AFRICAN AMERICAN REPRESENTATION ON THE BOARDS OF
BANKS AND MORTGAGE LOAN REJECTION RATES
Valentin Dimitrov
a
, Jiayin Li
b
and Darius Palia
a
`
February 2024
Abstract
With growing social activism and changing regulations, stakeholders are urging boards to
better reflect their firms’ diverse customers. Given the pivotal role banks play in home ownership
and wealth accumulation, we study the impact of African American directors on banks’ mortgage
rejection rates of African American applicants. We document that banks that have at least one
African American director are typically larger, have larger boards, a higher percentage of female
directors, and more African American executives within the headquarter state - all compared to
their counterparts without African American directors. Significantly, we find that banks with at
least one African American director have lower rejection rates for African American mortgage
applicants. We establish this relationship using three methods: 2SLS, two matching methods, and
by examining the effects of appointing a new African American director to the board. We also find
that a large majority of African American directors sit on key board committees which helps lower
borrower rejection rates. Finally, we find no evidence that diversifying the board comes at a cost
to the banks’ shareholders.
_____________________________________________________________________________________
a
Rutgers Business School and
b
University of International Business and Economics, respectively. We thank Yakov
Amihud, Divya Anantharaman, Neil Bhutta, Ivan Brick, Jennifer Carpenter, N.K. Chidambaran, Morris Davis, Vadim
Elenev, Suresh Govindaraj, Prem C. Jain, Kose John, David Matsa, Oded Palmon, Thomas Philippon, N. Prabhala,
Rajesh Narayanan, Stijn Van Nieuwerburgh, Adi Sundaram, Lukai Yang, David Yermack, Adam Yore, and seminar
participants at Florida State, John Hopkins, Rutgers, NYU Stern, NYC Real Estate Conference (Columbia), and the
NYCU International Finance Conference (Taiwan) for valuable comments and suggestions. Corresponding author:
Darius Palia, dp[email protected].
2
“While the calls for more board diversity apply to all kinds of public companies, banks are
arguably under the most pressure because they can play such a crucial role in closing the racial
wealth gap…”
American Banker, December 7, 2020
1
Starting July 1, 2020, Goldman Sachs will not underwrite companies to go public who do not have
at least one diverse director and at least two diverse directors in 2021.
David Solomon, CEO, on Goldman Sachs’ Commitment to Board Diversity
In late 2020, NASDAQ companies must appoint at least two diverse directors.
NASDAQ Press Release, December 1, 2020
1. Introduction
As social activism intensifies and regulations change, corporate stakeholders are
increasingly advocating for boards to reflect their diverse customer base and to take active steps
to address racial inequities. This paper investigates whether having an African American bank
director impacts the probability of obtaining a home mortgage loan for African American
borrowers. Additionally, we explore factors associated with the appointment of an African
American bank director, and assess the impact of having such a director on future bank
performance and risk.
Significant economic inequalities continue to persist between races in the United States.
The average per capita net worth for Whites is approximately $437,000, compared to a much lower
$105,000 for African Americans and $53,000 for Hispanics.
2
A large part of this difference can be
attributed to considerably higher home ownership rates for Whites (73.3%), compared to African
Americans (42.1%) and Hispanics (47.5%).
3
Given the importance of real estate, fair and equal
1
https://www.americanbanker.com/news/protests-and-mandates-push-banks-to-add-more-minorities-to-boards. See
Section 2.1 of this paper for more references and details of initiatives to increase the diversity of boards.
2
Federal Reserve: Distributional Financial Accounts.
3
U.S. Census Bureau 2019 Housing Vacancies and Homeownership Survey. Hubbard(1985), Campbell(2006),
Cochrane(2007) propose that real estate is an illiquid/nontraded asset that has a significant effect on a household’s
3
access to housing has been a priority among regulators.
4
Prior empirical studies have examined
whether African Americans and Hispanics are systematically discriminated against by mortgage
lenders (for example, Black et al., 1978; King, 1980; Munnell et al., 1996; Hubbard et al., 2011;
Quillian et al., 2020; Giacolleti et al. 2021). In a recent paired testing study by the newspaper
Newsday, equally matched White and minority testers posing as homebuyers approached real
estate agents in the liberal bastion of Long Island, New York (Ann Choi, 2019). In 49% of cases,
African American testers seemed to have received disparate treatment when compared to White
testers, whereas 39% of Hispanics received disparate treatment when compared to White testers.
Using a more comprehensive sample across five cities, a HUD (2012) study finds similar disparate
treatments between the races. Ross et al. (2008) find that African American and Hispanic testers
in Chicago received less information and assistance from their mortgage lending institutions when
compared to similar White borrowers. Martinez and Glantz (2018) analyze 31 million mortgage
records in the period 2015 and 2016. They find that African Americans, Hispanics, and Asians had
less access to conventional mortgages than similar White applicants in 45, 25, and nine metros,
respectively.
One potential way to ameliorate these stark inequalities is to have the boards of financial
institutions direct their companies to increase lending to minority home borrowers. However,
according to a 2018 survey of Fortune 500 companies, 80.5% of board seats are held by Whites,
portfolio choice and asset returns. Empirical evidence of such effects has been found in Flavin and Yamashita(2002),
and Palia, Qi and Wu(2014).
4
The Fair Housing Act of 1968 prohibits discrimination based on race and is actively enforced by the Office of Fair
Housing and Equal Opportunity in the U.S. Department of Housing and Urban Development (HUD). The Home
Mortgage Disclosure Act (HMDA) of 1975 requires financial institutions to maintain, report, and publicly disclose
loan-level information about mortgages. Public data are modified to protect applicant and borrower privacy, while
providing regulators and researchers data that might shed light on discriminatory lending patterns. The Community
Reinvestment Act (CRA) of 1977 encourages financial institutions to help meet the credit needs of the communities
in which they do business, including low- and moderate-income neighborhoods.
4
11.1% are held by African Americans, and 4.4% are held by Hispanics.
5
In light of these statistics,
the NASDAQ, SEC, state legislatures of California, Hawaii, Illinois, Massachusetts, Michigan,
New York and Washington, Congress, pension funds (e.g., NYC pension fund), proxy advisors
(e.g., Institutional Shareholder Services and Glass Lewis), and underwriters of companies going
public (e.g., Goldman Sachs) have increasingly put pressure on corporate America to diversify its
boards.
