enhanced economic resilience through the GFC (Petach and Weiler, 2019; Petach, Weiler,
and Conroy, 2021; Langford and Feldman, 2021). In sum, this research stream indicates
relationship strength, and community banks strengthened economic resilience through the GFC.
We extend this stream by considering how shared minority identity and minority-owned banks
enhanced economic resilience through the GFC.
2.6 – COVID-19 Crisis
The COVID-19 crisis is unique across several characteristics. It is a public health crisis, with
implications for the broader economy. It represents the largest economic shock in modern U.S.
history, as well as the shortest recession. Although this crisis caused a severe economic downturn,
no banking crisis was observed in 2020 (Berger and Demirgüç-Kunt, 2021). This has sparked a
rapidly expanding literature.
Considering bank size, Levine, Lin, and Wensi (2020) find counties with a higher
proportion of small banks experience less employment declines through the COVID-19 crisis. A
series of studies also considered how relationship lending has fared thus far through the COVID-
19 crisis. Across a range of outcomes, Berger, Bouwman, Norden, Roman, Udell, and Wang
(2022a) find relationship lenders take advantage of firms through the COVID-19 crisis. This
finding is in contrast with normal times (e.g., Kysucky and Norden, 2016) and previous crises
(e.g., Bolton, Freixas, Gambacorta, and Mistrulli 2016). Using consumer and small business credit
card data, Berger, Bouwman, Norden, Roman, Udell, and Wang (2022b) find relationships benefit
both groups through the COVID-19 crisis. A series of studies also examine the Paycheck
Protection Program (PPP), which was a key policy tool response (see Berger and Demirgüç-Kunt,
2021). Initial evidence indicates this program had a positive impact on employment (Barrazza,
Rossi, and Yeager, 2020; Li and Strahan, 2021) and increased conventional small business lending
(Karakaplan, 2021; Lopez and Spiegel, 2021; Marsh and Sharma, 2021). Relevant here, Atkins,
Cook, and Seamans (2021) find Black-owned firms received loans that were approximately half
the size of observationally similar White-owned firm in the early stages of the program. As
nontraditional lenders entered the program, this effect disappeared. Berger, Epouhe, and Roman
(2022) find subprime consumers with increased access to Troubled Asset Relief Program (TARP)
recipient banks increased their debt burden, while the opposite trend was observed for consumers
10
with higher access to PPP banks, demonstrating how the PPP was unique relative to TARP. These
results guide our study design; we condition our results on both per capita PPP and TARP values.
2.7 – Hypotheses and Empirical Approach
Based on the arguments discussed previously, in this study, we test three hypotheses: (1) minority-
owned banks positively impact employment growth through regular business cycle times; (2)
minority-owned banks mitigate employment losses through the GFC; and (3) minority-owned
banks mitigate employment losses through the COVID-19 crisis. We test these hypotheses using
data sets constructed at the county-year, bank-year, bank-county-year, and loan-bank-county-year
levels. At the county-year level, we test the effect of the minority-owned bank market share on
employment growth. At the bank-year, bank-county-year, and loan-bank-county-year levels, we
test the effect of bank minority-ownership status on small business and residential mortgage
lending activity. At each level, we compare regular business cycle times to each of the recent
crises. We also condition on a consistent set of socioeconomic, demographic, and bank-health
variables across data sets. Together, the analyses estimate the impact of minority-owned banks on
economic resilience and point to an underlying mechanism: the credit supply.
3. County-Level Economic Growth and Resilience Analysis
3.1 – Econometric Model
Our central argument is that minority-owned banks strengthen economic growth and resilience in
their communities, and this is achieved through more effective communication with borrowers via
a shared-minority-identity. Thus, our first objective is to estimate the effect of the minority-owned
bank market share on county-level economic growth and resilience, as captured by employment
growth. This model estimates the effect of minority-owned bank market share on overall, minority,
and White employment growth through regular business cycle times, the GFC, and the COVID-
19 crisis, and takes the following form:
(1) 𝛥 𝐸𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡𝑖𝑡 = 𝛽1 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 +
𝛽2 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 × 𝐺𝐹𝐶𝑡 +
𝛽3 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 × 𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 +
𝜁 × 𝑈𝑖𝑡−1 + 𝑆𝑡𝑎𝑡𝑒𝑖 + 𝑌𝑒𝑎𝑟𝑡 + 𝜀𝑖𝑡 .
11
We observe economic growth and resilience with 𝛥 𝐸𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡𝑖𝑡 , the year-over-year
percent change in the total number of employed individuals in county i and year t. We calculate
this value for all individuals, minority individuals, and White individuals. We observe the market
share of minority-owned banks with 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 , the proportion of deposits
held by minority-owned banks. 𝐺𝐹𝐶𝑡 is a GFC indicator variable (equal to one in 2007‒2009 and
zero otherwise). 𝐶𝑂𝑉𝐼𝐷 − 19𝑡 is a COVID-19 crisis indicator variable (equal to one in 2020 and
zero otherwise).
The effects of interest here are the minority-owned bank market share (𝛽1 ×
𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 ), as well as two interaction terms: between the minority-owned
bank market share and the GFC (𝛽2 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 × 𝐺𝐹𝐶𝑡 ), and the
minority-owned bank share and the COVID-19 crisis (𝛽3 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘 𝑀𝑎𝑟𝑘𝑒𝑡 𝑆ℎ𝑎𝑟𝑒𝑖𝑡 ×
𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 ), respectively. The minority-owned bank market share coefficient (𝛽1 )
represents the impact of the minority-owned bank market share on employment growth in regular
business cycle times, or economic growth. The coefficients of the interaction terms
(𝐺𝐹𝐶: 𝛽2 ; 𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠: 𝛽3 ) represent the effect of minority-owned bank market share on
employment growth in each crisis relative to regular business cycle times, or economic resilience.
In each case, positive coefficient estimates indicate minority-owned banks have positive impacts
on employment growth or resilience. Given that the model includes year fixed effects, GFC and
COVID-19 Crisis indicators do not appear in non-interacted form.
𝑈𝑖𝑡−1 represents a vector of county-level time-variant socio-economic, demographic, and
bank health control variables (see Table 1B for a list and definitions). 𝑆𝑡𝑎𝑡𝑒𝑖 represents state-level
fixed effects, thus conditioning our results on time-invariant state-level differences. 𝑌𝑒𝑎𝑟𝑖
represents year-level fixed effects, thus conditioning our results on national trends. So, we
condition on time-invariant state-level characteristics, time-variant national trends, as well as
county-level time-variant trends in socioeconomic, demographic, and bank health characteristics.3
3.2 – Data Set Construction
One of our chief identification concerns is omitted variable bias. In particular, bank health may be
an intervening influence. Chodorow-Reich (2014), using exposure to Lehman Brothers as an
instrument for bank health, find that firms with lending relationships to less healthy banks were
3
In unreported results, we also use the labor market as unit of analysis instead of county, and our main results hold.
12
less likely to obtain credit, paid higher interest rates, and experienced reduced employment. Kiser,
Prager, and Scott (2015) provide evidence indicating that as bank health increases, small business
loan volume is maintained. If we fail to account for bank health, and minority-owned banks are
healthier than nonminority-owned banks, then we may overestimate the effect of interest.
Conversely, if nonminority-owned banks are healthier than minority-owned banks, then we may
underestimate the effect of interest. To mitigate these concerns, we condition our analyses on bank
health by including a vector of time-variant bank health control variables.
We obtain bank data, including the minority-bank ownership status (RSSD9320), as well
as bank health variables from the FFIEC Regulatory Call Reports for the period 2006‒2020. For
all banks in our sample, we obtain Call Report data for each quarter and construct the annual value
for each variable as the mean value of all quarters in that year.4 The bank-level data set is at the
commercial bank level. We obtain TARP data (2008‒2010) from the U.S. Department of the
Treasury’s website (U.S. Treasury). We use the corresponding RSSD9001 (Call Reports ID) to
match each data set. We exclude observations that have missing data for total assets or common
equity. All financial variables have been adjusted using the GDP deflator to be in real 2021 dollars.
In the county-level data set, we convert these bank data to the county-level based on the proportion
of their deposits across counties in which they have operations using the FDIC Summary of
Deposits database for the given year.
Local characteristics could also be a source of omitted variable bias. We use this concern
to guide our selection of additional control variables. We obtain county-level socioeconomic and
demographic control variables from the U.S. Census American Community Surveys (ACS) 5-Year
Estimates. These are available between 2009 and 2019, and thus we hold them constant to their
2009 values for 2006‒2008 and their 2019 values for 2020. We collect industrial composition data
for the 2006‒2020 time period from the Quarterly Census of Employment and Wages (QCEW).
We collect the Housing Price Index (HPI) from the Federal Housing Finance Agency (FHFA). The
final data set is constructed between 2006 and 2020. The control variables use a one-year lag, so
these variables were also collected for 2005. To construct our dependent variables, we collect
county-level employment growth data for the period 2006‒2020 from the Quarterly Workforce
Indicators (QWI). Similar to the bank-level variables, we obtain QWI data for each quarter and
construct the annual value for each variable using the mean value of all quarters in that year.
4
In unreported results, we also use variables as of the fourth quarter in each year instead, and our main results hold.
13
We collect data for each county in the contiguous U.S. We exclude counties with missing
data for the control variables in our analysis. Our final data set contains 2,552 counties (82.1% of
all counties; 91.8% of the population).
3.3 – Summary Statistics
Table 1 provides variable definitions, while Table 2 provides summary statistics for each variable.
In terms of employment growth, the overall mean value is 0.02%, while the minority mean value
is 2.41%. The Minority Bank Market Share mean value, approximately 0.0043%, indicates that
most counties have low minority bank market shares.
In terms of socio-economic and demographic measures, the average proportion of minority
individuals is 22.33%, Working Age is 51.38%, Bachelor’s Degree is 14.50%, Population Density
is 0.13, and HPI is 293.49. In terms of industrial composition, the mean proportion of workers in
Manufacturing is 16.12%, Arts, Entertainment, and Recreation is 1.53%, Accommodation and
Food Services is 9.97%, and Public Administration is 7.44%. Considering bank CAMELS proxies,
the declared set of financial criteria used by regulators to assess the health of banking
organizations, we find the average county over our time period has aggregated bank Capital
Adequacy of 10.91%, Asset Quality of 1.50%, Management Quality of 0.96%, Earnings of 0.85%,
Liquidity of 26.08%, and Sensitivity to Market Risk of 15.02%. Considering other bank
characteristics, we find the average county over our time period has Bank Deposits ($100s per
capita) of $1,029.35, Bank Assets (Log) of 27.56 and Bank Age of 97.57. In terms of bank
ownership, 75.29% are Bank Holding Company Owned. In terms of regulation, 39.35% are
regulated by the OCC, 42.93% are regulated by the FDIC and 17.71 % are regulated by the Federal
Reserve System. Considering market characteristics, we find the average county over our time
period has aggregated Herfindahl-Hirschman Index (HHI) of 0.16, and proportion of banks with
the majority of deposits in metropolitan areas of 69.01%. Considering the magnitude of each policy
level, the TARP Amount (per capita) is $2.25, while the PPP Amount ($1,000s per capita) is
$65.50.
3.4 – Determinants of Minority-Owned Bank Presence
Table 2 also provides the mean value of each variable for counties with and without minority-
owned banks. Counties with minority-owned banks have higher overall employment growth rates
(difference: 0.45 p-value < 0.01), though lower minority employment growth rates (difference: -
14
0.89%; p-value < 0.01). This indicates minority-owned banks are located in communities with
stronger overall economic growth, though with weaker minority employment growth. These
counties also have significantly more minority individuals (26.84%; p-value < 0.01). Of course,
selection is an issue here; the remaining variables are also significantly different between the
groups.
In Table 3, we provide four regressions, which identify determinants of minority-owned
bank presence. Model 1 includes socioeconomic, demographic, and industrial composition
variables, while Model 2 also includes Minority Ancestry (our instrumental variable used in later
analyses in Section 3.6). Minority Ancestry is the proportion of residents with ancestry classified
as a minority (ACS). We classify residents as having minority ancestry if their ancestry is African,
Hispanic, Asian-Pacific, or Arab, based on data from the U.S. Census ACS, which reports the first
ancestry of people residing in a county. Model 3 includes socioeconomic, demographic, industrial
composition, and bank health variables, while Model 4 also includes Minority Ancestry. These
analyses reveal several trends. Conditional on the proportion of minority residents, Minority
Ancestry positively impacts the Minority Bank Market Share. This is a key result. Minority
Ancestry is determined many years prior to our sample, local residents cannot influence it, and it
is expected to have little direct impact on economic growth or resilience, yet it impacts the Minority
Bank Market Share. We also observe that Bachelor’s Degree, Arts, Entertainment, and Recreation,
Capital Adequacy, Sensitivity to Market Risk and Deposits in Metropolitan Regions are positively
correlated with the Minority Bank Market Share, while Population Density, Median Household
Income (Natural Log), BHC Ownership, and OCC Regulation are negatively correlated with the
Minority Bank Market Share. In sum, these results provide insight into the determinants of the
location of minority-owned banks.
3.5 – Effects of Minority-Owned Bank Market Share on Economic Growth and Resilience –
OLS and Propensity Score Matching (PSM) Regression Results
Table 4 provides the main estimation results for equation (1) (state and year fixed effects excluded
for brevity). Here, we test whether minority-owned banks impact economic growth and resilience
for the community overall, as well as for minority and nonminority individuals (Whites). We test
these by constructing the dependent variable, the employment growth rate, using each group
(Overall: Models 1‒2; Minority: Models 3‒4; White: Models 5‒6). Here, we provide two
specifications for each dependent variable: (1) the Full Sample, which includes all counties within
15
our final data set and (2) the PSM Sample. The purpose of the second specification is to mitigate
selection concerns. In this specification, we use characteristics in 1990 (minority population,
proportion of workers in the finance industry, per capita income, total population, rurality) to
predict the probability of having a minority-owned bank and limit the sample to those counties on
common support. We also limit the sample to those counties that either contain a minority-owned
bank or are adjacent to a county containing a minority-owned bank. In the PSM sample, we
compare counties that were similar along a range of characteristics in 1990 and evolved to either
establish a minority-owned bank or did not. The purpose of this step is to reduce the bias in the
observable characteristics. We test this in Figure S1, showing that the bias in several key
socioeconomic and demographic variables is reduced. In sum, across specifications, we find
evidence consistent with minority-owned banks strengthening economic resilience, though not
economic growth.
The Minority Bank Market Share coefficient estimates reflect how minority-owned banks
influence employment growth through regular business cycle times. The coefficient estimates do
not consistently reach statistical significance. Thus, these results do not provide evidence
consistent with minority-owned banks contributing to economic growth through regular business
cycle times. We do not suggest that minority-owned banks do not have positive economic impacts
on their communities through regular business cycle times. Rather, our analyses do not provide
strong evidence consistent with this; other analyses may capture positive impacts.
The Minority Bank Market Share × GFC coefficient estimate is positive and statistically
significant across specifications, thus providing evidence that minority-owned banks positively
impacted economic resilience through the GFC. These results reveal additional points here as well.
First, the magnitude of the coefficient estimates is the largest in the minority specifications
(Models 3‒4). This indicates minority-owned banks strengthen economic resilience the most
within their own communities. Second, the coefficient estimates are consistently positive in the
overall and White specifications as well (Models 1‒2; 5‒6). This indicates minority-owned banks
have positive impacts beyond their own communities.