6
These institutions are encouraging, and more recently mandating, that companies appoint
female directors and directors from underrepresented minority groups such as African Americans
and Hispanics.
Having minorities on bank boards may increase lending to minorities for two reasons. First,
a more diverse board may be more sensitive to existing inequalities in mortgage lending, and to
the unique needs and circumstances of minority borrowers. Second, a more diverse board may
signal the firm’s commitment to diversity, inclusion, and social justice (Ferreira, 2010; Lamkin
Broome & Krawiec, 2008). Alternatively, having minorities on the board may not increase lending
to minorities if board diversity is used to ‘whitewash’ existing discrimination. For example, board
diversity may be used as a false signal to divert attention from existing discriminatory lending
practices (Brown et al., 2012).
7
In this paper, we focus on African American directors, given that there are too few directors
from other minority groups in our sample. For African American board directors, we examine the
following five related questions:
5
Missing Pieces Report: The 2018 Board Diversity Census of Women and Minorities on Fortune 500 Boards, Deloitte
and the Alliance for Board Diversity.
6
See Section 2.1 of this paper for details on these initiatives.
7
Consider the case of Wells Fargo, which received rewards for having a diverse board while facing lawsuits for
preying on minority borrowers. See https://www.cnbc.com/2017/05/04/wells-fargo-amid-minority-targeting-
lawsuits-wins-diversity-award.html.
5
(1) We are the first paper to examine the determinants of having an African American on
the board of directors of a bank. Prior research has shown that the board composition is not random
(see, for e.g., Boone et al., 2007; Coles et al., 2008; Linck et al., 2008). We provide ex ante testable
hypotheses for bank characteristics that might correlate with whether or not a bank has an African
American board member.
(2) We are the first paper to examine the impact of African American bank board
representation on the rejection rates faced by African American mortgage applicants. We do so
using three estimation approaches. First, we estimate a two-stage least squares (2SLS) model,
using the percentage of African American executives in the bank's headquartered state as an
instrument variable (IV) for African American bank board representation. Our dependent variable
is the rejection rate of African American mortgage applicants excluding applicants from the state
where the bank is headquartered. The supply of qualified African American executives has been
shown to be an important determinant of board membership in the literature on non-bank firms
(Giannetti & Zhao, 2019; Glass & Cook, 2017; Knyazeva et al., 2013; Rose & Bielby, 2011).
Giannetti & Zhao (2019) argue that firms choose their headquarters locations early in their life
cycle, while they select directors in response to specific challenges and opportunities they
encounter. Second, we estimate two matching models, Kernel matching and Nearest Neighbor
matching. When there are significant differences in the means and variances of the covariates in
the treated and control groups, as is the case in our sample, matching models provide more efficient
and unbiased inferences (Imbens, 2004; Rosenbaum, 2002; Rubin, 2006; Stuart, 2010). Third, we
examine changes in rejection rates when a new African American director is appointed to the board.
In doing so, we control for any time invariant mortgage risk or bank-specific variable that we have
omitted to include in the first two approaches.
6
(3) We investigate two mechanisms by which banks with an African American board
member might potentially lower the rejection rates of African American mortgage applicants. First,
an African American board member might influence bank credit policy by being a key member of
the board (see Chidambaran et al., 2022). Second, the presence of an African American board
member might be used to obscure a bank's reduced lending to low- and moderate-income groups,
like those covered by the Community Reinvestment Act. In this scenario, lower rejection rates for
African American borrowers could paradoxically indicate increased discrimination.
(4) We test whether rejection rates for White and other minority borrowers
8
are related to
having an African American on the board.
(5) We examine whether having an African American on the board is costly to shareholders?
To answer these questions, we construct a sample of 99 bank holding companies
9
where
we can identify the racial identity of all directors on each board for the years 2010 to 2021. This
corresponds to 680 bank-year observations, 41.79 million mortgage loan applications, and 429
bank-year observations with at least one African American director. In this sample, we find:
(1) African Americans account for 6.7% of board directors on average, and 63.1% of all
bank-years have at least one African American board director. African American board
representation is correlated with a number of bank characteristics. Small banks (asset size less than
$10 million) are less likely to have African American board representation than medium and large
banks. Banks with larger boards and more female directors are more likely to have at least one
African American director. We find that having an African American on the board is unrelated to
the prior year’s bank performance and risk, deposits, and real estate mortgage portfolio. We also
8
Others are mortgage applicants belonging to any of the following ethnic groups: Alaska Native, Asian, Hispanic,
Native American, and Pacific Islander.
9
In this paper we use “bank holding companies” and “banks” interchangeably.
7
find strong evidence that the labor supply of potential African American board members has a
significant impact on African American board representation. Specifically, a higher percentage of
African American executives in the state where a bank's headquarters is located significantly
increases the likelihood of the bank having at least one African American director.
(2) Importantly, we find that banks with at least one African American director have
significantly lower rejection rates for African American borrowers who apply for home mortgage
loans. Using 2SLS, we find that having at least one African American director is associated with
a 24.8% lower rejection rate for African American mortgage applicants form states where the
banks is not headquartered. Using matching models, the reduction in rejection rates ranges from
3.4% to 6.2%.
10
We also observe that five years following the addition of an African American
director to the board (including the year of their appointment), the rejection rates for African
American mortgage applicants are 8.3% to 9% lower compared to the five years prior. We also
find that large banks have higher rejection rates for African American borrowers when compared
to medium and small banks. Better capitalized banks have lower rejection rates for African
American borrowers, as do banks with a higher fraction of real estate loans. Finally, banks with a
greater fraction of female directors on the board have higher rejection rates of African American
mortgage applicants.
(3) Within banks with at least one African American director, African American directors
serve on key committees 90.2% of the time. Furthermore, banks where African American directors
serve on key committees have lower rejection rates for African American mortgage applicants
10
2SLS regression coefficients often significantly exceed those in OLS and matching models. In our case, 2SLS
coefficients reflect the average treatment effect on banks that change board diversity in response to shifts in the
instrumental variable, rather than the effect on the entire population. If these banks show a greater reaction to the
treatment, 2SLS results in a larger estimated effect. Section 5.2.2 delves deeper into this issue.