To consider the economic significance of the minority-owned bank market share on
economic resilience through the GFC, we use the PSM Sample Minority Employment coefficient
16
estimate (Model 4: 63.72). This coefficient estimate is smaller in magnitude relative to the Full
Sample coefficient estimate (Model 3: 111.9), thus the following analysis provides a conservative
estimate of marginal effects. This coefficient estimate indicates relative to normal times, through
the GFC, counties with a 100 percentage point increase in Minority Bank Market Share
experienced a 6.37 percentage point higher-minority employment growth. Given the minority
employment growth rate (-1.30%) over the GFC, this value is significant in an economic sense. In
the average county (population: ~140,000; minority employees: ~13,000), we expect this to result
in approximately 800 more employed minority individuals per year. Applied to the entire sample,
this would result in ~1.9 million more employed individuals per year (800 minority
individuals/county × 2,316 counties = ~1.9 million employed individuals). Alternatively phrased,
our analyses indicate that, if all nonminority-owned banks in the U.S. behaved in a manner
consistent with minority-owned banks, at least 1.9 million more minority individuals would have
maintained employment through each year of the GFC.
The Minority Bank Market Share × COVID-19 Crisis coefficient estimate provides similar
results. In the minority specification, this coefficient estimate is positive and statistically
significant in each specification. Thus, these results also provide evidence consistent with
minority-owned banks strengthening economic resilience through the current COVID-19 crisis.
Similar to the previously described results, we also observe positive impacts for overall and White
employment; however, these latter coefficient estimates reach statistical significance only in the
PSM specification. Thus, although these results suggest minority-owned banks have a positive
impact beyond their own communities through the current COVID-19 crisis, we treat these results
with caution.
To consider the marginal effect of the minority-owned bank market share on economic
resilience through the COVID-19 crisis, we use the PSM Sample Minority Employment coefficient
estimate (Model 4: 79.21). This coefficient estimate indicates relative to normal times, through the
COVID-19 crisis, counties with a 100 percentage point increase in Minority Bank Market Share
experienced a 7.92 percentage point increase in minority employment growth. Given the minority
employment growth rate (-2.55%) over the COVID-19 crisis, this value is significant in an
economic sense. In the average county (population: ~140,000; minority employees: ~13,000), we
expect this to result in approximately 1,000 more employed minority individuals per year. Applied
to the entire sample, this would result in ~2.4 million more employed minority individuals per year
17
(1,000 minority individuals/county × 2,316 counties = ~2.4 million employed individuals).
Alternatively phrased, our analysis indicates that if all banks in the U.S. behaved in a manner
consistent with minority-owned banks, at least 2.4 million more minority individuals would have
maintained employment through each year of the COVID-19 crisis.
We also provide the coefficient estimates for the socioeconomic, demographic, and bank
health control variables. These coefficient estimates reflect the impact of the given variable on the
employment growth rate in regular business cycle times and demonstrate several key trends. First,
the minority share of the population negatively impacts the minority employment growth rate
through regular business cycle times, and several other socioeconomic and demographic factors
(Population Density, HPI, Median Household Income, Manufacturing) consistently have an
impact across groups. Second, in the nonmatched samples (Models 1, 3, and 5), many of the bank
health coefficient estimates are significant; however, these effects either disappear or are reduced
in magnitude in the matched sample (Models 2, 4, and 6). This builds confidence in our matching
technique. In sum, the first test of our hypotheses indicates minority-owned banks strengthen
economic resilience in their own communities, as well as nonminority communities, through both
the GFC and the COVID-19 crisis.
3.6 – Additional Identification Tests: PSM and Instrumental Variable Analysis
As previously outlined, in Table 4 (Models 2, 4, and 6), we use a PSM technique to address
selection concerns. This technique mitigates selection concerns around the local community —
minority-owned banks may locate in unique areas. A second selection concern exists here —
minority-owned-banks may be unique. In particular, minority-owned banks are relatively small.
In 2020, 84.8% (89 of 105) of minority-owned banks had fewer than $2 billion in assets. Thus, in
our prior results, the observed effect may be driven by differences in bank behavior across the size
dimension. To mitigate this concern, we conduct two additional PSM analyses in which we instead
confine the universe of banks to those with less than $10 billion ($2) billion in assets (see Table 5
for results). Thus, in these analyses we limit our sample to banks that are similar in size to our
banks of interest. In these analyses, our coefficient estimates are driven by competition between
similarly sized banks. These results are statistically and economically significant and highly
consistent with our main findings. Notably, the magnitude of the coefficient estimates increase as
more banks are included in the analysis. This result is consistent with expectations — the
coefficient estimates estimated in the more limited samples use the comparison between minority-
18
owned-banks to similarly sized counterparts. We expect smaller banks to more effectively use soft
information, and thus better compete with minority-owned banks. Thus, since the sample is limited
to smaller banks, we expect the effect size to be diminished.
Reverse causality is also a concern here. Local conditions, economic and otherwise, may
drive the location of minority-owned banks. To address this concern, we use an instrumental
variable (IV) model as an additional identification strategy (see Table 6 for results). Our
instrument, Minority Ancestry, represents the proportion of residents with ancestry classified as a
minority (ACS). As mentioned in Section 3.4, we classify residents as having minority ancestry if
their ancestry is African, Hispanic, Asian-Pacific, or Arab, based on data from the U.S. Census
ACS, which reports the first ancestry of people residing in a county. We expect counties with a
higher proportion of people with minority ancestry to be more likely to have a higher proportion
of minority-owned banks. The key terms in our model are the Minority Bank Market Share and
the interactions between the Minority Bank Market Share and the GFC as well as COVID-19
Crisis. Thus, we construct three instruments: (1) Minority Ancestry; (2) Minority Ancestry × GFC;
and (3) Minority Ancestry × COVID-19 Crisis, resulting in three separate First-Stage IV equations
(Models 1‒3).
This identification strategy relies on two assumptions: (1) Minority Ancestry influences the
Minority-Bank Ownership Share; (2) Minority Ancestry only impacts economic resilience through
the Minority Bank Market Share. We provide empirical evidence (see Table 5; Models 1‒3) to
support the first assumption. Each of the key IV First-Stage coefficient estimates, Minority
Ancestry × GFC and Minority Ancestry × COVID-19 Crisis, are sufficiently strong (F-Statistic >
10). A violation of the second assumption would require Minority Ancestry to impact economic
growth or resilience, either directly or indirectly. Ancestry is established well in advance of
minority-bank formation, and current residents cannot influence it. Furthermore, it is not expected
to directly impact current economic outcomes in a local area. Finally, given we condition our
results on time-invariant state-level characteristics, as well as a battery of county-level time-variant
characteristics, we mitigate concerns a violation of the second assumption occurs via these
characteristics.
Our main results hold for overall and minority employment growth, further strengthening
confidence in our results. Notably, the Minority Bank Market Share coefficient estimate is
consistently positive and significantly different than zero. This indicates that Minority Bank
19
Market Share positively impacts employment growth in regular business cycle times. This model
estimates the local average treatment effect. Here, this represents the impact of the Minority Bank
Market Share on employment growth, for those counties induced to have higher shares by Minority
Ancestry. Alternatively phrased, although a causal claim can be made here, it narrows the relevance
of the conditions. Thus, we treat this coefficient estimate with caution.
3.7 – Additional Identification and Other Robustness Checks
To address variable construction and omitted variable concerns, we conduct several robustness
checks. In our first two robustness checks, we consider variable construction concerns by
constructing alternative specifications of the GFC and minority-owned banks.
In the first robustness check, we construct the GFC indicator using an alternative definition.
In our main results, we define the GFC as occurring between 2007 and 2009. The GFC definitions
we use (NBER Business Cycle Committee: December 2007‒June 2009; Berger and Bouwman,
2013: Q3 2007‒Q4 2009) each indicate the crisis began in 2007. Thus, our current variable
captures some noncrisis times, and a more appropriate definition may be 2008‒2009. In our first
robustness check, we use this alternative definition (see Table S1.A), and our main results hold.
In the second robustness check, we construct the Minority-Owned Bank Share variable
using an alternative definition. In our main results, minority-owned banks are identified from the
Call Reports. The FDIC maintains a list of Minority Depository Institutions (MDIs), which relies
on a similar, though unique definition. The MDI designation has two requirements: (1) the majority
of the voting stock be owned by minorities or (2) the board of directors is comprised predominately
of minorities and the community the bank serves is predominantly minority (FDIC 2021). As a
robustness check, we use this alternative definition to construct our minority-owned bank market
share (see Table S1.B), and our main results hold.
In our main results, our treatment group consists of banks owned by racial and ethnic
minorities, while women-owned banks are included in the control group. A concern here is that
the observed effect is driven by banks owned by racial and ethnic minorities outperforming
women-owned banks. To address this concern, we also consider an alternative specification where
we exclude women-owned banks from the control group (see Table S1.C). Our main results hold.
In our main results, we use a PSM technique to reduce selection bias. This specification
uses one neighbor, without replacement on common support. To build confidence in this
specification, we conducted one more analysis using another PSM specification. This specification
20
conducts the matching with three neighbors, with replacement. We provide the results of these
analysis in Table S1.D, and our main results hold.
Finally, it is also possible racial biases serve as an omitted variable in this analysis.
Communities with higher minority-owned bank market shares may form in regions with lower
racial biases. Communities with lower racial biases may also have stronger economic growth and
resilience. To consider this, we conduct a series of regressions that include measures of racial bias:
fair housing law, interracial marriage, racial bias index, slave state, racial animosity index, and
Black Lives Matter protests (see Table S2.A for variables and definitions, and Table S2.B-D for
results). These measures are primarily constructed at the state level; thus, we remove state-level
fixed effects in these specifications. Our main results hold, indicating racial biases do not drive the
observed result.
4. Underlying Channels: Bank-Level Analysis Using Call Reports
4.1 – Econometric Model
Our proposed mechanism through which minority-owned banks influence economic growth and
resilience is the credit supply. To consider this, we test whether minority-owned banks lend to
firms at higher rates in regular business cycle times, as well as through the GFC and the COVID-
19 crisis in comparison to regular business cycle times. We estimate the effect of minority-bank
ownership status on small business lending activity in regular business cycle times and through
each economic downturn using the model provided below:
(2) 𝛥 𝑆𝑚𝑎𝑙𝑙 𝐵𝑢𝑠𝑖𝑛𝑒𝑠𝑠 𝐿𝑒𝑛𝑑𝑖𝑛𝑔𝑏𝑡 = 𝜃1 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 + 𝜃2 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 × 𝐺𝐹𝐶𝑡 +
𝜃3 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 × 𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 + 𝜆 × 𝑈𝑏𝑡−1 + 𝑌𝑒𝑎𝑟𝑡 + 𝜉𝑏𝑡 .
We capture the credit supply using 𝛥 𝑆𝑚𝑎𝑙𝑙 𝐵𝑢𝑠𝑖𝑛𝑒𝑠𝑠 𝐿𝑒𝑛𝑑𝑖𝑛𝑔𝑏𝑡 , which represents the
annual change in small business lending activity, as measured in dollar amounts.
𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 represents the minority-bank ownership status (equal to one if the bank is
minority owned, and zero otherwise). 𝐺𝐹𝐶𝑡 represents the GFC (equal to one 2007‒2009 and zero
otherwise). 𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 represents the COVID-19 crisis (equal to one in 2020 and zero
otherwise). The key effects here are minority-owned bank status (𝜑1 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 ) as well
as the interactions between this and the GFC (𝜑2 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 × 𝐺𝐹𝐶𝑡 ) and the COVID-
21
19 crisis (𝜑3 × 𝑀𝑖𝑛𝑜𝑟𝑖𝑡𝑦 𝐵𝑎𝑛𝑘𝑏𝑡 × 𝐶𝑂𝑉𝐼𝐷 − 19 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 ). The minority-owned bank market
share coefficient (𝜑1 ) represents the effect of minority-owned bank ownership status on small
business loan activity in regular business cycle times. The coefficients of the interaction terms
(𝐺𝐹𝐶: 𝜑2 ; 𝐶𝑂𝑉𝐼𝐷 − 19: 𝜑3 ) represent the effect of minority-owned bank ownership on small
business lending in each crisis relative to regular business cycle times. Given that the model
includes year fixed effects, GFC and COVID-19 Crisis indicators do not appear in noninteracted
form.
Consistent with the previous analyses, we incorporate three series of control variables here:
(1) bank health variables, (2) socioeconomic and demographic controls, and (3) industrial
composition variables. Using branch-level data, we aggregated county-level socioeconomic,
demographic, and industrial composition characteristics up to the bank-level based on the banks’
deposit shares in different local markets of operation. Thus, these variables represent the
characteristics of the branch network. We use the same socioeconomic, demographic, and bank
health variables as outlined in the county-level analysis (see Table 1 for a list and definitions).5
4.2 – Data Set Construction
We collected financial data for each commercial bank in the U.S. from the FFIEC regulatory Call
Reports (9,301). Our sample is limited to commercial banks (RSSD9331 = 1) with nonmissing
data for the key variables in the analysis over 2006‒2020; thus, the final data set has 58,443
observations for 5,573 banks (59.9% of all banks). For all of the banks in our sample, we obtain
Call Report data for each quarter and construct the annual value for each variable using the mean
value of all quarters over each year.6
4.3 – Summary Statistics
Table 7 provides bank-level summary statistics, both overall and by bank ownership status. This
provides information on differences between minority and nonminority-owned banks. Notably,
the Small Business Loan Activity Growth is higher for minority-owned banks. In terms of assets
and CAMELS, we find minority-owned banks have fewer assets, higher Capital Adequacy, Asset
Quality, Management Quality, and Sensitivity to Market Risk while minority-owned banks have
5
In unreported results, using an alternative definition for minority-ownership status (MDI banks) yields consistent
findings.
6
In unreported results, we also use variables as of the fourth quarter in each year instead, and our main results hold.
22
lower Earnings and Liquidity. Minority-owned banks have more Deposits in Metropolitan
Regions, are located in less concentrated markets, and are younger. They are also less likely to be
owned by a bank holding company. The results demonstrate minority-owned banks differ
significantly from nonminority-owned banks across a range of outcomes.
4.4 – Regression Results
Table 8 provides the main estimation results for equation (2) (year fixed effects excluded for
brevity). We use the overall small business lending growth as the dependent variable across
specifications. In Models 1‒3, we incorporate additional sets of control variables in each
subsequent model, thus testing our main hypotheses across several specifications. In Model 4, we
examine whether minority-owned banks have stronger effects in counties with higher-minority
populations. We do this by interacting our key coefficients with bank exposure to High Minority
Area, an indicator variable equal to one when the minority population exceeds the national mean
and is zero otherwise.
The Minority Bank coefficient estimate is consistently negative, though not consistently
statistically significant. This suggests that in regular business cycle times, minority-owned banks
lend less to small businesses than nonminority-owned banks. This may be driven by demand,
rather than supply; minority-owned banks may operate in markets with less small business lending
activity.
In Models 1‒3, the Minority Bank × GFC coefficient estimate is positive and statistically
significant, thus indicating that minority-owned banks’ small business lending activity was higher
through the GFC. The coefficient estimate indicates the small business lending activity of a
Minority Bank in the GFC continued to grow at approximately 13 percentage points, relative to a
nonminority-owned bank in regular business cycle times. Given the average annual small business
lending amount ($72.34 million), this implies approximately $9.40 million in additional credit
availability per bank. Applied to the entire sample, this implies approximately $52.41 billion in
additional capital availability per year. In short, minority-owned banks maintained their small
business lending activity at higher rates through the GFC. If nonminority-owned banks had acted
in a manner consistent with minority-owned banks, significantly more capital would have been
available to businesses.
In Models 1‒3, the Minority Bank × COVID-19 Crisis coefficient estimate is positive,
though not statistically significant, thus suggesting minority-owned banks lent more through the
23
COVID-19 crisis. Consistent with our prior county-level results, collectively, these results provide
suggestive evidence that minority-owned banks’ small business lending activity differed
substantially from nonminority-owned banks through the COVID-19 crisis as well.
One of our central hypotheses is that minority-owned banks can more effectively acquire
information via a shared minority identity. Based on this, we expect minority-owned banks to lend
at higher rates in areas with higher proportions of minority residents and minority-owned firms.
We test this in Model 4 by inserting an additional set of interaction terms, as described previously.
The Minority Bank × High Minority Area coefficient estimate is negative and significantly
different than zero. This indicates minority-owned banks with branch networks located in high-
minority neighborhoods lend at lower rates.
The key coefficient estimate here, Minority Bank × High Minority Area × GFC, is positive
and significantly different than zero. This indicates minority-owned banks with branch networks
located in counties with higher proportions of minority residents continued lending at higher rates
through the GFC. The coefficient estimate indicates the small business lending activity of a
Minority Bank located in a High Minority Area through the GFC continued to grow at
approximately 21.34 percentage points relative to a Non-Minority Bank located in a Low Minority
Area through regular business cycle times. Relative to the mean Small Business Loan Growth
(11.12%), this value has economic significance.
The second key coefficient estimate here, Minority Bank × High Minority Area × COVID-
19 Crisis, is positive, though not significantly different than zero. This suggests minority-owned
banks with branch networks located in counties with higher proportions of minority residents
continued lending at higher rates through the COVID-19 crisis. Collectively, these results provide
evidence that minority-owned banks improved economic resilience through the credit supply.
They achieve this by maintaining the credit supply in minority communities.
5. Underlying Channels: Additional Analyses
5.1 – Small Business Lending Analysis Using the Community Reinvestment Act Data
The primary channel through which we expect minority-owned banks to impact the local economy
is the small business credit supply. Our prior analyses demonstrate minority-owned banks enhance
economic resilience using county-level analyses, and they continued to lend at higher rates
through crises using bank-level analyses. It is possible minority-owned banks have branch
24
locations in regions that differ from nonminority-owned banks in ways that we are unable to
account for in our previous analyses.
To mitigate this concern, in Table 9, we conduct a series of analyses at the bank-county-
year level using Community Reinvestment Act (CRA) Data. These analyses enable us to compare
the credit supply of banks operating within the same local environment. This data set provides
information on loan originations to commercial and industrial (C&I) firms for 2006‒2020. These
data include bank loans with sizes up to $1 million and specifies loans to businesses with less than
$1 million revenue, denoting small businesses. While the Call Report includes both new
originations and loans from other earlier periods, this data set includes new loan originations,
making it easier to derive final conclusions on credit supply. To condition for local credit demand,
we include state fixed effects, and similar to the prior analyses, we include many county-level
socioeconomic, demographic, and industrial control variables. Similar to prior analyses, to
condition on bank health, we also include bank-level health variables.
We match CRA data to bank characteristics from the regulatory Call Reports, including
minority ownership information and county characteristics. We use the same bank and county
characteristics as of the year-end immediately prior to the loan origination year, as well as state,
and year fixed effects. Our analytic sample has 621,449 total bank-county-year observations
covering 918 banks and 2,551 counties that have CRA information and nonmissing data for the
key variables. We use a similar econometric model as noted previously in equation (2), but we
use bank-county-year level data on small business loan originations as dependent variables, along
with the previously noted bank minority indicator and the interactions with the two crises. We
also include bank and county controls, as well as state and year fixed effects to condition on
unobservable factors.
In Model 1, the Minority Bank coefficient estimate is negative and significantly different
than zero. This indicates relative to nonminority-owned banks, minority-owned banks are less
engaged in small business loan origination. We observed similar results in prior analyses; however,
given those results were observed at the bank-level, they may have reflected local demand. These
results are conditioned on local demand, indicating minority-owned banks may be less engaged in
small business loan origination in regular business cycle times. The Minority Bank × GFC and
Minority Bank × COVID-19 Crisis coefficient estimates each indicate minority-owned banks
continued to lend at higher rates through each crisis. Again, these results condition on local credit
25
demand. Thus, these results demonstrate, conditional on local credit demand, minority-owned
banks continued to lend at higher rates through each crisis.
Similar to prior analyses, in Model 2, we estimate the effect of minority ownership on the
credit supply during the two crises in communities with higher proportions of minority residents.
Consistent with prior results, the key coefficient estimates here, Minority Bank × High Minority
Area × GFC and Minority Bank × High Minority Area × COVID-19 Crisis are each positive and
significantly different than zero. These results provide evidence consistent with our prior results
— minority-owned banks continue lending at higher rates in areas with a higher proportion of
minority residents through crises.
We expect minority-owned banks and firms to more effectively communicate soft
information, which we observe through smaller loans. In Models 3‒8, we consider this by
replicating the analyses conducted in Models 1‒2 with various loan sizes. In the odd-numbered
models (Models 3, 5, and 7), we test our main hypotheses, whether minority-owned banks
continued lending at higher rates through each crisis. In the even-numbered models (Models 4, 6,
and 8), we test our secondary hypotheses, that minority-owned banks lend at higher rates through
these crises in communities with a higher proportion of minority residents.
In Models 3, 5, and 7, the Minority Bank × GFC and Minority Bank × COVID-19 Crisis
coefficient estimates are positive and reach statistical significance in the < $100K and $100‒$250K
specifications, though the magnitude is larger in the < $100K specification. This indicates that our
earlier results are primarily driven by the smallest of loans (< $100K). This may be consistent with
minority-owned banks enhancing economic resilience via shared-identity using specific soft
information. This is because the smallest loans are often regarded as a proxy for small firm-bank
soft information usage in the relationship lending literature mentioned previously.
Similar to previous analyses, in Models 4, 6, and 8, we test whether minority-owned banks
lend at higher rates through each crisis in communities with higher proportions of minority
residents. The Minority Bank × High Minority Area × GFC and Minority Bank × High Minority
Area × COVID-19 Crisis coefficient estimates are positive and consistently reach statistical
significance in the <$100K and $100‒$250K specifications, though the < $100K specification
coefficient estimates are larger in magnitude. This is a key result; it provides evidence consistent
with minority-owned banks continuing to lend via small loans through crises in neighborhoods
with a higher proportion of minority residents, conditional on local demand. In sum, it provides
26
rigorous evidence consistent with our central hypotheses, driven by shared-identity, minority-
owned banks enhance economic resilience via specific soft information.
5.2 – Residential Mortgage Lending Using the Home Mortgage Disclosure Act Data
In addition to the small business credit supply, minority-owned banks could enhance economic
resilience via the residential mortgage credit supply. By maintaining the residential mortgage
supply, minority-owned banks could directly enhance economic resilience by ensuring economic
activities associated with homeownership (e.g., construction) continue. They could indirectly
enhance economic resilience by ensuring that homeowners gain access to home equity, which can
contribute to wealth accumulation and can also be used as a source of business credit. To consider
this, in Table 10, we analyze the effects of bank-level minority ownership on household lending
activity, namely, mortgage loan approvals and interest rate. To conduct these tests, we use the
Home Mortgage Disclosure Act (HMDA) data, which represent approximately 90% of mortgage
lending in the U.S.
We match Call Report bank data with the HMDA mortgage application data using the
lender file developed by Robert Avery. We follow the prior literature to filter the mortgage
applications data by excluding: (1) loans insured by government agencies (i.e., FHA, VA, FAS, or
RHS), (2) refinancing and home improvement loans, (3) loans neither approved nor rejected, (4)
loans to finance nonowner-occupied units, and (5) loans sold by the lender upon origination to
another party. Our final data set contains information on 3,034,722 loan applications for
households in 2,532 counties sent to 817 banks.
We use a similar econometric model as in equation (2), but we use loan-bank-county level
data on household mortgage application, origination, and interest rate outcomes, along with the
previously noted bank minority indicator and the interactions with the two crises. In addition to
the bank and county characteristics used in the prior models, we also control here for additional
loan and household characteristics (loan size, borrower income, coapplicant indicator, race, and
gender). Whereas prior analyses conditioned on the local characteristics and credit demand, these
analyses condition on individual characteristics. We include state and year fixed effects to control
for unobservable factors.
In these regressions, we use a series of dependent variables (loan approval (Models 1‒2),
27
origination (Models 3‒4) and interest rate (Model 5)) to investigate this.7 We find minority-owned
banks consistently increase household mortgage loan approvals and originations during the GFC.
However, they do not affect or decrease these outcomes as well as increase interest rates during
the COVID-19 crisis.8 The Minority Bank coefficient estimate is positive and significantly
different than zero, indicating minority-owned banks are more likely to approve and originate
loans through regular business cycle times. The Minority Bank × GFC coefficient estimate is
positive and significantly different than zero across loan approval and origination specifications.
This indicates minority-owned banks continued to approve and originate home mortgages through
the GFC at higher rates relative to nonminority-owned banks. Given some categories of home
mortgages were contributing factors to the GFC, this is a particularly notable result.
Similar to prior results, in secondary models (Models 2 and 4), we also interact each key
variable with Minority HH, a minority household indicator variable. The key coefficient estimates
here are Minority Bank × Minority HH × GFC and Minority Bank × Minority HH × COVID-19
Crisis. The former is positive, though not significantly different from zero. This provides
suggestive evidence that minority-owned banks continued lending to minority applicants at higher
rates through the GFC. The latter is positive and significantly different than zero. This provides
evidence that minority-owned banks continued lending to minority applicants at higher rates
through the COVID-19 crisis. These results are consistent with minority-owned banks using
shared minority-identity information obtained through superior communication with minority
borrowers to maintain the credit supply.
6. Conclusions
In the past two decades, the U.S. has experienced two severe crises, which caused significant
economic downturns and damage to local communities. Some local communities were more
economically resilient and better able to withstand the crisis, thus experiencing fewer employment
losses. Our research demonstrates that minority-owned banks presence is one important factor that
helped mitigate minority employment losses through the GFC, and in most cases, during the
7
We use the OLS method for our analysis of household mortgage loan approvals. Our choice of a linear rather than
nonlinear model of loan approvals is in line with recent research and is motivated by the fact that nonlinear models
tend to produce biased estimates in panel data sets with many fixed effects, leading to incidental parameter problems
and inconsistent estimates.
8
Model 5 has fewer observations because models include loans from 2018 onward, which report interest rates.
28
COVID-19 crisis. These results add to recent research on community banks and economic
outcomes (Petach and Weiler, 2019; Levine, Lin, and Winsi, 2020; Petach, Weiler, and Conroy,
2021; Langford and Feldman, 2021), by showing that bank ownership characteristics also matter
for economic resilience.
We also find that during times of crisis, minority-owned banks have the additional effect
of mitigating nonminority employment losses. In our main specification, we condition our results
on a battery of socioeconomic, demographic, and bank health variables. This specification tests
minority-owned banks against all banking establishments. These results stand up to two
identification strategies: (1) a PSM technique, which limits our sample to those counties that either
contain minority-owned banks or are adjacent to those containing minority-owned banks and had
similar characteristics in 1990, and (2) an IV technique, where we instrument for minority-bank
market share using minority ancestry. In sum, we demonstrate minority-owned banks mitigate
employment losses across both the GFC and COVID-19 crisis and these results are robust.
We demonstrate that the mechanism that enhances economic resilience is the credit supply;
minority-owned banks maintain the small business and residential real estate credit supplies
through each crisis. In the small business credit supply, stronger effects are observed for regions
with a higher proportion of minority residents. In the household credit supply, stronger effects are
observed for minority households. Each result is consistent with minority-owned banks using more
effective communication to acquire and process information via a shared minority identity.
These results are consistent with both supply-side and demand-side mechanisms. On the
supply side, minority-owned banks more effectively communicate with borrowers, providing a
more nuanced evaluation of risk and reduce taste-based unfavorable racial discrimination. On the
demand side, minority borrowers might also be less likely to strategically default on loans from
minority-owned banks. We are unable to disentangle these using our available data. We encourage
future research that distinguishes between these mechanisms.
The crises studied here differ substantively across a range of dimensions, including causes,
policy solutions, depth, and duration. While the GFC originated within the financial sector, the
COVID-19 crisis was a public health crisis with economic implications. The primary policy
solution for the GFC, TARP, and assisted banks, while the key policy solution for the COVID-19
crisis and the PPP directly supported the real sector. Finally, while the GFC was protracted and
less severe, the COVID-19 crisis was short and more severe. These characteristics point to unique
29
circumstances driving these crises; thus, minority-owned banks may impact the economy
differently through each crisis. Our findings are consistent with this: Minority-owned banks show
even more positive impacts through the GFC than through the COVID-19 crisis.
These results have implications for both finance and policy. First, our research
demonstrates that bank ownership characteristics are an important factor in driving lending
decisions through economic downturns. That is, minority-owned banks behave differently than
nonminority-owned banks through economic downturns, a relevant point for researchers.
In particular, we note two key results: (1) minority-owned banks strengthen economic
resilience for both minorities and nonminorities (Whites), despite effects being larger for
minorities, and (2) minority-owned banks have stronger impacts on economic resilience in
communities with more minority residents. The first key result may be explained by minority-
owned banks simply not discriminating. They approve loans at higher rates through crises and at
similar rates for minority and nonminority-owned firms. “A rising tide may also lift all boats”:
Minority-owned banks, which strengthen minority-owned firms, have the downstream effect of
positively impacting nonminority employment, too. This distinction points to the need for further
research to distinguish between the two mechanisms. The second result is consistent with minority-
owned banks providing stronger relief to minority business owners and homeowners. This points
to financial institutions with stronger community ties enhancing economic resilience. It also points
to minority-owned banks being a potential conduit for strengthening the economic resilience of
minority communities.
Finally, results imply that if all U.S. banks behaved in a manner consistent with minority-
owned banks through the GFC, at least 1.9 million more minority jobs would have been kept and
$50 billion more in credit extended to small businesses on an annual basis. In sum, these results
demonstrate the importance of minority-bank ownership in economic terms.
Notably, we do not find strong evidence consistent with minority-owned banks
contributing to economic growth through regular business cycle times. It is not possible to
econometrically identify and/or analyze a nonresult. However, these findings suggest that
minority-owned banks do not promote economic growth for minorities through regular business
cycle times. We do not suggest minority-owned banks have no long-term positive impacts on the
communities they serve, simply that the analyses conducted here do not provide strong evidence
of this through regular business cycle times. This points to the need for future research.
30
While these results are significant and robust, they have several caveats. One caveat for
our analyses is that in order to construct the minority category, we aggregate several racial and
ethnic minorities into one classification. This decision was driven by the small number of minority-
owned banks, and associated statistical power problems; however, the analysis may oversimplify
nuanced differences between groups. We encourage future research to consider differences across
minority groups using enhanced data sets. In our most robust specifications, we either limit our
sample using the PSM technique or alter the interpretation of the coefficient estimate using the IV
model. Thus, we estimate a local average treatment effect, limiting the generalizability of our
results. Nonetheless, these results demonstrate the importance of minority-owned banks in their
communities.
This research focuses on banks, which is only one aspect of small business finance. Banks
represent an intermediate source of finance in terms of information opacity, collecting both hard
and soft information. Notably, our overall small business credit supply results are driven by small
loans (< $100K). By finding that small loans, which embody support to small firms, are the primary
channel through which minority-owned banks act, our study is relevant to how small businesses
behave through economic shocks. Future studies should consider how the ownership qualities of
other sources of finance (e.g., venture capital) impact their financing behavior, both overall and
through crises. This result, coupled with the employment results, also demonstrates how minority-
owned banks impact the overall economy via small loans. Alternatively phrased, this result
demonstrates how small amounts of capital can have large employment gains.
The research presented here provides evidence demonstrating the role of minority-owned
banks in enhancing economic resilience and highlight the importance of policy that encourages the
formation and survival of minority-owned banks. Our results confirm the Maggie Lena Walker
quote at the beginning of the paper, that communities are strengthened when minorities put their
money together. Finally, results may have broader implications beyond the U.S. banking landscape
alone. It may suggest that a shared minority identity mechanism along many different dimensions
beyond race and ethnicity (e.g., gender, orientation, age, income status) could apply to banks
worldwide that researchers may consider. Such shared minority identity may be powerful in
unleashing positive economic benefits to communities.
31
References
Ambrose, B. W., Conklin, J. N., and Lopez, L. A. (2021) “Does Borrower and Broker Race Affect
the Cost of Mortgage Credit?” Review of Financial Studies 34(2) 790‒826.