8
compared to banks where African American directors do not serve on key committees. These
results support the argument that African American directors influence bank credit policy towards
greater equity. In contrast, within our sample, CRA lending is too small in magnitude to play any
role in explaining our main findings. Hence, we find no evidence that banks appoint African
American directors to the board to camouflage potentially poor lending records in CRA areas.
(4) We find that having an African American board member is associated with lower
rejection rates for mortgage loan applications of White borrowers. The results suggest that African
American directors may help adjust credit standards to benefit all disadvantaged borrowers,
irrespective of their race. However, we do not find any benefit to other racial minority borrowers.
This might be because of relatively few such applications (low test power), or because combining
these diverse racial groups into a single category is inappropriate.
(5) We find no significant relationship between African American board representation and
the average two-year future bank risk (excess volatility) and bank performance (Tobin’s Q, ROA,
and non-performing loans). We also perform an event study on the appointment of new African
American directors to the board; we find no abnormal stock returns during the 10-day period
around the appointment date. In other words, there is no evidence that diversifying the board
comes at a cost to the bank’s shareholders.
The above results consistently show that having an African American director is associated
with lower rejection rates for African American mortgage applicants. The results suggest that
African American directors likely play an important role in alleviating historical inequities in home
mortgage lending. Even though we have used a 2SLS model with a strong IV that is conceptually
similar to Giannetti and Zhao (2019), we view our results as strong correlations and not causal.
9
Studies of new legislation or future campaigns by large institutional shareholders for appointing
African American directors to bank boards might allow for stronger causal interpretation.
11
This rest of the paper is structured as follows: Section 2 reviews recent legislative efforts
and literature on board diversity. Section 3 outlines hypotheses concerning the factors influencing
the presence of African American directors on bank boards. Section 4 details the data and sample
characteristics, followed by Section 5 which presents our empirical findings. Finally, Section 6
offers our conclusions.
2. Board diversity and mortgage lending
2. 1 Recent initiatives to increase board diversity
Despite years of pressure from regulators and social activists, progress on appointing more
minority directors on corporate boards remains slow (Adams & Ferreira, 2009; Field et al., 2020;
Kim & Starks, 2016; Lemayian et al., 2020; Peterson et al., 2007). Faced with such slow progress,
regulators, Congress, stock exchanges and institutional investors are increasingly adopting a more
aggressive approach wherein board diversity goals are mandated rather than encouraged. In this
section, we briefly review the recent initiatives to increase board diversity.
On December 1, 2020, NASDAQ proposed rules that would require all listed companies
to disclose information regarding the diversity of their board.
12
Additionally, NASDAQ
companies must appoint at least two diverse directors, one a person who self-identifies as female,
11
See the interesting paper by Gormley et al. (2023), who find that the three biggest index fund families, Blackrock,
State Street, and Vanguard, explicitly started campaigns to increase gender diversity on the board of directors. This
caused companies to significantly increase the representation of women on boards.
12
https://www.nasdaq.com/press-release/nasdaq-to-advance-diversity-through-new-proposed-listing-requirements-
2020-1
10
and the other a person who self-identifies as either an underrepresented minority
13
or as lesbian,
gay, bisexual, transgender or queer. All NASDAQ companies would have two years to appoint
one diverse director. Companies on the NASDAQ Global Market and Global Select Market would
need two diverse directors within four years. Companies on the NASDAQ Capital Market would
need two diverse directors within five years. Failure to satisfy the requirement could result in
delisting. The SEC’s recently released Item 401 and Item 407 of Regulation S-K Compliance and
Disclosure Interpretations
14
which recommend that where director nominees have self-identified
diversity characteristics and consent to their disclosure, the company’s disclosures should identify
those characteristics, along with other qualifications or attributes, to the extent they are considered
by the board or nominating committee in evaluating board membership. The Improving Corporate
Governance Through Diversity Act of 2021 (H.R. #1277), introduced in the US House of
Representatives on February 24, 2021, mandates that issuers of securities disclose the racial,
ethnic, and gender composition of their boards of directors and executive officers, as well as
whether any members are veterans.
15
It also requires the disclosure of any plan to promote racial,
ethnic, and gender diversity among these groups. Additionally, the SEC must establish a Diversity
Advisory Group to report on strategies to increase gender, racial, and ethnic diversity among board
members.
Institutional Shareholder Services (ISS), the largest and most prominent proxy advisory
firm, recently introduced rules for Russell 3000 or S&P 1500 companies that will be effective for
meetings on and after February 1, 2022.
16
ISS will generally recommend a vote "against" or
13
Members of an "underrepresented minority" are at least one of the following races or ethnicities: African American,
Alaska Native, Hispanic, Native American, and Pacific Islander.
14
https://www.sec.gov/divisions/corpfin/guidance/regs-kinterp.htm#116-11
15
https://www.congress.gov/bill/117th-congress/house-bill/1277
16
https://www.mofo.com/resources/insights/210127-iss-glass-lewis-nasdaq-board-diversity.html
11
"withhold" for the chair of the nominating committee (or other directors on a case-by-case basis)
where the board has no apparent racially or ethnically diverse members. Mitigating factors include
the presence of a racial and/or ethnic minority on the board at the preceding annual meeting and a
commitment to appoint at least one racially and/or ethnically diverse board member. Similarly,
starting January 1, 2021, Glass Lewis, another prominent proxy advisory firm, identified board
diversity as a factor of concern. It will assess board diversity based on race and ethnicity and a
company’s disclosure of such data in its proxy statement. Glass Lewis' guidelines indicate that its
recommendations will comply with relevant state law board composition requirements. Among
investment banks, Goldman Sachs has pledged that starting July 1, 2020, they would not
underwrite companies to go public who do not have at least one diverse director, and at least two
diverse directors in 2021.
17
States such as California, Hawaii, Illinois, Massachusetts, Michigan, New York, and
Washington have proposed similar rules. For example, California passed a law in September 2020
that requires public companies headquartered in the state to have two or three directors from
underrepresented groups by the end of 2022.