Ammons, L. (1996) “The Evolution of Black-Owned Banks in the US between the 1880s and
1990s.” Journal of Black Studies 26(4) 467‒489.
Atkins, R. M. B., Cook, L .D., and Seamans, R. (2021) Discrimination in Lending? Evidence from
the Paycheck Protection Program. Small Business Economics.
Baradaran, M. (2017) The Color of Money: Black Banks and the Racial Wealth Gap. Harvard
University Press.
Barrazza, S., Rossi, M. A., and Yeager, T. J. (2020) “The Short-Term Effect of the Paycheck
Protection Program on Unemployment.” Working Paper. University of Arkansas.
Barth, J., and Xu, J. (2020) “U.S. Minority Banks: Why So Few — After 150 Years?” SSRN
Working Paper.
Bartlett, R., Morse, A., Stanton, R., and Wallace, N.E. (2022) “Consumer-Lending Discrimination
in the Fintech Era.” Journal of Financial Economics 143(1) 30-56.
Bates, T, and Bradford, W. (1980) “An Analysis of the Portfolio Behavior of Black-Owned
Commercial Banks.” The Journal of Finance 35(3) 753‒768.
Beck, T., Behr, P., and Madestam, A. (2018) “Sex and Credit: Do Gender Interactions Matter for
Credit Market Outcomes?” Journal of Banking and Finance 87, 380‒396.
Beck, T., Degryse, H., De Haas, R., and Van Horen, N. (2018) “When Arm’s Length Is Too Far:
Relationship Banking over the Credit Cycle.” Journal of Financial Economics 127(1) 174‒
196.
Becker, G. S. (1957) The Economics of Discrimination, University of Chicago Press.
Begley, T. A., and Purnanandam, A. (2021) “Color and Credit: Race, Regulation, and the Quality
of Financial Services.” Journal of Financial Economics 141(1) 48‒65.
Berger, A. N., and Bouwman, C. H. S. (2013) “How Does Capital Affect Bank Performance
During Financial Crises?” Journal of Financial Economics 109(1) 146‒176.
Berger, A. N., Bouwman, C. H. S., and Kim, D. (2017) “Small Bank Comparative Advantages in
Alleviating Financial Constraints and Providing Liquidity Insurance over Time.” Review
of Financial Studies 30, 3416‒3454.
32
Berger, A. N., Bouwman, C. H. S., Norden, L., Roman, R. A., Udell, G. F., and Wang, T. (2022a)
“Is a Friend in Need a Friend Indeed? How Relationship Borrowers Fare during the
COVID-19 Crisis.” Working Paper. Federal Reserve Bank of Philadelphia.
Berger, A. N., Bouwman, C. H. S., Norden, L., Roman, R. A., Udell, G. F., and Wang, T. (2022b)
“Piercing Through Opacity: Relationships and Credit Card Lending to Consumers and
Small Businesses During Normal Times and the COVID-19 Crisis.” Working Paper.
Federal Reserve Bank of Philadelphia.
Berger, A .N., and Demirgüç-Kunt, A. (2021) “Banking Research in the Time of COVID-19.”
Journal of Financial Stability 57, 100939.
Berger, A. N., Epouhe, O., and Roman, R. A. (2022) “A Tale of Two Bailouts: Effects of Tarp and
PPP on Subprime Consumer Debt.” Working Paper. Federal Reserve Bank of Philadelphia.
.
Berger, A. N., Hasan, I., and Klapper, L. F. (2004) “Further Evidence on the Link between Finance
and Growth: An International Analysis of Community Banking and Economic
Performance.” Journal of Financial Services Research 25, 169‒202.
Berger, A. N., Miller, N. H., Petersen, M. A., Rajan, R. G., and Stein, J. C. (2005) “Does Function
Follow Organizational Form? Evidence from the Lending Practices of Large and Small
Banks.” Journal of Financial Economics 76, 237‒269.
Berger, A. N., and Roman, R. A. (2018) “Finance and the Real Economy: Evidence from the US.”
Handbook of Finance and Development, 261‒288.
Berger, A. N., and Udell, G. F. (1998) “The Economics of Small Business Finance: The Roles of
Private Equity and Debt Markets in the Financial Growth Cycle.” Journal of Banking and
Finance 22, 613‒673.
Bhutta, N., Hizmo, A., and Ringo, D. (2021) “How Much Does Racial Bias Affect Mortgage
Lending? Evidence from Human and Algorithmic Credit Decisions.” Working Paper.
Board of Governors of the Federal Reserve System.
Black, H. A., Robinson, B. L., and Schweitzer, R. L. (2001) “Comparing Lending Decisions of
Minority-Owned and White-Owned Banks.” Review of Financial Economics 10(1) 23‒39.
Black, H. A., Collins, M. C., and Cyree, K. B. (1997) “Do Black-Owned Banks Discriminate
Against Black Borrowers?” Journal of Financial Services Research 11, 189‒204.
Bolton, P., Freixas, X., Gambacorta, L., and Mistrulli, P. (2016) “Relationship and Transaction
Lending in a Crisis.” The Review of Financial Studies 29(10) 2643‒2676.
33
Boot, A. W. A. (2000) “Relationship Banking: What Do We Know?” Journal of Financial
Intermediation 9, 7‒25.
Boschma, R. (2015). “Towards an Evolutionary Perspective on Regional Resilience.” Regional
Studies 49(5) 733‒751.
Brodeur, A., Gray, D., Islam, A., and Bhuiyan, S. (2021) “A Literature Review of the Economics
of COVID-19.” Journal of Economic Surveys 35, 1007‒1044.
Brunnermeier, M. (2021) The Resilient Society. Endeavor Literary Press.
Célerier, C., and Matray, A. (2019) “Bank-Branch Supply, Financial Inclusion, and Wealth
Accumulation.” The Review of Financial Studies 32(12) 4767‒4809.
Chen, B. S., Hanson, S. G., and Stein, J. C. (2017) “The Decline of Big-Bank Lending to Small
Business: Dynamic Impacts on Local Credit and Labor Markets.” NBER Working Paper.
Chodorow-Reich, G. (2014) “The Employment Effects of Credit Market Disruptions: Firm-Level
Evidence from the 2008-9 Global Financial Crisis.” The Quarterly Journal of Economics
129(1) 1‒59.
Christopherson, S., Michie, J., and Tyler, P. (2010) “Regional Resilience: Theoretical and
Empirical Perspectives.” Cambridge Journal of Regions, Economy and Society 3, 3‒10.
Chu, Y., Ma, X., and Zhang, T. (2022) “Bank Public Status and Mortgage Lending
Discrimination.” Working Paper, University of North Carolina-Charlotte.
Clark, J., Huang, H., and Walsh, J. (2010) “A Typology of ‘Innovation Districts’: What It Means
for Regional Resilience.” Cambridge Journal of Regions, Economy and Society 3, 121‒
137.
Cole, R. A., Goldberg, L. G., White, L. J. (2004) “Cookie Cutter vs. Character: The Micro
Structure of Small Business Lending by Large and Small Banks.” Journal of Financial and
Quantitative Analysis 39(2) 227‒251.
Collins, W. J. (2004) “The Housing Market Impact of State-Level Anti-Discrimination Laws,
1960-1970.” Journal of Urban Economics 55(3) 534‒564.
Cotugnoa, M., Monferrà, S., and Sampagnaro, G. (2013) “Relationship Lending, Hierarchical
Distance and Credit Tightening: Evidence from the Global Financial Crisis.” Journal of
Banking and Finance 37(5), 1372‒1385.
Couch, K. A., Fairlie, R., and Xu, H. (2018) “Racial Differences in Labor Market Transitions and
the Great Recession.” Research in Labor Economics 1‒53.
34
DeYoung, R., Gron, A., Torna, G., and Winton, A. (2015) “Risk Overhang and Loan Portfolio
Decisions: Small Business Loan Supply Before and During the Global Financial Crisis.”
The Journal of Finance 70(6), 2451‒2488.
Drexler, A., and Schoar, A. (2014) “Do Relationships Matter? Evidence from Loan Officer
Turnover.” Management Science 60(11), 2722‒2736.
Elyasiani, E., and Mehdian, S. (1992) “Productive Efficiency Performance of Minority and
Nonminority-Owned Banks: A Nonparametric Approach.” Journal of Banking and
Finance, 16(5), 933‒948.
FDIC (2021) Minority Depository Institutions Program.
https://www.fdic.gov/regulations/resources/minority/mdi.html.
Federal Housing Finance Agency (FHFA) (2022) House Price Index.
https://www.fhfa.gov/DataTools/Downloads/Pages/House-Price-Index.aspx.
Federal Reserve Bank of New York (2021) Banking Research Datasets.
https://www.newyorkfed.org/research/banking_research/datasets.html.
Fisman, R., Paravisini, D., and Vig, V. (2017) “Cultural Proximity and Loan Outcomes.” American
Economic Review 107(2), 457‒492.
Frame, W. S., Huang, R., Mayer, E. J., and Sunderam, A. (2021) “Minority Loan Officers and
Minorities’ Access to Mortgage Credit.” Working Paper. Federal Reserve Bank of Dallas.
Fryer, R. G. (2007) “Guess Who’s Been Coming to Dinner? Trends in Interracial Marriage Over
the 20th Century.” Journal of Economic Perspectives 21(2), 71‒90.
Gardner, M. J. (1982) “Black-Owned Commercial Banks: A New Look at Their Performance and
Management.” The Review of Black Political Economy 12(1), 91‒101.
Gerena, C. (2007) “Opening the Vault.” Region Focus — Federal Reserve Bank of Richmond
11(2), 46‒49.
Giacoletti, M., Heimer, R., and Yu, E. G. (2021) “Using High-Frequency Evaluations to Estimate
Discrimination: Evidence from Mortgage Loan Officers.” Working Paper. Federal Reserve
Bank of Philadelphia.
Gjerde, K. A. P., Prescott, P., and Rice, J. (2019) “The Impact of State Fiscal Policy on State’s
Resilience Entering the Global Financial Crisis.” Journal of Regional Analysis & Policy
49(1), 13‒30.
35
Henderson, C. C. (1999) “The Economic Performance of African-American-Owned Banks: The
Role of Loan Loss Provisions.” American Economic Review 89(2), 372‒376.
Holdren, D. P., and Heyliger, W. E. (1993) “The Performance of Minority Banks in a Deregulated
Banking Environment.” The Review of Black Political Economy 22(2), 89‒107.
Hoynes, H., Miller, D. L., and Schaller, J. (2012) “Who Suffers During Recessions?” Journal of
Economic Perspectives 26(3), 27‒48.
Iqbal, Z., Ramaswamy, K. V., and Akhigbe, A. (1999) “The Output Efficiency of Minority-Owned
Banks in the United States.” International Review of Economics & Finance 8(1), 105‒114.
Jiang, E. X., Lee, Y., and Liu, W. S. (2022) “Disparities in Consumer Credit: The Role of Loan
Officers in the FinTech Era.” Working Paper. University of Southern California.
Karakaplan, M. U. (2021) “This Time is Really Different: The Multiplier Effect of the Paycheck
Protection Program (PPP) on Small Business Bank Loans.” Journal of Banking and
Finance 133.
Kashian, R., and Drago, R. (2016) “Minority-Owned Banks and Bank Failures after the Financial
Collapse.” Economic Notes 46(1), 5‒36.
King, R. G., and Levine, R. (1993) “Finance and Growth: Schumpeter Might Be Right.” The
Quarterly Journal of Economics 108(3), 717‒737.
Kiser, E. K., Prager, R. A., and Scott, J. R. (2015) “Supervisory Ratings and Bank Lending to
Small Businesses During the Global Financial Crisis and Great Recession.” Journal of
Financial Services Research 50(2) 163‒186.
Kwast, M. L., and Black, H. (1983) “An Analysis of the Behavior of Mature Black-Owned
Commercial Banks.” Journal of Economics and Business 35(1) 41‒54.
Kysucky, V., and Norden, L. (2016) “The Benefits of Relationship Lending in a Cross-Country
Context: A Meta-Analysis.” Management Science 62, 90‒110.
Langford, W. S., and Feldman, M. P. (2021) “We Miss You George Bailey: The Effect of Local
Banking Conditions on the County-Level Timing of the Great Recession.” Working Paper.
University of North Carolina – Chapel-Hill.
Lawrence, E. C. (1997) “The Viability of Minority-Owned Banks.” The Quarterly Review of
Economics and Finance 37(1), 1‒21.
Lewin, P. A., Watson, P., and Brown, A. (2017) “Surviving the Global Financial Crisis: the
Influence of Income Inequality in US Urban Counties.” Regional Studies 52(6), 781‒792.
36
Levine, R., Rubinstein, Y., and Levkov, A. (2014) “Bank Deregulation and Racial Inequality in
America.” Critical Finance Review 3(1), 1‒48.
Levine, R., Lin, C., and Wensi X. (2020) “Local Financial Structure and Economic Resilience.”
Working Paper. University of California – Berkeley.
Li, L., and Strahan, P. (2021) “Who Supplies PPP Loans (and Does it Matter)? Banks,
Relationships and the COVID Crisis.” Journal of Financial and Quantitative Analysis 56,
2411‒2438.
Lopez, J. A., and Spiegel, M. M. (2021) “Small Business Lending Under the PPP and PPPLF
Programs.” Working Paper.
Marsh, W. B., and Sharma, P. (2021) “Government Loan Guarantees During a Crisis: The Effect
of the PPP on Bank Lending and Profitability.” Working Paper. Federal Reserve Bank of
Kansas City.
Martin, R. (2012) “Regional Economic Resilience, Hysteresis and Recessionary Shocks.” Journal
of Economic Geography 12(1) 1‒32.
Martin, R., and Sunley, P. (2015) “On the Notion of Regional Economic Resilience:
Conceptualization and Explanation.” Journal of Economic Geography 15(1), 1‒42.
Montenovo, L., Jiang, X., Lozano-Rojas, F., Schmutte, I.M., Simon, K.I., Weinberg, B.A., and
Wing, C. (2021) “Determinants of Disparities in COVID-19 Job Losses.” NBER Working
Paper.
Munnell, A. H., Tootell, G. M., Browne, L. E., and McEneaney, J. (1996) “Mortgage Lending in
Boston: Interpreting HMDA Data.” American Economic Review 25‒53.
NBER Business Cycle Committee. (2022) US Business Cycle Expansions and Contractions. Link:
https://www.nber.org/cycles/cyclesmain.html.
Nguyen, H-L. Q. (2019) “Are Credit Markets Still Local? Evidence from Bank Branch Closings.”
American Economic Journal: Applied Economics 11(1), 1‒32.
Opportunity Insights (2021) Economic Tracker.
https://github.com/OpportunityInsights/EconomicTracker/tree/main/data.
Orrenius, P. M., and Zavodny, M. (2010) “Mexican Immigrant Employment Outcomes Over the
Business Cycle.” American Economic Review 100(2), 316‒320.
37
Petach, L., and Weiler, S. (2019) “Geography Matters: The Impact of Geographic Expansion on
Bank Performance During the Global Financial Crisis.” Working Paper. Belmont
University.
Petach, L., Weiler, S., and Conroy, T. (2021) “It’s a Wonderful Loan: Local Financial
Composition, Community Banks, and Economic Resilience.” Journal of Banking and
Finance 126, 106077.
Pike, A., Dawley, S., and Tomaney, J. (2010) “Resilience, Adaptation and Adaptability.”
Cambridge Journal of Regions, Economy and Society 3, 59‒70.
Quarterly Census of Employment and Wages. (2022) U.S. Bureau of Labor Statistics.
https://www.bls.gov/cew/downloadable-data-files.htm.
Quarterly Workforce Indicators (2021) U.S. Census Bureau.
https://www.census.gov/data/developers/data-sets/qwi.html.
Rahe, M. L., Weber, B., Wu, X., and Fisher, M. (2019) “Income Inequality and County Economic
Resistance to Job Loss during the Global Financial Crisis.” Review of Regional Studies.