18
However, the LA County Superior Court ruled on
April 1, 2022 that the law violates the state’s constitution.
19
Finally, pension funds are also starting to exert their influence towards more diverse boards.
The New York City Pension Fund targeted 151 companies in the S&P 500, 49 of whom have
elected 60 new directors who identify as female or as a person of color; including 45 women, 16
African Americans, four Hispanic Americans, and two Asian Americans. Additionally, 24
17
https://www.goldmansachs.com/our-commitments/diversity-and-inclusion/launch-with-gs/pages/commitment-to-
diversity.html.
18
https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201920200AB979.
19
https://www.reuters.com/world/us/la-court-rules-state-corporate-diversity-law-unconstitutional-2022-04-02/.
12
companies have publicly committed to include women and people of color in the candidate pool
for every board search going forward.
20
2.2 Literature on mortgage lending discrimination and board diversity
This paper intersects with existing research in two broad areas, mortgage lending
discrimination and board diversity. In this section, we briefly discuss the relevant research in each
area.
The early literature on mortgage discrimination showed that lenders reject significantly
more mortgage applications from minorities than from white borrowers (Black et al., 1978; King,
1981; Schafer & Ladd, 1981). Banks and conservative commentators argued that the discrepancy
in rejection rates were due to omitted variables that reflect differences in the credit histories of
minority and white borrowers rather than discrimination. In response, Munnell et al. (1996)
combined HMDA data on loan applications in Boston with additional borrower data collected via
survey by the Federal Reserve Bank of Boston. They confirm that minority borrowers have less
wealth, lower credit scores, and higher loan-to-value ratios than white applicants. After controlling
for these and other factors, Munnell et al. (1996) still find that minorities are 8% more likely to be
rejected than white borrowers. Hubbard et al. (2011) examine mortgage loans in New Jersey and
find that minorities are more likely to be rejected than white borrowers in the prime market, but
less likely to be rejected than white borrowers in the subprime market. Quillian et al. (2020) find
that the higher rejection rate faced by minorities has declined only slightly from 1976 to 2016.
Giacoletti et al. (2021) examine the differential effects of loan approval quotas at the end of the
month compared to other times when these quotas are not binding. Their clever argument is that
20
https://www.30percentcoalition.org/news/the-coalition-in-the-news/nyc-pension-funds-push-for-board-diversity-
yields-impressive-results.
13
the personal judgment of loan officers about borrowers, which is not easily seen or measured,
wouldn't change much in short time spans, such as within a month or at the end of the month. They
find that in the first seven days of a month, loan rejections are 20% higher for African American
borrowers compared to White borrowers. Even at the end of the month, when loan approval quotes
are binding, African American borrowers still have a 10% higher chance of being rejected than
White borrowers.
Many papers show that the rejection rates for minority borrowers are lower when the
lending institution uses minority loan officers to work with minority borrowers (Fisman et al.,
2017; Fisman et al., 2020.) Jiang et al., (2021) observe fewer mortgage loan approvals when a
senator from the state where a bank's headquarters are located gains significant political power,
specifically by serving as a committee chair in the previous two years. This decline in approval
rates is more pronounced for minority applicants.
There is also a literature on price-based discrimination that shows that minority borrowers
are charged higher interest rates, points and fees on their home mortgages than White borrowers
(Ambrose et al., 2021; Bartlett et al., 2022; Bayer et al., 2018; Bhutta & Hizmo, 2020; Black et al.,
2003; Courchane & Nickerson, 1997; Ghent et al., 2014; Reid et al., 2017; Willen & Zhang,
2022).
21
There are three theories offered to explain discrimination. First, the taste-based
discrimination theory of Becker (1957) posits that employers discriminate because of their explicit
prejudice or animus towards minority candidates. Importantly, employers are willing to pay a
financial penalty to avoid interacting with minorities. Second, Bertrand et al. (2005) allow for
21
We do not test for price-based discrimination in this paper because we lack detailed data on loan terms and borrower
characteristics.
14
“implicit” discriminatory attitudes, which are unconscious mental associations toward agents of a
certain group. Bordalo et al. (2016) model stereotypes based on Kahneman and Tversky (1972)’s
representativeness heuristic. A decision maker assesses a target group by overweighting its
representative type, defined as a type that occurs more frequently in that group when compared to
a baseline reference group. Stereotypes might reflect real differences accurately, or exaggerate real
differences, which, the authors argue, is why they contain a ‘‘kernel of truth.’’ In the case when
stereotype exaggerates differences and causes belief distortions, it leads to inaccurate depictions
of the target group.
Third, the information-based or statistical discrimination theory of Arrow (1971) and
Phelps (1972) posits that discrimination is profit-maximizing in a world with imperfect
information. When information is limited and costly to gather, economic agents discriminate
against minorities because of salient group characteristics. The information-based theory predicts
that lenders are more likely to reject minorities because minorities have worse credit characteristics.
As explained above, in the case when stereotypes accurately reflect credit differences between
groups, Bordalo et al. (2016)’s model would be consistent with information-based
discrimination.
22
To understand why minority directors could help improve lending outcomes for minorities,
it is useful to consider the role of directors. Directors have a dual role as advisors and monitors of
managers (Adams & Ferreira, 2007). Minority directors may advocate for more equitable lending
practices in both roles. As advisors, minority directors can help banks tailor their lending activities
to better meet the needs of minority borrowers. Robinson and Dechant (1997) and Carter et al.
(2003) argue that matching the diversity of a company’s leadership to the diversity of the
22
We do not test which discrimination theory best explain our results because their ex ante predictions are the same.
15
company’s customers helps increase market share. Consistent with this hypothesis, Carter et al.
(2003) find a positive relationship between the fraction of minorities on the board and firm value.
23
Giannetti and Zhao (2019) examine board ancestry and find that firms with more ancestral
diversity have greater returns, more patents, more well-cited patents, and higher stock return
volatility. Cook and Glass (2015) find a positive association between race/ethnic board diversity
and product development and innovation. Minority directors also bring different perspectives and
priorities due to their upbringing and professional experience. They are more likely to have
experienced discrimination and bias than their white colleagues (Collins, 1997; Smith & Nkomo,
2003). As a result, minority directors encourage firms to put more weight on the interests of the
firm’s outside stakeholders (Post et al., 2011; Wang & Coffey, 1992). In the context of mortgage
lending, they may champion policies that expand access to credit to underserved communities.