Rajan, R. G. (1992) “Insiders and Outsiders: The Choice Between Relationship and Arm’s Length
Debt.” Journal of Finance 47, 1367‒1400.
Robb, A. M., and Robinson, D. T. (2014) “The Capital Structure Decisions of New Firms.” Review
of Financial Studies 27, 153‒179.
Ruggles, S., Fitch, C., Goeken, R., Hacker, J., Nelson, M., Roberts, E., Schouweiler, M., and
Sobek, M. (2021) IPUMS Ancestry Full Count Data: Version 3.0 [data set]. Minneapolis,
MN: IPUMS.
Simmie, J., and Martin, R. (2010) “The Economic Resilience of Regions: Towards an Evolutionary
Approach.” Cambridge Journal of Regions, Economy and Society 3(1), 27‒43.
Social Explorer. (2022) Black Lives Matter Protests Data. https://www.socialexplorer.com/.
Stein, J. C. (2002) “Information Production and Capital Allocation: Decentralized versus
Hierarchical Firms.” The Journal of Finance 57(5), 1891‒1921.
U.S. ACS (2022) U.S. Census Bureau Link: https://www.census.gov/programs-surveys/acs.
U.S. Census Bureau (2022) Decennial Survey. https://www.census.gov/programs-
surveys/decennial-census/data.html.
38
U.S. Treasury (2022) Troubled Assets Relief Program (TARP).
https://home.treasury.gov/data/troubled-assets-relief-program.
Vatsa, P. (2021) “Do Minority Banks Matter?” Working Paper.
Von Thadden, E-L. (1995) “Long-Term Contracts, Short-Term Investment, and Monitoring.”
Review of Economic Studies 62(4), 557‒575.
Walker, M. L. (1901) National Park Service — The St. Luke Penny Savings Bank.
https://www.nps.gov/mawa/the-st-luke-penny-savings-bank.html.
Ziorklui, S. Q. (1994) “The Performance of Black-Owned Commercial Banks: A Comparative
Analysis.” The Review of Black Political Economy 23(2) 5‒23.
39
Figure 1: Number of Minority-Owned Banks over Time
This figure shows the evolution over time of the number of minority-owned (≥ 50% minority ownership) banks over
our sample period 2006‒2020. Minority refers to any of the categories of ownership denoted as Black, Hispanic,
Asian, and Other (Native-American, Eskimo, Aleut, Arab, Multi-Racial). In the figure, data on minority bank
ownership come from the FFIEC Call Reports.
40
Figure 2: Location of Minority-Owned Banks Across U.S. Counties
This figure shows a county-level heat map for the location of minority-owned banks in 2020 via an indicator variable equal
to one when a county has a minority-owned bank present, and zero otherwise. Minority refers to any of the categories of
ownership denoted as Black, Hispanic, Asian, Native-American, Eskimo, Aleut, Arab, Multi-Racial. In the figure, blue
shows counties where we have minority-owned banks, white shows counties where we do not have minority-owned banks,
while gray indicates counties with no data. Data on minority bank ownership is from the FFIEC Call Reports, while local
bank presence is from the FDIC Summary of Deposits.
Figure 3: Minority Population (%) Across U.S. Counties
This figure shows a county-level heat map for the minority proportion of the population in 2020. In the figure, blue shows
counties where we have higher percent of minority population (minority represents any population that is not reported as
White), white shows counties where we have little to no percentage of minority population, while gray indicates counties
with no available data. Data on minority population is from the U.S. Census ACS.
41
Table 1. Variable Definitions
This table provides definitions for the variables used in our analyses. Panel A refers to independent, dependent, socioeconomic and demographic controls, while Panel B
refers to bank health and other controls.
Panel A. Independent, Dependent, Socioeconomic and Demographic Control Variables Definitions
Name Definition Source
Δ Employment The year-over-year percent change in the number of employed persons. The data set used to observe
employment, the QWI, is limited to employment by private firms and state and local governments.
QWI (2021)
To adjust coefficient estimates, in our main regression results, this variable has been multiplied by a
factor of 1,000.
Minority-Owned The deposit-weighted proportion of bank branches with greater than 50% ownership by a minority
Summary of Deposits
Bank Market Share individual. Minority refers to any of the categories of ownership denoted as Black, Hispanic, Asian,
FFIEC Call Reports
and Other (Native American, Eskimo, Aleut, Arab, Multi-Racial).
Δ Small Business Lending The year-over-year percent change in total loan activity, as measured by the value of the outstanding
FFIEC Call Reports
loans.
GFC An indicator variable equal to one through 2007‒2009 and zero otherwise (NBER Business Cycle
Committee: December 2007-June 2009; Berger and Bouwman 2013: Q3 2007-Q4 2009). In a NBER Business Cycle Committee
robustness check, we specify this variable as through 2008‒2009. (2022); Berger and Bouwman (2013)
COVID-19 Crisis An indicator variable equal to one in 2020, and zero otherwise.
Minority Population (%) The proportion of residents identified as non-White. Ruggles, Fitch, Goeken, Hacker,
Working Age (25-64; %) The proportion of residents between the ages of 25 and 64. Nelson, Roberts, Schouweiler. and
Bachelor’s Degree (%) The proportion of residents age 25 years and older holding a bachelor’s degree. Sobek (2021); U.S. Census ACS
Population Density The number of residents per square kilometer. (2022)
House Price Index (HPI) The FHFA county-level, annual HPI. To adjust coefficient estimates, in our main regression results,
FHFA (2022)
this variable has been divided by a factor of 1,000.
Median Household Income (Log) The natural log of the family income. U.S. Census ACS (2022)
Industrial Compositions (%) The proportion of workers employed in the manufacturing (2-digit NAICS Code: 31-33), services
QCEW (2022)
(71, 72) and public administration (92) industries by 2-digit NAICS code.
Minority Population (1990) The proportion of residents identified as non-White in 1990.
Ruggles, Fitch, Goeken, Hacker,
Finance (1990) The proportion of workers identified as employed in the finance industry in 1990.
Nelson, Roberts, Schouweiler,
Per Capita Income (1990) The average individual income in 1990.
andSobek (2021); U.S. Census
Total Population (1990) The total number of residents in 1990.
Bureau (2022)
Rurality (1990) The proportion of land classified as rural in 1990.
COVID-19 Death Rate The per capita number of COVID-19 deaths. Opportunity Insights (2021)
42
Panel B. Local Bank Health and Other Variables
Variable Definition Source
Deposits (per Capita) The per capita value of bank deposits. To adjust coefficient estimates, in our main regression results,
FDIC Summary of Deposits
this variable has been divided by a factor of 1,000,000.
Bank Assets (Log) The natural logarithm of bank total assets.
Capital Adequacy (%) The mean equity ratio, defined as the ratio of total equity capital to total assets.
Asset Quality (%) The mean nonperforming loans to total loans ratio. Nonperforming loans are defined as loans, or
leases past due more than 90 days or are no longer accruing interest.
Management Quality (%) The mean overhead costs ratio. The overhead costs ratio is defined as the ratio of premises and fixed
assets expenses to total income.
FFIEC Call Reports
Earnings (%) The mean return on assets. The return on assets ratio is defined as the ratio of total interest and non-
interest income to total assets.
Liquidity (%) The mean ratio of liquid assets to total assets. Liquid assets included here are Currency and Coin,
Money Market Mutual Funds, and Total Investment Securities.
Sensitivity to Market Risk (%) The mean ratio of the absolute difference (gap) between short-term assets and short-term liabilities to
gross total assets.
Bank Herfindahl–Hirschman Index (HHI) The HHI, calculated using branch deposits.
FDIC Summary of Deposits
Deposits in Metropolitan Regions (%) The proportion of deposits held by banks with the majority of their deposits in metropolitan areas.
Bank Age The number of years the bank has been operating. To adjust coefficient estimates, in our main
FFIEC Call Reports
regression results, this variable has been divided by a factor of 1,000.
Bank Holding Company (BHC) The proportion of bank branches in a BHC.
FDIC Summary of Deposits
Ownership (%)
OCC Regulation (%) The proportion of banks regulated by the OCC.
FDIC Regulation (%) The proportion of banks regulated by the FDIC. FFIEC Call Reports
Federal Reserve Regulation (%) The proportion of banks regulated by the Federal Reserve System.
TARP Amount (per Capita) The per capita value of TARP funding. To adjust coefficient estimates, in our main regression results,
U.S. Treasury (2022)
this variable has been divided by a factor of 1,000,000.
PPP Loan Balance (per Capita) The per capita value of originated PPP loans. To adjust coefficient estimates, in our main regression
FFIEC Call Reports
results, this variable has been divided by a factor of 1,000,000.
Minority Ancestry The proportion of residents with ancestry classified as a minority. We classify residents as having
minority ancestry if their ancestry is African, Hispanic, Asian-Pacific, American Indian, or Arab,
U.S. Census ACS (2021)
based on data from the U.S. Census ACS, which reports the first ancestry of people residing in a
county.
43
Table 2: Summary Statistics for Minority-Owned Bank Presence and Employment
This table provides summary statistics for the variables used in our county-level analysis (county-year). The number of observations is 28,129. The time period is 2006‒
2020. Variables using dollar amounts are expressed in real 2020 dollars using the implicit GDP price deflator.
Full Minority Non-Minority Diff: Minority–
Variable: Sample Mean Bank Mean Bank Mean Non-Minority Stdv. p1 p25 p50 p75 p99
Δ Overall Employment Growth 0.02 0.44 -0.01 0.45*** 3.82 -10.58 -1.73 0.35 2.01 8.88
Δ Minority Employment Growth 2.41 1.58 2.47 -0.89*** 7.59 -16.32 -1.04 2.58 5.62 22.67
Δ White Employment Growth -0.19 0.13 -0.22 0.35*** 3.64 -10.26 -1.84 0.06 1.68 8.38
Minority Bank Market Share 0.0043 0.0662 0.0000 -0.0662*** 0.0386 0.0000 0.0000 0.0000 0.0000 0.1366
Minority Population (%) 22.33 47.45 20.61 26.84*** 18.03 2.50 7.87 16.30 32.99 74.56
Working Age (25-64; %) 51.38 52.27 51.32 0.95*** 3.27 41.57 49.74 51.64 53.36 59.11
Bachelor’s Degree (%) 14.50 18.63 14.22 4.42*** 6.32 5.85 10.00 12.93 17.43 35.54
Population Density 0.13 0.70 0.09 0.61*** 0.56 0.00 0.01 0.03 0.07 1.53
Housing Price Index 293.49 459.03 282.12 176.90*** 179.10 111.26 181.34 235.01 348.53 981.68
Median Household Income (Log) 10.98 11.05 10.97 0.08*** 0.23 10.50 10.82 10.96 11.11 11.61
COVID-19 Death Rate 1.97 3.25 1.89 1.37*** 9.72 0.00 0.00 0.00 0.00 47.42
Manufacturing (%) 16.12 9.58 16.57 -6.98*** 12.14 0.00 6.80 13.41 22.69 52.15
Arts, Entertainment, and Recreation (%) 1.53 1.66 1.52 0.14*** 1.82 0.00 0.60 1.12 1.83 9.58
Accommodation and Food Services (%) 9.97 10.16 9.96 0.20 5.53 0.00 7.49 9.63 12.10 29.36
Public Administration (%) 7.44 6.59 7.50 -0.91*** 5.97 0.20 3.86 5.85 8.98 30.87
Bank Deposits ($100s per Capita) 1,029.35 3,993.39 825.75 3,167.64*** 13,560.16 54.03 154.32 220.87 335.66 12,474.53
Bank Assets (Log) 27.56 30.00 27.39 2.61*** 3.01 20.05 26.02 28.09 29.61 33.14
Capital Adequacy (%) 10.91 11.08 10.90 0.17*** 1.30 8.04 10.09 10.88 11.65 14.55
Asset Quality (%) 1.50 1.62 1.49 0.14*** 1.39 0.13 0.67 1.02 1.78 6.67
Management Quality (%) 0.96 0.97 0.96 0.01 0.25 0.62 0.79 0.89 1.07 1.63
Earnings (%) 0.85 0.88 0.85 0.03** 0.62 -1.57 0.71 0.98 1.19 1.81
Liquidity (%) 26.08 27.93 25.95 1.98*** 7.27 12.43 21.32 25.35 29.70 49.31
Sensitivity to Market Risk (%) 15.02 13.94 15.10 -1.15*** 7.87 3.71 9.06 12.68 19.88 36.94
HHI Deposits 0.16 0.12 0.16 -0.05*** 0.14 0.01 0.07 0.12 0.21 0.67
Deposits in Metropolitan Regions (%) 69.01 87.52 67.74 19.78*** 33.10 0.00 40.92 83.15 100.00 100.00
Bank Age 97.57 91.18 98.01 -6.83*** 21.60 42.00 83.80 98.74 111.82 146.35
BHC Ownership (%) 75.29 59.07 76.40 -17.33*** 22.67 7.09 61.37 79.14 95.46 100.00
OCC Regulation (%) 39.35 50.04 38.62 11.42*** 26.66 0.00 18.47 37.21 57.86 100.00
FDIC Regulation (%) 42.93 32.26 43.66 -11.40*** 26.62 0.00 21.82 41.52 61.60 100.00
Federal Reserve Regulation (%) 17.71 17.71 17.70 0.019 19.65 0.00 0.00 12.13 28.50 80.95
TARP Amount (per Capita) 2.25 2.83 2.21 0.62 69.53 0.00 0.00 0.00 0.00 0.00
PPP Loans ($1,000s per Capita) 65.50 62.25 112.74 -50.49*** 321.49 0.00 0.00 0.00 0.00 1,781.96
Minority Ancestry 11.30 10.11 28.55 -18.43*** 13.50 0.87 3.27 6.29 13.20 63.78
44
Table 3: Determinants of Minority-Owned Bank Presence
This table presents estimates from regressions analyzing the determinants of minority bank market share in a county using
a county‒year sample over the period 2006 to 2020. The dependent variable is the Minority Bank Market Share, the deposit-
weighted proportion of branches of minority-owned banks (≥ 50% minority ownership) in a county. Minority Ancestry (%)
is the proportion of residents with ancestors classified as a minority. We include a broad set of other local market (county)
controls that could influence our dependent variable, including Minority Population (%), Working Age (%), Bachelors’
Degree (%), Population Density, HPI, Median Household Income (Log), COVID-19 Death Rate, and several variables for
important industries in the county such as Manufacturing (%), Arts, Entertainment, and Recreation (%), Accommodation
and Food Services (%), and Public Administration (%). All specifications include control variables lagged by one year, as
well as state and year fixed effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at
county level are reported in parentheses. Significance at the 10%, 5%, and 1% level is indicated by *, **, and ***,
respectively.
Model [1] [2] [3] [4]
Minority Bank Minority Bank Minority Bank Minority Bank
Dependent Variable:
Market Share (%) Market Share (%) Market Share (%) Market Share (%)
Independent Variables:
Minority Ancestry 102.0*** 96.84***
[2.63] [2.72]
Minority 58.08*** 8.39 55.26*** 8.07
[3.72] [0.84] [3.85] [0.80]
County Controls:
Working Age -9.00 18.31 -12.39 14.34
[-0.42] [1.08] [-0.61] [0.89]
Bachelor’s Degree 46.05** 65.05*** 54.67** 73.80***
[2.23] [2.64] [2.45] [2.81]
Population Density -2.82* -4.41* -3.02* -4.39**
[-1.65] [-1.96] [-1.76] [-2.02]
Housing Price Index -1.61 -6.19 -4.18 -8.35
[-0.33] [-1.18] [-0.90] [-1.61]
Median Household Income (Natural Log) -24.28*** -33.18*** -28.21*** -37.53***
[-2.74] [-3.04] [-3.05] [-3.28]
COVID-19 Death Rate 0.042 0.041 0.043 0.043
[0.98] [0.99] [0.98] [1.01]
Manufacturing 0.84 -0.44 3.59 1.88
[0.16] [-0.08] [0.72] [0.36]
Arts, Entertainment, and Recreation 60.11* 64.61** 46.91 51.55*
[1.87] [2.03] [1.60] [1.77]
Accommodation and Food Services 20.53 20.06 18.93 18.05
[1.61] [1.60] [1.40] [1.37]
Public Administration 8.27 26.75 10.97 28.52
[0.56] [1.44] [0.73] [1.55]
45
Local Bank Controls:
Deposits -6.04 -8.33
[-0.52] [-0.69]
Assets (Log) 0.14 0.026
[0.49] [0.09]
Capital Adequacy 118.1* 120.5**
[1.89] [2.02]
Asset Quality 66.83 38.31
[1.43] [0.84]
Management Quality 762.1 858.0
[0.95] [1.11]
Earnings -21.31 -69.99
[-0.25] [-0.76]
Liquidity 2.02 1.24
[0.23] [0.15]
Sensitivity to Market Risk 65.32** 53.92**
[2.25] [2.27]
HHI -2.18 -3.59
[-0.38] [-0.64]
Deposits in Metropolitan Regions 10.29** 11.13***
[2.56] [2.74]
Bank Age -0.12** -0.074**
[-2.38] [-2.07]
BHC Ownership -11.75** -9.83**
[-2.20] [-2.15]
OCC Regulation -13.12*** -13.06***
[-2.59] [-2.65]
FDIC Regulation -1.11 -1.15
[-0.29] [-0.30]
TARP Amount (per Capita) -1,251.8 1,132.0
[-1.38] [0.79]
PPP Loan Balance (per Capita) -0.29 -0.60
[-0.40] [-0.82]
State FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 28,129 28,129 28,129 28,129
R2 0.11 0.14 0.13 0.16
46
Table 4: Effect of Minority-Owned Banks on Employment Growth
(Overall, Minority, and White)
This table reports regression estimates for analyzing the effects of minority bank market share on employment growth using
OLS estimators for a county–year sample over the period 2006‒2020. We show results using both full sample in models [1],
[3], and [5] and a propensity-score matched sample in models [2], [4], and [6]. The PSM matched sample is limited to
counties either containing minority-owned banks, or counties adjacent to counties containing minority-owned banks.
Counties are matched on the 1990 minority population (%), proportion of workers in the finance industry (%), per capita
income, total population as well as rurality (%). The PSM method limits observations to those on common support, without
replacement. The dependent variables are Δ Overall Employment, Δ Minority Employment, and Δ White Employment,
defined as the year-over-year percent change in the number of employed persons, Overall, Minority, and White, respectively.
The key independent variables are Minority Bank Market Share, the deposit-weighted proportion of branches of minority-
owned banks (≥ 50% minority ownership) in a county, as well as interactions between Minority Bank Market Share with
GFC and COVID-19 Crisis. GFC is an indicator variable equal to one through 2007-2009 and zero otherwise. COVID-19
Crisis is an indicator variable equal to one in 2020, and zero otherwise. We include a broad set of other local market (county)
control variables: Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median
Household Income (Log), COVID-19 Death Rate, and several variables for important industries in the county such as
Manufacturing (%), Arts, Entertainment, and Recreation (%), Accommodation and Food Services (%), and Public
Administration (%). We also include a broad set of bank characteristics aggregated at the local market (county) level based
on banks’ proportions of deposits in each county: Bank Deposits, Assets (Log), Capital Adequacy (%), Asset Quality (%),
Management Quality (%), Earnings (%), Liquidity (%), Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan
Regions (%), Bank Age, BHC Ownership (%), OCC Regulation (%), FDIC Regulation (%), TARP Amount (per Capita), and
PPP Loan Balance (per Capita). All specifications include county and bank control variables lagged by one year, as well as
state and year fixed effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at county
level are reported in parentheses. Significance at the 10%, 5%, and 1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4] [5] [6]
Sample Full Sample PSM Full Sample PSM Full Sample PSM
Δ Overall Δ Overall Δ Minority Δ Minority Δ White Δ White
Dependent Variable:
Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -1.50 -9.39 -14.21 -23.98*** 3.99 -4.50
[-0.25] [-1.47] [-1.46] [-2.87] [0.77] [-0.72]
Minority Bank Market Share × GFC 39.57** 34.25* 111.9*** 63.72** 35.22** 26.67
[2.34] [1.91] [3.69] [2.49] [2.17] [1.51]
Minority Bank Market Share × COVID-19 Crisis 10.96 54.20** 39.27** 79.21*** 7.96 48.80**
[0.68] [2.56] [2.21] [3.16] [0.48] [2.28]
County Controls:
Minority Population -1.54 0.89 -43.40*** -21.17* -3.87* 0.64
[-0.71] [0.10] [-12.67] [-1.91] [-1.81] [0.07]
Working Age -14.97* -9.15 -24.03* 5.21 -17.05** -16.65
[-1.88] [-0.22] [-1.67] [0.11] [-2.20] [-0.41]
Bachelor’s Degree 7.84 59.00** -11.53 43.74 5.38 56.35**
[1.01] [2.10] [-0.83] [1.17] [0.73] [2.07]
Population Density 2.84*** 2.91** 2.61*** 2.39* 2.90*** 3.24***
[5.00] [2.37] [3.61] [1.71] [5.04] [2.91]
Housing Price Index -9.20*** -13.93* -5.18 -17.71* -9.61*** -16.30**
[-4.10] [-1.86] [-1.45] [-1.86] [-4.39] [-2.25]
Median Household Income (Log) 19.35*** 5.75 16.59*** 10.98 18.85*** 6.32
[7.64] [0.65] [3.78] [0.98] [7.62] [0.73]
COVID-19 Death Rate -0.019 -0.18** -0.011 -0.18** 0.026 -0.18***
[-0.61] [-2.50] [-0.23] [-2.08] [0.90] [-2.74]
Manufacturing -14.14*** -8.34 -26.01*** -24.35 -14.06*** -0.24
[-4.73] [-0.49] [-3.91] [-1.01] [-4.98] [-0.02]
Arts, Entertainment, and Recreation -16.34 35.19 -2.11 20.14 -15.13 30.31
[-1.18] [0.59] [-0.09] [0.26] [-1.10] [0.53]
Accommodation and Food Services -2.89 28.09 -23.76** 15.12 -1.87 32.01
[-0.48] [1.10] [-2.24] [0.47] [-0.32] [1.29]
Public Administration -8.06* -18.63 -15.08* -16.60 -9.16** -25.60
[-1.78] [-0.93] [-1.71] [-0.69] [-2.06] [-1.28]
47
Local Bank Controls:
Bank Deposits 6.84 54.59 3.36 46.86 3.25 42.40
[0.93] [0.89] [0.17] [0.53] [0.49] [0.78]
Bank Assets (Log) 0.0090 -0.65 -0.22 0.26 -0.070 -0.92
[0.06] [-0.84] [-0.76] [0.27] [-0.48] [-1.23]
Capital Adequacy -48.25** 12.64 -61.82 29.63 -47.34** 16.27
[-1.97] [0.18] [-0.95] [0.33] [-2.05] [0.23]
Asset Quality -148.1*** 47.69 -317.9*** 32.85 -159.0*** 32.73
[-3.99] [0.17] [-4.62] [0.08] [-4.54] [0.15]
Management Quality 631.1** -612.1* 154.8 -718.1* 607.9** -546.5
[2.25] [-1.77] [0.33] [-1.83] [2.23] [-1.54]
Earnings 292.7*** 339.6 27.55 491.6 278.8*** 221.5
[4.06] [1.28] [0.18] [1.11] [4.09] [0.94]
Liquidity 14.71*** 2.44 16.88* -1.23 15.04*** 10.05
[3.28] [0.16] [1.88] [-0.06] [3.47] [0.69]
Sensitivity to Market Risk -29.78*** -5.80 -33.28*** 6.13 -26.96*** -3.34
[-5.16] [-0.26] [-2.83] [0.20] [-4.84] [-0.16]
HHI Deposits -3.27 -22.76** -4.37 -22.22** -1.25 -19.91**
[-1.17] [-2.55] [-0.83] [-2.15] [-0.46] [-2.33]
Deposits in Metropolitan Regions 3.18*** 8.12 3.96* 4.16 2.73*** 9.81*
[3.07] [1.46] [1.83] [0.57] [2.72] [1.76]
Bank Age -7.08 -1.18 -12.37 -49.94 -2.22 -5.33
[-0.53] [-0.02] [-0.46] [-0.59] [-0.17] [-0.09]
BHC Ownership -4.47*** -9.47* -10.24*** -10.53 -4.28*** -9.22**
[-3.16] [-1.95] [-3.69] [-1.55] [-3.15] [-1.99]
OCC Regulation 1.52 -1.75 3.64 -5.33 1.46 -2.52
[0.86] [-0.26] [1.10] [-0.58] [0.85] [-0.38]
FDIC Regulation 2.03 -6.54 5.04 -9.29 2.01 -7.91
[1.17] [-0.91] [1.45] [-1.04] [1.19] [-1.11]
TARP Amount (per Capita) -793.2 3,090.7 1,658.6 1,766.9 -344.8 6,788.9
[-0.50] [0.30] [1.03] [0.14] [-0.21] [0.60]
PPP Loan Balance (per Capita) -9.15*** -3.55 -10.19*** -3.57 -8.51*** -2.79
[-9.94] [-1.06] [-6.53] [-0.79] [-9.65] [-0.91]
PSM Match Sample No Yes No Yes No Yes
State & Year FEs Yes Yes Yes Yes Yes Yes
Observations 28,129 1,756 28,129 1,756 28,129 1,756
R2 0.26 0.46 0.17 0.34 0.24 0.43
48
Table 5: Effect of Minority-Owned Banks on Employment Growth
(Alternative Samples Using PSM)
This table reports regression estimates for analyzing the effects of minority bank market share on employment growth using OLS estimators for a county–year sample over
the period 2006 to 2020. We present results using a series of alternative samples. The “All Banks” sample includes banks of all sizes in the sample. The < $10B sample
excludes banks with more than $10 billion in assets, and those that are publicly traded. The < $2B sample excludes banks with more than $2 billion in assets, and those that
are publicly traded. We show results using a propensity-score matched sample in all models. The PSM matched sample is limited to counties either containing minority-
owned banks, or counties adjacent to counties containing minority-owned banks. Counties are matched on the 1990 minority population (%), proportion of workers in the
finance industry (%), per capita income, total population as well as rurality (%). The PSM method limits observations to those on common support, without replacement.
The dependent variables are Δ Overall Employment, Δ Minority Employment, and Δ White Employment, defined as the year-over-year percent change in the number of
employed persons, Overall, Minority, and White, respectively. The key independent variables are Minority Bank Market Share, the deposit-weighted proportion of branches
of minority-owned banks (≥ 50% minority ownership) in a county, as well as interactions between Minority Bank Market Share with GFC and COVID-19 Crisis. GFC is
an indicator variable equal to one through 2007-2009 and zero otherwise. COVID-19 Crisis is an indicator variable equal to one in 2020, and zero otherwise. We include a
broad set of other local market (county) control variables: Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household
Income (Log), COVID-19 Death Rate, and several variables for important industries in the county such as Manufacturing (%), Arts, Entertainment, and Recreation (%),
Accommodation and Food Services (%), and Public Administration (%). We also include a broad set of bank characteristics aggregated at the local market (county) level
based on banks’ proportions of deposits in each county: Bank Deposits (per Capita), Assets (Log), Capital Adequacy (%), Asset Quality (%), Management Quality (%),
Earnings (%), Liquidity (%), Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership (%), OCC Regulation (%),
FDIC Regulation (%), TARP Amount (per Capita), and PPP Loans Balance (per Capita). All specifications include county and bank control variables lagged by one year,
as well as state and year fixed effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at county level are reported in parentheses.
Significance at the 10%, 5%, and 1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4] [5] [6] [7] [8] [9]
Sample All Banks < $10 B < $2B All Banks < $10 B < $2B All Banks < $10 B < $2B
Dependent Variable: Δ Overall Employment Δ Minority Employment Δ White Employment
Independent Variables:
Minority Bank Market Share -9.39 -3.27 0.73 -23.98*** -11.48** -3.39 -4.50 0.14 2.46
[-1.47] [-0.78] [0.18] [-2.87] [-2.10] [-0.60] [-0.72] [0.04] [0.68]
Minority Bank Market Share × GFC 34.25* 24.31** 20.61** 63.72** 47.67*** 39.15*** 26.67 19.22 15.37
[1.91] [2.05] [2.13] [2.49] [2.84] [3.10] [1.51] [1.63] [1.55]
Minority Bank Market Share × COVID-19 Crisis 54.20** 25.26* 19.81* 79.21*** 36.71*** 31.64*** 48.80** 22.04* 16.22
[2.56] [1.96] [1.78] [3.16] [2.67] [2.70] [2.28] [1.70] [1.44]
County & Local Bank Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes
PSM Match Sample Yes Yes Yes Yes Yes Yes Yes Yes Yes
State & Year FEs Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 1,756 1,685 1,411 1,756 1,685 1,411 1,756 1,685 1,411
R2 0.46 0.45 0.51 0.34 0.32 0.35 0.43 0.42 0.47
49
Table 6: Effect of Minority-Owned Banks on Employment Growth – IV
This table reports regression estimates for analyzing the effects of minority bank market share on employment growth using IV models for a county–year sample over the
period 2006 to 2020. The IV first stages are shown in models [1]-[3], while the second stages are shown in columns [4]-[6]. The IV is Minority Ancestry (%), the proportion
of residents with ancestors classified as a minority, or interactions of Minority Ancestry (%) with GFC and COVID-19 Crisis. The dependent variables in our main
regressions are Δ Overall Employment, Δ Minority Employment, and Δ White Employment, defined as the year-over-year percent change in the number of employed persons,
Overall, Minority, and White, respectively. The key independent variables are Minority Bank Market Share, the deposit-weighted proportion of branches of minority-owned
banks (≥ 50% minority ownership) in a county, and interactions between Minority Bank Market Share with GFC and COVID-19 Crisis. GFC is an indicator variable equal
to one through 2007‒2009 and zero otherwise. COVID-19 Crisis is an indicator variable equal to one in 2020, and zero otherwise, We include a broad set of other local
market (county) controls: Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household Income (Log), COVID-19
Death Rate, and several variables for important industries in the county such as Manufacturing (%), Arts, Entertainment, and Recreation (%), Accommodation and Food
Services (%), and Public Administration (%). We also include a broad set of bank characteristics aggregated at the local market (county) level based on banks’ proportions
of deposits in each county: Bank Deposits (per Capita), Assets (Log), Capital Adequacy (%), Asset Quality (%), Management Quality (%), Earnings (%), Liquidity (%),
Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership (%), OCC Regulation (%), FDIC Regulation (%), TARP
Amount (per Capita), and PPP Loans Balance (per Capita). All specifications include county and bank control variables lagged by one year, as well as state and year fixed
effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at county level are reported in parentheses. Significance at the 10%, 5%, and
1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4] [5] [6]
IV Stage IV 1st Stage IV 2nd Stage
Minority Bank Share Minority Bank Share × Minority Bank Share × Δ Overall Δ Minority Δ White
Dependent Variable:
GFC COVID-19 Crisis Employment Employment Employment
Independent Variables:
Minority Ancestry 0.10*** -0.00056 0.0018
[2.72] [-0.24] [1.01]
Minority Ancestry × GFC -0.036*** 0.044*** -0.000070
[-2.60] [2.79] [-0.42]
Minority Ancestry × COVID-19 Crisis 0.0015 0.00032 0.079***
[0.12] [1.30] [3.38]
Minority Bank Market Share 0.24** 0.27** 0.25**
[2.48] [2.08] [2.56]
Minority Bank Market Share × GFC 0.32* 1.17** 0.25
[1.72] [2.33] [1.51]
Minority Bank Market Share × COVID-19 Crisis -0.10 0.20 -0.16*
[-1.11] [1.44] [-1.74]
County & Local Bank Controls Yes Yes Yes Yes Yes Yes
State & Year FEs Yes Yes Yes Yes Yes Yes
Observations 28,129 28,129 28,129 28,129 28,129 28,129
R2 0.16 0.05 0.09 -0.06 -0.06 -0.06
F-Statistic 8.20*** 14.79*** 14.12*** 19.29*** 10.50*** 17.39***
50
Table 7: Bank-Level Summary Statistics
This table provides summary statistics for the variables used in our bank-level analysis (bank-year) using Call Reports data. The number of observations is 58,443. Variables
using dollar amounts are expressed in real 2020 dollars using the implicit GDP price deflator.