Minority directors might also help select and reward CEOs that share these priorities.
Appointing minority directors can also signal to outside stakeholders that the bank is
committed to diversity, equity, and inclusion (Certo, 2003; Deutsch & Ross, 2003). Lamkin
Broome and Krawiec (2008) conduct 38 in-depth interviews with directors about their view of
board diversity. The authors report that the single most coherent narrative regarding diversity is
that diverse boards send a signal to employees that the company values and promotes diversity.
Further evidence consistent with a signaling effect is provided by Miller and del Carmen Triana
(2009), who show that firms with greater board racial diversity have higher reputation scores in
Fortune’s Corporate Reputation Survey after controlling for financial performance. In the context
23
In a subsequent study, Carter, Sinken & Simpson (2010) fail to find any significant association between the
percentage of minorities on the board and financial performance after controlling for the endogeneity of board
appointments.
16
of mortgage lending, diverse boards may signal to potential borrowers that the bank does not
discriminate against minority borrowers.
Appointing a single minority director on a large board may also help banks deflect criticism
from social activists without leading to any meaningful change regarding the treatment of minority
borrowers. Single minority directors are often considered “tokens” in the literature (Kanter, 1977);
being the only minority director can limit the director’s effectiveness because of negative
perceptions and stereotypes. If this is the case, then we may find no relation or even a positive
relation between board diversity and the rejection rate for minority borrowers.
Why do some banks choose to hire African American directors while others do not?
Empirical research on this topic is sparse. Rose and Bielby (2011) study the boards of S&P 500
companies over the period 1980 to 2000. They find that larger and more visible companies and
those with a more diverse workforce are more likely to appoint an African American director on
the board. In the cross-section, firms with larger boards, more interlocking directorships, finance
and utility firms, and firms with a greater percentage of African Americans in the headquarter state
are more likely to have at least one African American on the board. Using the same data, Glass
and Cook (2017) confirm that the representation of Black and Hispanic directors is predicted by
the racial/ethnic composition of the industry’s labor force. These studies do not analyze banks
separately. Knyazeva et al. (2013) find that the higher the availability of the local director pool the
higher the level of board independence which results in higher firm performance. Giannetti and
Zhao (2019) examine board ancestry and find that it tends to reflect the ancestral composition of
the location where the firms’ headquarters are located. Chidambaran et al. (2022) examine the
retention and appointment to key director positions within the board. Key director positions are
defined as chairperson, lead director, or chair of the compensation/nominating/audit committees.
17
The study then uses a skill diversity measure based on 18 skill characteristics, summarized as the
Euclidean difference between a director’s skill and the median board member’s skill. They find
that the likelihood of retaining and appointing a director increases with the director’s skill diversity.
However, none of these studies include banks in their analysis.
3. Testable hypotheses for having an African American on the board of directors
In this section, we develop testable hypotheses for having an African American member
on a bank’s board of directors. Additionally, we describe proxy variable(s) used to capture each
hypothesis, which are summarized in Table 1.
*** Table 1 ***
We estimate the following OLS regression model:
(1)
where i refers to banks and t refers to years, respectively.
is a vector that consists of
regression coefficients on variables that proxy for bank characteristics that may explain the choice
to include an African American director to the board.
is a vector that consists of regression
coefficients for control variables, year
t
are year fixed effects to capture systematic differences in
board diversity across time, and
is the regression error term.
24
All standard errors are clustered
by bank to account for potential correlations between observations of the same bank. The
dependent variable in the above equation is BoardAA, which is set to one if there is at least one
African American director on the board, and set to zero otherwise. We use an indicator variable
24
We are unable to use bank-level fixed effects models because we do not have a long enough panel data set.
18
rather than the percentage of directors who are African American because the latter variable varies
significantly with board size.
The different hypotheses and their associated empirical proxies are explained below. To
mitigate reverse causality concerns, all proxy and control variables in equation (1) above are
measured with a one-year lag.
Bank size: There are significant direct and indirect costs of finding and appointing a
qualified board member. Direct costs involve using a director search firm such as Korn Ferry
International, Heidrick and Struggles, or Spencer Stuart to help identify and short-list potential
director candidates. Indirect costs involve managerial time and effort to interview and get informal
recommendations from social networks as to the operating and managerial skills of the board
candidate. The larger the bank, the more resources it has to attract qualified African American
executives to its board of directors. This is especially true given the scarcity of African American
executives who are already board members or current CEOs/COOs of other firms. Additionally,
larger banks face greater regulatory and political scrutiny when they do not have minorities on
their board than smaller banks do. Accordingly, we would expect a large bank to have a higher
likelihood of having an African American board member than a small bank. To capture the non-
monotonic effect of bank size, we include two dummy variables that capture three bank size groups.
SmallSize equals one when a bank’s total assets are less than $10 million, and equals zero otherwise.
MediumSize equals one when a bank’s total assets are between $10 million and $100 million, and
equals zero otherwise. For large banks with total assets greater than $100 million, both SmallSize
and MediumSize are set to zero.
Board Size: The opportunity cost of appointing an African American board member is
higher when the total size of the board is smaller. For example, a bank board with only six directors
19
incurs a higher cost of managerial effort and time when it appoints an African American director
than a bank board with 12 members. To capture this effect, we include a variable BoardSize,
defined as the natural logarithm of the number of directors on the board.
Culture of diversity: It seems reasonable to hypothesize that banks that have a culture of
diversity are more likely to appoint an African American board member. Culture has been shown
to matter for firm performance by Edmans (2011), Guiso et al. (2015), Graham et al. (2022); these
results are summarized by Gorton et al. (2022). Fahlenbrach et al. (2012) demonstrate that culture
significantly impacts bank performance in particular. They find that a bank's buy-and-hold returns
during the 1998 Russian debt crisis strongly predict its performance in the 2007-08 financial crisis.