Full Minority Non-Minority Diff: Minority–
Variable: Sample Mean Bank Mean Bank Mean Non-Minority Std. Dev. p1 p25 p50 p75 p99
Small Business Loan Activity Growth (Total) 11.12 28.91 10.71 18.20** 254.67 -59.39 -11.45 0.53 15.33 203.60
Bank Assets (Log) 19.37 19.30 19.37 -0.06* 1.25 17.17 18.56 19.20 19.95 23.71
Capital Adequacy (%) 10.52 11.03 10.51 0.52*** 2.96 5.24 8.76 9.98 11.63 20.77
Asset Quality (%) 1.26 2.03 1.25 0.78*** 1.94 0.00 0.23 0.64 1.49 9.48
Management Quality (%) 1.08 1.21 1.08 0.13*** 0.42 0.47 0.81 1.01 1.29 2.28
Earnings (%) 0.74 0.41 0.75 0.34*** 1.22 -3.97 0.49 0.88 1.25 2.82
Liquidity (%) 24.76 22.27 24.82 2.55*** 13.95 1.12 14.72 22.64 32.69 65.12
Sensitivity to Market Risk (%) 18.94 22.69 18.86 3.84*** 12.44 0.40 9.20 17.19 26.80 53.24
HHI Deposits 10.33 7.72 10.39 -2.67*** 9.30 0.63 4.33 8.17 12.94 49.13
Deposits in Metropolitan Regions (%) 74.69 91.00 74.32 16.68*** 43.48 0.00 0.00 100.00 100.00 100.00
Bank Age 66.97 34.45 67.73 -33.27*** 45.17 2.00 20.00 74.00 106.00 152.00
BHC Ownership (%) 80.17 62.10 80.59 -18.49*** 39.87 0.00 100.00 100.00 100.00 100.00
OCC Regulation (%) 21.09 25.34 20.99 4.35*** 40.79 0.00 0.00 0.00 0.00 100.00
FDIC Regulation (%) 64.63 63.54 64.65 -1.11 47.81 0.00 0.00 100.00 100.00 100.00
TARP Amount ($1,000,000s) 2.46 4.57 2.42 2.15 102 0.00 0.00 0.00 0.00 0.00
Balance of PPP Loans ($1,000,000s) 6,897.83 5,628.01 6,927.22 -1,299.20 203,933.91 0.00 0.00 0.00 0.00 68,246.63
Minority Population (%) 26.92 60.03 26.15 33.88*** 19.17 2.97 10.85 22.41 39.35 81.95
Working Age (25-64; %) 52.00 53.53 51.97 1.56*** 2.98 42.50 50.49 52.35 53.81 59.05
Bachelor’s Degree (%) 17.42 21.11 17.33 3.78*** 6.39 6.83 12.63 16.61 21.03 36.87
Population Density 40.55 177.13 37.39 139.74*** 165.26 0.39 2.97 8.39 28.65 511.22
Housing Price Index 390.06 644.43 384.17 260.26*** 222.03 134.65 243.52 331.47 457.20 1,283.16
Median Household Income (Log) 11.05 11.05 11.05 0.00 0.20 10.57 10.92 11.04 11.18 11.58
COVID-19 Death Rate 1.58 3.51 1.54 1.97*** 8.58 0.00 0.00 0.00 0.00 41.48
Manufacturing (%) 13.63 7.78 13.76 -5.99*** 8.69 1.26 7.46 11.28 18.15 41.38
Arts, Entertainment, and Recreation (%) 1.58 1.74 1.58 0.17*** 1.13 0.00 0.99 1.41 1.89 5.58
Accommodation and Food Services (%) 9.90 9.76 9.91 -0.15 3.43 0.00 8.19 9.48 11.17 22.87
Public Administration (%) 6.02 6.02 6.02 0.00 4.06 0.96 3.72 5.00 7.00 23.24
51
Table 8: Effect of Bank Minority-Ownership Status on Small Business Lending Growth
(FFIEC Call Reports Data)
This table reports regression estimates for analyzing the effects of bank minority ownership status on small business lending
growth using OLS models for a bank–year FFIEC Call Report sample over the period 2006 to 2020. We show results using
models with different controls and fixed effects, where models [3] and [4] contain the most complete specifications. The
dependent variable is Δ Small Business Lending, defined as the year-over-year percent change in the bank small business
lending based on the FFIEC Call Reports. The key independent variables are Minority Bank, an indicator for minority-owned
banks (≥ 50% minority ownership), and interactions between Minority Bank with GFC and COVID-19 Crisis. GFC is an
indicator variable equal to one through 2007-2009 and zero otherwise. COVID-19 Crisis is an indicator variable equal to
one in 2020, and zero otherwise. In Model [4] we provide further interactions of each of these terms with High Minority
Area, an indicator variable equal to one if the bank branch network minority population exceeds the national average, and
zero otherwise. We include a broad set of other local market (county) controls: Minority Population (%), Working Age (%),
Bachelor’s Degree (%), Population Density, HPI, Median Household Income (Log), COVID-19 Death Rate, and several
variables for important industries in the county such as Manufacturing (%), Arts, Entertainment, and Recreation (%),
Accommodation and Food Services (%), and Public Administration (%). We also include a broad set of bank characteristic
controls: Assets (Log), Capital Adequacy (%), Asset Quality (%), Management Quality (%), Earnings (%), Liquidity (%),
Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership, OCC
Regulation, FDIC Regulation, TARP Amount, and PPP Loans Balance. Specifications include county and/or bank control
variables lagged by one year as indicated, as well as year fixed effects. All variables are defined in Table 1.
Heteroskedasticity-robust t-statistics clustered at bank level are reported in parentheses. Significance at the 10%, 5%, and
1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4]
Δ Small Δ Small Δ Small Δ Small
Dependent Variable: Business Business Business Business
Lending Lending Lending Lending
Independent Variables:
Minority Bank -1.89 -7.96*** -7.07** 2.06
[-1.05] [-2.73] [-2.41] [0.48]
Minority Bank × GFC 12.50** 13.10** 13.73** -6.90
[2.14] [2.28] [2.40] [-1.11]
Minority Bank × COVID-19 Crisis 193.1 199.8 130.6 48.69
[1.53] [1.60] [1.05] [1.10]
High Minority Area 5.30
[1.27]
Minority Bank × High Minority Area -9.46*
[-1.77]
High Minority Area × GFC -0.41
[-0.25]
Minority Bank × High Minority Area × GFC 21.34**
[2.42]
High Minority Area × COVID-19 Crisis -21.67
[-0.41]
Minority Bank × High Minority Area × COVID-19 Crisis 96.18
[0.69]
Bank Controls No Yes Yes Yes
County Controls No No Yes Yes
Year FE Yes Yes Yes Yes
Observations 58,443 58.443 58,443 58,443
R2 0.010 0.032 0.040 0.040
52
Table 9: Effect of Bank Minority-Ownership Status on Small Business Origination Growth
(Community Reinvestment Act (CRA) Data)
This table reports regression estimates for analyzing the effects of bank minority ownership status on small business lending growth using OLS models for a bank-county-
year sample using the Community Reinvestment Act (CRA) data over the period 2006‒2020. We show results using models for overall small business lending growth in
models [1]-[2], as well as models with loans of different sizes in models [3]-[8]. The dependent variables are Δ Overall Small Business Lending, Δ < $100K Small Business
Lending, Δ $100-250K Small Business Lending, and Δ $250K-1M Small Business Lending, defined as the year-over-year percent change in the new bank small business
originations based on the CRA data. The key independent variables are Minority Bank, an indicator for minority-owned banks (≥ 50% minority ownership), and interactions
between Minority Bank with GFC and COVID-19 Crisis. GFC is an indicator variable equal to one through 2007-2009 and zero otherwise. COVID-19 Crisis is an indicator
variable equal to one in 2020, and zero otherwise. In Models [2], [4], [6] and [8], we provide further interactions of each of these terms with High Minority Area, an
indicator variable equal to one if the bank branch network minority population exceeds the national average, and zero otherwise. We include a broad set of other local
market (county) controls: Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household Income (Natural Log), COVID-
19 Death Rate, and several variables for important industries in the county such as Manufacturing (%), Arts, Entertainment, and Recreation (%), Accommodation and Food
Services (%), and Public Administration (%). We also include a broad set of bank characteristics: Assets (Log), Capital Adequacy (%), Asset Quality (%), Management
Quality (%), Earnings (%), Liquidity (%), Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership, OCC Regulation,
FDIC Regulation, TARP Amount, and PPP Loans Balance. Specifications include county and bank control variables lagged by one year as indicated, as well as county and
year fixed effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at county level are reported in parentheses. Significance at the 10%,
5%, and 1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4] [5] [6] [7] [8]
Δ Overall Δ Overall Δ < $100K Δ < $100K Δ $100-250K Δ $100-250K Δ $250K-1M Δ $250K-1M
Small Small Small Small Small Small Small Small
Business Business Business Business Business Business Business Business
Dependent Variable: Origination Origination Origination Origination Origination Origination Origination Origination
Independent Variables:
Minority Bank -4.52*** -3.52*** -5.26*** -4.45*** -0.50*** -0.44*** -0.17*** -0.16**
[-8.52] [-5.76] [-10.32] [-7.04] [-9.44] [-3.23] [-4.73] [-2.10]
Minority Bank × GFC 1.46** 0.078 1.61*** -0.074 0.17* 0.064 0.011 -0.18
[2.19] [0.09] [3.72] [-0.11] [1.95] [0.29] [0.13] [-0.99]
Minority Bank × COVID-19 Crisis 2.63*** 0.16 4.80*** 0.10 0.83** -0.99*** -0.48*** -0.85***
[3.77] [0.12] [7.66] [0.08] [2.52] [-3.95] [-6.55] [-7.91]
High Minority Area -1.36 -0.70 -0.052 -0.078**
[-1.57] [-1.04] [-1.48] [-2.27]
Minority Bank × High Minority Area -1.25* -1.01* -0.054 -0.020
[-1.96] [-1.85] [-0.38] [-0.25]
High Minority Area × GFC -1.39*** -0.83*** -0.10*** -0.086***
[-4.20] [-4.21] [-3.77] [-2.68]
Minority Bank × High Minority Area × GFC 2.23* 2.36*** 0.16 0.27
[1.88] [2.69] [0.67] [1.28]
High Minority Area × COVID-19 Crisis -1.03*** -1.12*** 0.10*** 0.012
[-2.98] [-4.37] [2.61] [0.40]
Minority Bank × High Minority Area × COVID-19 Crisis 3.13** 5.49*** 1.86*** 0.40***
[2.25] [3.88] [4.39] [2.98]
Bank & County Controls Yes Yes Yes Yes Yes Yes Yes Yes
County & Year FE Yes Yes Yes Yes Yes Yes Yes Yes
Observations 621,449 621,449 537,501 537,501 236,012 236,012 242,076 242,076
R2 0.003 0.003 0.004 0.004 0.027 0.027 0.016 0.016
53
Table 10: Effect of Bank Minority-Ownership Status on Household Lending Origination
(Home Mortgage Disclosure Act (HMDA) Data)
This table reports regression estimates for analyzing the effects of minority bank market share on household (HH) lending origination in
the mortgage market using OLS models for a bank–county-year sample from the Home Mortgage Disclosure Act (HMDA) data over the
period 2006‒2020. We show results using models for mortgage approvals, originations, and mortgage interest rate over 2006‒2020 in
models [1]-[4] and 2018-2020 in model [5], the latter focusing on the COVID-19 crisis only and noting that interest rates only start to be
reported from 2018 onward. The dependent variables are Approved, an indicator that equals 1 if a loan application was approved by the
lender (action_type = 1 or 2), and 0 if it was denied (action_type = 3), Originated is an indicator that equals 1 if a loan application was
originated by the lender (action_type = 1), and 0 if it was denied (action_type = 3), and Interest Rate is the interest rate offered by the
lender on approved mortgages. The key independent variables are Minority Bank, an indicator for minority-owned banks (≥ 50% minority
ownership), GFC, an indicator variable equal to one through 2007‒2009 and zero otherwise, COVID-19 Crisis, an indicator variable equal
to one in 2020, and zero otherwise, and interactions between Minority Bank with GFC and COVID-19 Crisis and further interactions of
each of these terms with Minority HH, an indicator variable equal to one if a household is minority. We include a set of standard HMDA
borrower characteristics at loan application time are: Female, indicator for female applicants, Metro, indicator for metropolitan areas,
Ln(Loan Amount), and Ln(Loan Amount) Sq, the natural logarithm of the loan amount that the applicant requested and its squared term,
Co-Applicant, indicator for the presence of a coapplicant on the application, Ln(Applicant Income), the natural logarithm of the applicant
income, and Debt-to-Income, the ratio of debt to income of the applicant. We include a broad set of other local market (county) controls:
Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household Income (Natural Log),
COVID-19 Death Rate, and several variables for important industries in the county such as Manufacturing (%), Arts, Entertainment, and
Recreation (%), Accommodation and Food Services (%), and Public Administration (%). We also include a broad set of bank
characteristics: Assets (Log), Capital Adequacy (%), Asset Quality (%), Management Quality (%), Earnings (%), Liquidity (%), Sensitivity
to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership, OCC Regulation, FDIC Regulation,
TARP Amount (Log), and PPP Loans Balance (Log). Specifications include county and bank control variables lagged by one year as
indicated, as well as state and year fixed effects. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at
county level are reported in parentheses. Significance at the 10%, 5%, and 1% level is indicated by *, **, and ***, respectively.
Model [1] [2] [3] [4] [5]
Dependent Variable: Approved Approved Originated Originated Interest Rate
Independent Variables:
Minority Bank 0.027** 0.038*** 0.019 0.029** 0.131
[2.27] [3.26] [1.39] [2.16] [1.55]
Minority Bank × GFC 0.086*** 0.076*** 0.115*** 0.105***
[5.01] [4.72] [5.89] [5.60]
Minority Bank × COVID-19 Crisis 0.009 -0.020 0.019 -0.022 0.641***
[0.51] [-1.19] [0.96] [-1.01] [9.47]
Minority HH -0.036*** -0.033*** -0.037*** -0.033*** -0.156***
[-7.36] [-6.52] [-7.21] [-6.06] [-13.15]
Minority HH × GFC -0.053*** -0.079***
[-7.32] [-7.01]
Minority HH × COVID-19 Crisis 0.008*** 0.006* 0.042***
[2.74] [1.68] [3.95]
Minority Bank × Minority HH -0.037* -0.037* 0.020
[-1.86] [-1.66] [0.36]
Minority Bank × Minority HH × GFC 0.031 0.032
[1.12] [0.91]
Minority Bank × Minority HH × COVID-19 Crisis 0.069** 0.091*** -0.216***
[2.32] [2.70] [-3.14]
Female 0.005*** 0.006*** 0.006*** 0.006*** -0.026***
[4.15] [4.23] [4.04] [4.14] [-4.57]
Metro 0.032*** 0.032*** 0.037*** 0.037*** -0.029*
[7.07] [7.10] [7.17] [7.21] [-1.78]
Ln(Loan Amount) 0.066*** 0.067*** 0.040*** 0.041*** 0.081
[7.07] [7.10] [4.10] [4.14] [0.71]
Ln(Loan Amount)2 0.001 0.001 0.003*** 0.003*** -0.049***
[0.94] [0.92] [3.20] [3.17] [-5.16]
Co-Applicant 0.006*** 0.006*** 0.007*** 0.007*** 0.013***
[3.79] [3.82] [4.11] [4.15] [3.15]
Ln(Applicant Income) 0.045*** 0.045*** 0.053*** 0.053*** 0.209***
[13.57] [13.67] [15.36] [15.48] [7.56]
Debt-to-Income -0.036*** -0.036*** -0.034*** -0.034*** 0.001
[-18.11] [-18.29] [-16.85] [-17.06] [0.12]
Bank & County Controls Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes
Observations 3,034,722 3,034,722 2,776,079 2,776,079 765,827
R2 0.10 0.10 0.12 0.12 0.35
54
Appendix S: Supplementary Robustness Tests and Analyses
Table S1: Effects of Minority-Owned Banks on Employment Growth – Robustness Checks
(Overall, Minority, and White)
This table reports regression estimates for analyzing the effects of minority bank market share on employment growth using
OLS models for a county–year sample over the period 2006‒2020. We show results using several robustness checks. Panel
A shows results using an alternative GFC definition; the GFC is defined as an indicator variable equal to one through 2008-
2009 and zero otherwise. Panel B uses an alternative definition for minority-owned bank, a Minority Depository Institution
(MDI) based on the FDIC (2021) definition. MDIs are defined according to two requirements: (1) the majority of the voting
stock is owned by minorities or (2) the board of directors is comprised predominately of minorities and the community the
bank serves is predominantly minority. Panel C shows results using an alternative specification for our main propensity
score matching (PSM) technique, where we exclude White Women from the PSM control group. Panel D shows results
using an alternative propensity score matching (PSM) technique. The PSM matched samples are limited to counties either
containing minority-owned banks, or counties adjacent to counties containing minority-owned banks. Counties matched on
the 1990 minority population (%), proportion of workers in the finance industry (%), per capita income, total population and
rurality (%). Specification 1, used in our main results, uses one neighbor, without replacement on common support.
Specification 2 uses three neighbors, with replacement.
The dependent variables are Δ Overall Employment, Δ Minority Employment, and Δ White Employment, defined as the year-
over-year percent change in the number of employed persons, overall, minority, and white, respectively. The key
independent variables are Minority Bank Market Share, the deposit-weighted proportion of branches of minority-owned
banks (≥ 50% minority ownership) in a county, GFC, an indicator variable equal to one through 2007-2009 and zero
otherwise, COVID-19 Crisis, an indicator variable equal to one in 2020, and zero otherwise, and interactions between
Minority Bank Market Share with GFC and COVID-19 Crisis. We include a broad set of other local market (county):
Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household Income
(Natural Log), Male (%), COVID-19 Death Rate and several variables for important industries in the county such as
Manufacturing (%), Arts, Entertainment, and Recreation (%), Accommodation and Food Services (%), and Public
Administration (%). We also include a broad set of bank characteristics aggregated at the local market (county) level based
on banks’ proportions of deposits in each county: Bank Deposits, Capital Adequacy (%), Asset Quality (%), Management
Quality (%), Earnings (%), Liquidity (%), Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%),
Bank Age, BHC Ownership(%), OCC Regulation (%), FDIC Regulation (%), Total Loans to Assets (%), TARP Amount (per
Capita), and PPP Loans Balance (per Capita). Specifications include county and bank control variables lagged by one year,
as well as state and year fixed effects, unless specified otherwise. All variables are defined in Table 1. Heteroskedasticity-
robust t-statistics clustered at county level are reported in parentheses unless specified otherwise. Significance at the 10%,
5%, and 1% level is indicated by *, **, and ***, respectively.
Panel S1.A: Alternative GFC Definition
Model [1] [2] [3] [4] [5] [6]
Δ Overall Δ Overall Δ Minority Δ Minority Δ White Δ White
Dependent Variable:
Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.00150 -0.00939 -0.0142 -0.0240*** 0.00400 -0.00450
[-0.25] [-1.47] [-1.46] [-2.87] [0.77] [-0.72]
Minority Bank Market Share × GFC 0.0396** 0.0342* 0.112*** 0.0637** 0.0352** 0.0267
[2.34] [1.91] [3.69] [2.49] [2.17] [1.51]
Minority Bank Market Share × COVID-19 Crisis 0.0110 0.0542** 0.0393** 0.0792*** 0.00797 0.0488**
[0.68] [2.56] [2.21] [3.16] [0.48] [2.28]
PSM Match Sample No Yes No Yes No Yes
County & Bank Controls Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes
Observations 28,129 1,756 28,129 1,756 28,129 1,756
R2 0.26 0.46 0.17 0.34 0.24 0.43
55
Table S1.B: Alternative Minority Bank Definition: MDI Bank
Model [1] [2] [3] [4] [5] [6]
Δ Overall Δ Overall Δ Minority Δ Minority Δ White Δ White
Dependent Variable:
Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.000829 -0.00834 -0.00975 -0.0167* 0.00396 -0.00535
[-0.13] [-1.19] [-0.99] [-1.84] [0.71] [-0.80]
Minority Bank Market Share × GFC 0.0517*** 0.0440*** 0.108*** 0.0637*** 0.0467*** 0.0357**
[4.21] [3.22] [6.78] [2.79] [3.60] [2.42]
Minority Bank Market Share × COVID-19 Crisis 0.00540 0.0472** 0.0379** 0.0755*** -0.000987 0.0376*
[0.34] [2.14] [2.21] [2.96] [-0.06] [1.75]
PSM Match Sample No Yes No Yes No Yes
County & Bank Controls Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes
Observations 28,129 1,756 28,129 1,756 28,129 1,756
R2 0.26 0.46 0.17 0.34 0.24 0.43
Table S1.C: PSM Analysis: Excluding White Women in the Control Group
Model [1] [2] [3]
Sample PSM PSM PSM
Δ Overall Δ Minority Δ White
Dependent Variable:
Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.00939 -0.0240*** -0.00450
[-1.47] [-2.87] [-0.72]
Minority Bank Market Share × GFC 0.0342* 0.0637** 0.0267
[1.91] [2.49] [1.51]
Minority Bank Market Share × COVID-19 Crisis 0.0542** 0.0792*** 0.0488**
[2.56] [3.16] [2.28]
Women Excluded in PSM Control Yes Yes Yes
County & Bank Controls Yes Yes Yes
Observations 1,756 1,756 1,756
R2 0.46 0.34 0.43
Table S1.D: Alternative PSM Specifications
Model [1] [2] [3] [4] [5] [6]
Sample [PSM Type] PSM1 PSM2 PSM1 PSM2 PSM1 PSM2
Δ Overall Δ Overall Δ Minority Δ Minority Δ White Δ White
Dependent Variable:
Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.00939 -0.00894 -0.0240*** -0.0231*** -0.00450 -0.00413
[-1.47] [-1.42] [-2.87] [-2.82] [-0.72] [-0.67]
Minority Bank Market Share × GFC 0.0342* 0.0331* 0.0637** 0.0622** 0.0267 0.0261
[1.91] [1.86] [2.49] [2.46] [1.51] [1.49]
Minority Bank Market Share × COVID-19 Crisis 0.0542** 0.0514** 0.0792*** 0.0762*** 0.0488** 0.0460**
[2.56] [2.47] [3.16] [3.08] [2.28] [2.18]
County & Bank Controls Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes
Observations 1,756 1,808 1,756 1,808 1,756 1,808
R2 0.46 0.46 0.34 0.34 0.43 0.43
56
Table S2: Effects of Minority-Owned Banks on Employment Growth –
Conditioning on Racial Animosity Variables
(Overall, Minority, and White)
These tables report regression estimates for analyzing the effects of minority bank market share on employment growth
using OLS models for a county‒year sample over the period 2006‒2020. We show results for the full sample when
conditioning on racial animosity variables: Fair Housing Law, Interracial Marriage, Racial Bias Index, Slave State, Racial
Animosity, Black Lives Matter (BLM) Indicator. Panel A shows definitions for the additional animosity variables, while
Panels B-D reports results for the overall, minority, and White employment growth, respectively when conditioning on
these additional animosity variables.
The dependent variables are Δ Overall Employment, Δ Minority Employment, and Δ White Employment, defined as the year-
over-year percent change in the number of employed persons, overall, minority, and White, respectively. The key
independent variables are Minority Bank Market Share, the deposit-weighted proportion of branches of minority-owned
banks (≥ 50% minority ownership) in a county, GFC, an indicator variable equal to one through 2007‒2009 and zero
otherwise, COVID-19 Crisis, an indicator variable equal to one in 2020, and zero otherwise, and interactions between
Minority Bank Market Share with GFC and COVID-19 Crisis. We include a broad set of other local market (county):
Minority Population (%), Working Age (%), Bachelor’s Degree (%), Population Density, HPI, Median Household Income
(Natural Log), COVID-19 Death Rate, and several variables for important industries in the county such as Manufacturing
(%), Arts, Entertainment, and Recreation (%), Accommodation and Food Services (%), and Public Administration (%). We
also include a broad set of bank characteristics aggregated at the local market (county) level based on banks’ proportions of
deposits in each county: Bank Deposits, Capital Adequacy (%), Asset Quality (%), Management Quality (%), Earnings (%),
Liquidity (%), Sensitivity to Market Risk (%), Bank HHI, Deposits in Metropolitan Regions (%), Bank Age, BHC Ownership
(%), OCC Regulation (%), FDIC Regulation (%), TARP Amount (per Capita), and PPP Loans Balance (per Capita).
Specifications include county and bank control variables lagged by one year, as well as state and year fixed effects, unless
specified otherwise. All variables are defined in Table 1. Heteroskedasticity-robust t-statistics clustered at county level are
reported in parentheses unless specified otherwise. Significance at the 10%, 5%, and 1% level is indicated by *, **, and ***,
respectively.
Panel S2.A: Definitions for Additional Racial Animosity Variables
Variable Definition Source
Fair Housing Law Equal to one if the state did not curb discriminatory practices in Célerier and Matray (2019);
housing until the Fair Housing Act of 1968, and zero otherwise. Collins (2004)
Interracial Marriage Equal to one if interracial marriage was banned in the state in 1967, Célerier and Matray (2019);
and zero otherwise. Fryer (2007)
Racial Bias Index Equal to one if the state is above the median for interracial marriage Célerier and Matray (2019);
bias, and zero otherwise. Levine, Rubinstein, and
Levkov (2014)
Slave State Equal to one if slavery was legal in the state at the beginning of Célerier and Matray (2019)
Civil War (1861), and zero otherwise.
Racial Animosity We use principal component analysis to construct an index of the
Index above four racial animosity variables.
Black Lives Matter An indicator variable equal to one if the county experienced a Black Social Explorer (2020)
(BLM) Protests Lives Matter (BLM) protest between May 29 and June 9 2020.
57
Table S2.B: Effects on Overall Employment Growth, Conditioning on Racial Animosity
[1] [2] [3] [4] [5] [6] [7]
Δ Overall Δ Overall Δ Overall Δ Overall Δ Overall Δ Overall Δ Overall
Dependent Variable:
Employment Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.00656 -0.00509 -0.00312 -0.00884 -0.000834 -0.00312 -0.00618
[-1.12] [-0.84] [-0.51] [-1.51] [-0.15] [-0.51] [-1.04]
Minority Bank Market Share × GFC 0.0389** 0.0388** 0.0393** 0.0385** 0.0395** 0.0393** 0.0391**
[2.36] [2.35] [2.33] [2.35] [2.37] [2.33] [2.38]
Minority Bank Market Share × COVID-19 Crisis 0.0123 0.0118 0.0106 0.0124 0.0108 0.0106 0.0119
[0.77] [0.74] [0.67] [0.77] [0.67] [0.67] [0.75]
Fair Housing Law 0.00214***
[3.83]
Interracial Marriage 0.00469***
[6.81]
Racial Bias Index 0.00360***
[7.12]
Slave State 0.00437***
[6.43]
Racial Animosity 0.00234***
[6.81]
BLM Indicator -0.00158***
[-2.80]
County & Bank Controls Yes Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes Yes
Observations 28,129 28,129 28,129 28,129 28,129 28,129 28,129
R2 0.24 0.24 0.25 0.25 0.25 0.25 0.24
58
Table S2.C: Effects on Minority Employment Growth, Conditioning on Racial Animosity
Model [1] [2] [3] [4] [5] [6] [7]
Δ Minority Δ Minority Δ Minority Δ Minority Δ Minority Δ Minority Δ Minority
Dependent Variable:
Employment Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.000450 -0.00412 -0.00445 -0.00357 -0.00692 -0.00445 -0.000460
[-0.04] [-0.36] [-0.39] [-0.29] [-0.56] [-0.39] [-0.04]
Minority Bank Market Share × GFC 0.111*** 0.111*** 0.110*** 0.110*** 0.110*** 0.110*** 0.111***
[3.52] [3.54] [3.56] [3.52] [3.53] [3.56] [3.52]
Minority Bank Market Share × COVID-19 Crisis 0.0389** 0.0403** 0.0409** 0.0389** 0.0406** 0.0409** 0.0389**
[2.26] [2.34] [2.39] [2.25] [2.38] [2.39] [2.26]
Fair Housing Law -0.00534***
[-4.60]
Interracial Marriage -0.00546***
[-4.17]
Racial Bias Index 0.00493***
[4.81]
Slave State -0.00494***
[-3.87]
Racial Animosity -0.00273***
[-4.17]
BLM Indicator 0.0000408
[0.04]
County & Bank Controls Yes Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes Yes
Observations 28,129 28,129 28,129 28,129 28,129 28,129 28,129
R2 0.16 0.16 0.16 0.16 0.16 0.16 0.16
59
Table S2.D: Effects on White Employment Growth, Conditioning on Racial Animosity
Model [1] [2] [3] [4] [5] [6] [7]
Δ White Δ White Δ White Δ White Δ White Δ White Δ White
Dependent Variable:
Employment Employment Employment Employment Employment Employment Employment
Independent Variables:
Minority Bank Market Share -0.00177 -0.000149 0.00181 -0.00396 0.00424 0.00181 -0.00134
[-0.34] [-0.03] [0.33] [-0.76] [0.83] [0.33] [-0.26]
Minority Bank Market Share × GFC 0.0342** 0.0341** 0.0346** 0.0338** 0.0348** 0.0346** 0.0345**
[2.15] [2.13] [2.13] [2.14] [2.17] [2.13] [2.16]
Minority Bank Market Share × COVID-19 Crisis 0.00919 0.00858 0.00744 0.00923 0.00756 0.00744 0.00866
[0.57] [0.53] [0.46] [0.57] [0.46] [0.46] [0.54]
Fair Housing Law 0.00237***
[4.34]
Interracial Marriage 0.00489***
[7.28]
Racial Bias Index 0.00345***
[6.97]
Slave State 0.00459***
[6.91]
Racial Animosity 0.00244***
[7.28]
BLM Indicator -0.00180***
[-3.31]
County & Bank Controls Yes Yes Yes Yes Yes Yes Yes
State & Year FE Yes Yes Yes Yes Yes Yes Yes
Observations 28,129 28,129 28,129 28,129 28,129 28,129 28,129
R2 0.23 0.23 0.23 0.23 0.23 0.23 0.23
60
Figure S1: Propensity Score Matching (PSM) Diagnostics
This figure shows standardized % bias across key covariates of treatment-group observations versus control-group
observations after the propensity score matching method was employed.
61
File and source
- File
- berger-2023-let-us-put-our-moneys-together-minority-owned-banks-and-resilience-to-crises.pdf
- Size
- 1,527,845 bytes
- SHA-256
- 957dfedc93b1895c83f48cc6b078a6953ef07c0b757caa6dcff3f1f839036a99
- Our copy
- berger-2023-let-us-put-our-moneys-together-minority-owned-banks-and-resilience-to-crises.pdf
- Original
- doi.org