In our context, we consider bank characteristics that proxy for the bank’s culture with respect to
diversity. The first proxy variable is the proportion of board members who are female
(BoardFemale). We anticipate that boards with more female members are more inclined to hire an
African American board member, reflecting their stronger commitment to diversity. The second
proxy variable, BoardOtherMinority, measures the proportion of board members from racial
minority groups other than African American. This includes those identifying as Alaskan Native,
Asian, Hispanic, Native American, or Pacific Islander. We posit that boards with a higher
proportion of other minority board members also have a stronger commitment to diversity.
Following Hurtado and Sakong (2022) and Berger et al. (2023), we consider another proxy
variable, MDI, to indicate if a bank is a Minority Depository Institution.
25
Minority owners may
favor more African American directors on the board due to aligned beliefs and preferences.
25
According to the FDIC, an MDI is a federal insured depository institution with either 51% percent or more of the
voting stock owned by minority individuals, or with minority individuals holding the majority of board seats and
serving a predominantly minority community. See https://www.fdic.gov/regulations/resources/minority/mdi.html.
20
However, some might feel their existing non-discriminatory views towards minority borrowers
negate the need for African American directors to influence lending policies. Unfortunately, we
cannot include MDI in our tests because there is only a single MDI bank in our sample
(International Bankshares Corporation). Our findings remain unchanged regardless of whether this
bank is included in the analysis or not.
Ownership by the three biggest index fund families: In 2017, the three biggest index fund
families, Blackrock, State Street, and Vanguard, explicitly started campaigns to increase gender
diversity on the board of directors of US firms. Gormley et al. (2023) find that these campaigns
significantly boosted the representation of women on boards. Specifically, in 2019, American
corporations had at least 2.5 times more women serving on their boards compared to the numbers
in 2016. Accordingly, we create a variable, Big3, defined as the total ownership of these three big
index fund families. We expect that banks with greater ownership by Blackrock, State Street, and
Vanguard are more likely to include an African American director on the board.
Director labor supply: Research has shown that the availability of qualified African
American executives greatly influences their representation on company boards (Giannetti & Zhao,
2019; Glass & Cook, 2017; Knyazeva et al., 2013; Rose & Bielby, 2011). Such supply constraints
might be even more binding in the banking industry. Each subsidiary bank in the bank holding
company is separately chartered with its own board (Adams & Mehran, 2012). The National
Banking Act requires that the majority of bank directors reside in the state, territory or district of
the bank or within 100 miles of it for at least a year prior to, and during their term.
26
At the same
time, the Federal Reserve’s Regulation L restricts directors from serving simultaneously on the
26
See 12 USC 72, “Qualifications,” which outlines the role of banks and banking in the US Code.
21
board of unaffiliated banks in the same community or metropolitan statistical area.
27
These two
requirements significantly reduce the potential pool of qualified bank directors. To quantify this,
we use a measure we call AAExecutives, defined as the ratio of African-American executives to
total executives in the state where the bank is headquartered.
Our analysis includes control variables for bank characteristics that could affect having an
African American board member. First, we account for a bank's past performance, measured by
return on assets (ROA). Second, we incorporate two metrics to control for bank risk. One is the
bank's past annual excess stock return volatility (IVOL). This is calculated as the square root of 12,
multiplied by the standard deviation of the bank's monthly excess stock returns determined using
the Fama-French three-factor model. The other risk metric is the bank's leverage, measured as a
bank's equity capital to total assets ratio (Capital). In addition, we include the ratio of deposits to
total assets (Deposits). This measures how reliant the bank is on deposits for funding, as opposed
to other borrowing methods like repos or equity. To account for a bank's dependency on real estate
loans, we also control for the ratio of real estate loans to total assets (RealEstateLoans).
28
We
acknowledge the possibility that our instrument variable director labor supply may reflect not just
the labor supply in the bank's headquartered state, but also the state's wealth. This could mean a
higher proportion of African American executives are available due to higher wealth levels in that
state. To control for this, we include two additional control variables. The first is the logarithm of
per capita income (logGDP) for the bank’s headquarter state (in 2012 US dollars), and the second
measures the income disparity between African American and White households in the bank's
27
See §203 of the Depository Institutions Management Interlocks Act.
28
We also include a bank’s CRA rating, a dummy variable for possessing a CRA rating, and the percentage of
mortgage applications from CRA-targeted areas in our analysis. None of these variables are statistically significant.
These results are not reported but are available upon request from the authors.
22
headquartered state (IncDisparity). It is defined as the ratio of median household income over the
past 12 months between White and African American households, according to the American
Community Survey (ACS) of the U.S. Census Bureau.
4. Data and sample characteristics
4.1 Sample construction
Table 2 shows how we constructed our sample. The initial sample comprises of all
financial companies included in the S&P 1500 index with board composition information in the
Institutional Shareholder Services (ISS) director file. Each calendar year, the ISS director file
provides a list of current directors as of the proxy date. We augment this list with (1) the names
of directors that are appointed during the same calendar year but after the proxy date, and (2) the
names of directors that step down during the year but before the proxy date. As a result, we define
a board as the group of individuals who served as directors for any part of the calendar year.
*** Table 2 ***
The ISS director file provides information on director ethnicity. ISS determines director
ethnicity using several data sources. First, it seeks company feedback and explicit public disclosure
of director ethnicity which is voluntarily provided by each director. In the absence of such
disclosure, ISS utilizes other sources such as director biographies, company documents and
websites of other companies or associations where an individual also serves as a director or as an
employee. If a clear determination still cannot be made, ISS uses the director’s photo in public
filings (DEF 14A, 10-K) to validate his/her ethnicity.
The ethnicity of some directors in some years is missing in the dataset. In many cases, we
can backfill and forward-fill the missing data based on the data for the director in other years.
23
Following Bernile et al. (2018), we exclude companies with fewer than four directors on the board.
We also exclude companies with missing director ethnicity data. The sample period, 2010 to 2021,
corresponds to the period with available loan origination data from the Home Mortgage Disclosure
Act Database (HMDA).
The Federal Reserve assigns a unique identification code, RSSD, to each financial
organization that it oversees. We use the PERMCO-RSSD link table provided by the Federal
Reserve Bank of New York (FRBNY) to link the ISS directors file data to the HMDA loan
originator data. In our data, 2,400 firm-years do not have an RSSD code assigned by the Federal
Reserve; these observations are dropped from our sample. We drop another 266 firm-years because
the bank has no loans application records in the HMDA dataset. The resulting sample consists of
1,173 bank-year observations and 44,600,096 mortgage loan applications. We only consider
mortgage loan applications for properties with one to four family units. These properties are more
likely to be for personal use rather than commercial use. The sample excludes loan purchases
because our focus is on the rejection rates at the loan origination stage.
We require that data is available on CRSP, Compustat, and the Federal Reserve’s form
Y9C, which reports consolidated financial statements for bank holding companies. This
requirement further reduces the sample by 228 bank-year observations. Next, we require that
banks have at least 1,000 loan applications in any given year. This requirement eliminates 84
bank-year observations and ensures that our results are not driven by small loan originators who
may have only a handful of mortgage loan applications by minority borrowers. We also restrict
the sample to bank-years with at least 50 African American loan applications to ensure a
meaningful representation of potential African American borrowers among the pool of applicants.
24
The final sample consists of 681 bank-year observations and 41.99 million mortgage loan
applications from 2010 to 2021. The number of unique banks in our sample is 99.
Data on the ethnicity of executives in the headquarter state is provided by the Equal
Employment Opportunity Commission (EEOC).
4.2 Sample characteristics
Table 3 reports the summary statistics for the variables in the sample. We find that the
rejection rates are significantly higher for African American borrowers. For African Americans,
the average rejection rate is 51.8%, relative to 34.6% for White borrowers. The average rejection
rate is also higher for Other borrowers, at 44%. In terms of board diversity, we find 63.1% of all
bank-years in our sample have at least one African American director. However, African
American board members account for only 6.7% of all directors in the sample. Out of the 429
bank-year observations with at least one African American director, 247 have only one African
American director, 159 have two, 21 have three, and 2 have four.
29
*** Table 3***
The data clearly demonstrate that minorities’ representation on bank boards is significantly
lower than their proportion in the general population. A potential explanation for this could be the
scarce number of minority executives in states where these banks are headquartered. For instance,
the average percentage of African American executives (AAExecutives) in these states is a mere
3.4%. Furthermore, minorities represent only a small fraction of all mortgage applications: African
Americans account for 5%, and other minority groups make up 12.1%. Along with high rejection
rates, this suggests that minorities face substantial challenges in securing mortgage financing. In
29
Trust Financial Corp (as of 2021) and TCF Financial Corp (as of 2020) are the banks with the most significant
African American representation on their boards. Each bank has four African American directors.
25
the following section, we examine the relationship between having an African American director
on the board and home mortgage rejection rates.
5. Empirical Results
5.1 Determinants of having an African American on the board
We begin our empirical tests by estimating equation (1), with the results reported in Table
4. Columns (1) reports estimates when we include variables that might be endogenously chosen
by the bank such as board diversity and size, whereas column (2) excludes them. In column (1),
we find that large- and medium-size banks have the highest likelihood of having an African
American on their board, when compared to small banks. We also find that that larger boards
(BoardSize)
and boards with a greater fraction of female directors (BoardFemale) are more likely
to include an African American director. The ownership of the Big 3 index fund families, Big3,
is statistically insignificant. The coefficient on the labor supply variable, namely the availability
of African American executives in the bank’s headquarters state (AAExecutives), is positive and
highly significant. Increasing AAExecutives from the 25
th
to the 75
th
percentile increases the
probability of having an African American director by 7%, all else equal (0.014*5.004). This
finding suggests that a lack of African American executives in a bank's home state is a major
obstacle to forming more diverse boards. None of the control variables we used in our analysis
show statistical significance.
*** Table 4 ***
Column (2) of Table 4 excludes the potentially endogenously chosen variables BoardSize,
BoardFemale, and BoardOtherMinority. We now find that large banks have the highest likelihood
of having an African American on their board, followed by medium-size banks, and finally small-
26
size banks. The results on the other determinants remain similar to the results in column (1). In
summary, we find that larger banks, banks headquartered in states with more African American
executives, and banks with larger boards and a greater fraction of female directors are more likely
to have an African American board member. We find no evidence that stock ownership of the Big
3 index fund families or any of the control variables help explain why some banks have African
American directors while others do not.
Our analysis in column (3) shows that the coefficients on these additional state
characteristics are not statistically significant. Importantly, including these variables in the model
does not alter the results presented in the first column.
5.2 African American directors and rejection rates of African American applicants
5.2.1 Two-Stage Least Squares (2SLS) regressions
We begin by examining the rejection rate for African American borrowers using 2SLS with
AAExecutives used as an instrument for BoardAA. The rejection rates for African American
borrowers, denoted as AfricanAmericanRR, represent the percentage of mortgage applications
from African American borrowers that were rejected by bank i in year t. We use the following
model:
where
represents the predicted value of BoardAA from the specification in
column (1) of Table 4, and
is its corresponding coefficient. The vector of regression coefficients
for the proxy variables is given by
,
is a vector of regression coefficients for control variables,
and
is the regression error term, respectively. We include year fixed effects and all standard
27
errors are clustered by bank to account for potential correlations between observations of the same
bank.
In Table 5, we present two estimates of equation (2). Column (1) of Table 5 shows the
estimates for equation (2), where AfricanAmericanRR is calculated after excluding loan
applications from the state where the bank is headquartered. This exclusion aims to address
potential uncontrolled state characteristics. For instance, states with larger African American
populations might exhibit both a higher supply of African American directors and increased
rejection rates for African American mortgage applicants due to historical socioeconomic
disparities. In such cases, AAExecutives might correlate with AfricanAmericanRR regardless of the
presence of an African American director on the bank's board. This concern is minimized when
analyzing rejection rates in states other than where the bank is headquartered, which is the basis
for calculating AAExecutives. The findings in column (1) reveal that banks with at least one African
American director have a 24.8% lower rejection rate for African American mortgage applicants
from states other than where the bank is headquartered. This evidence supports the hypothesis that
increased board diversity helps mitigate racial disparities in mortgage lending.
In column (2) of Table 5, our dependent variable AfricanAmericanRR is calculated for all
loan applications received by the bank, including loan applications from the bank’s headquarter
state. The results are nearly identical to the results in column (1). Banks with at least one African
American director have a 23.6% lower rejection rate for African American mortgage applicants.
30
*** Table 5 ***
30
We repeat the 2SLS estimates in Table 5 using the predicted value of BoardAA from the specification in column
(3) of Table 4. None of our results changed significantly. These results are not reported but are available from the
authors.
28
Our analysis also reveals that large banks exhibit the highest rejection rates of African
American borrowers, followed by medium-sized banks, while small banks have the lowest. Board
size does not significantly affect these rejection rates. Interestingly, banks with a higher proportion
of female board members, despite their efforts in board diversity including appointing African
American directors (as shown in Table 4), tend to have higher rejection rates. This result may seem
counterintuitive, but the choice to appoint female directors is likely endogenous and influenced by
various factors. Further research is needed to unravel these complex dynamics. This complexity is
echoed in our finding that boards with directors from other minority groups do not show a
significant correlation with rejection rates, indicating that the interplay of board composition and
lending practices may vary across different demographics.
In analyzing the role of the control variables, we observe no significant influence of past
bank performance on African American borrowers' loan rejection rates. Banks with better
capitalization show lower rejection rates for these borrowers, a pattern also seen in banks with a
higher volume of real estate loans. The amount of deposits held by banks does not significantly
impact African American borrowers' loan rejection rates.
31
.
5.2.2 Matching models
We estimate two matching models, namely, Kernel matching and Nearest Neighbor matching.
The goal is to closely match treatment and control observations on observable covariates so that
the two are only different from each other with respect to having an African American director on
31
From 2018 onwards, HMDA data includes additional borrower characteristics including FICO scores, combined
loan-to-value ratios, and debt-to-income ratios. However, the limited number of observations in this subset featuring
an African American board member prevents us from testing the robustness of our results against these additional
borrower variables.
29
the board. In doing so, we also include the director labor supply variable as one of our independent
variables.
Before estimating the matching models, we first check whether the two distributions
(treatment and control groups) have common support using the Min-Max condition. Common
support requires that the probability of receiving treatment for each possible value is within the
same interval as the probability of not receiving treatment. Using column (1) in Table 4, we find
common support for 582 of the 680 bank-year observations in the sample, respectively.
32
The
matching results are based on this restricted sample.
The results of both matching methods are given in Table 6. Consistent with the 2SLS results
in Table 5, we find that having an African American director on the board is associated with a
lower rejection rate for African American mortgage applicants. The magnitude of the effect of
having an African American director on the board is larger when we exclude loan applicants from
the state where the bank is headquartered then when we include them.
The disparity in rejection rates for African American applicants between banks with and
without an African American director is less pronounced in Table 6 compared to Table 5. Using
Kernel matching, we find an average effect of 6.3% when we consider loans outside the bank’s
headquarter state, and 3.4% for all loans. The results are similar using Nearest Neighbor matching.
Recall that for 2SLS, the effect is approximately 25%. It's not uncommon to see larger effects in
2SLS regressions compared to those estimated through matching methods. This could be attributed
to the endogeneity of the explanatory variable due to factors such as omitted variables, reverse
causality, or simultaneity. In such scenarios, the estimates using matching methods may be biased
32
For example, the range of probability of having an African American board member in the control group using
column (1) in Table 4 is 0.0192 to 0.9715 while for the treatment group is 0.039621 to 0.9999234. Hence, using the
Min-Max condition we restrict our sample to observations with probability in the range of 0.039621 to 0.9714886.
30
towards zero - they may underestimate the true effects. 2SLS corrects for endogeneity, leading to
larger absolute estimated effects. Moreover, the treatment effect estimated in 2SLS models
measures the average treatment effect for banks that alter their board diversity due to fluctuations
in the instrumental variable (AAExecutives), as opposed to measuring the effect for the entire
population. If these banks exhibit a larger response to the treatment, the estimated effects in 2SLS
can be larger than those derived from matching methods.
*** Table 6 ***
5.3 Changes in rejection rates when an African American director is appointed to the board
The above results are based on panel data of loan rejection rates and board composition. It
is still possible that we have not included some bank-specific and/or mortgage risk variables that
might impact our results on loan rejection rates. Ideally, we would like to estimate a bank-level
fixed-effects model to control for such time-invariant risk factors and for the credit risk of the pool
of mortgage applicants. However, we are constrained by the paucity of director ethnicity data
currently available as the disclosure of board member ethnicity is not mandated. Additionally,
banks in our sample need to have ethnicity data for all board members to be included in the analysis.
As an alternative approach, we explore how hiring an African American director to the
bank’s board affects African American loan rejection rates. Within our sample, 47 banks appoint
an African American director during the sample period. However, the number of banks in the
analysis varies each year with the availability of data. The results of this analysis are shown in
Table 7. Panel A presents the outcomes for all banks that appoint an African American director in
year 0, irrespective of their board composition. In the five years before year 0, the mean loan
rejection rate for African American borrowers is 57%. In the five years that follow (including year
0), the mean loan rejection rate for African American borrowers is reduced to 50%. The difference
31
between the pre- and post-rejection rates is statistically significant at the 1% level. Panel B focuses
on banks without any African American directors that appoint an African American director to
their board in year 0, but without any African American directors. Consistent with the results in
Panel A, we find the mean loan rejection rate for African American borrowers is 55% in the five
years before year 0, and 48% in the five years that follow. The difference between the pre- and
post-rejection rates is statistically significant at the 1% level. These results show that hiring an
African American director is associated with a decrease in the rejection rate for African American
borrowers.
*** Table 7 ***
The above results are based on differences in mean loan rejection rates. In Table 8, we
perform a similar analysis in an OLS regression framework. Specifically, we create a dummy
variable, AfterBoardAA, that equals one for years 0 to 4, and zero for years -5 to -1. The sample
in column (1) includes all banks that appoint an African American director in year 0. The sample
in column (2) includes banks without any African American directors that appoint an African
American director in year 0. In both columns, the coefficient on AfterBoardAA is significantly
negatively related to loan rejection rates. The magnitude of the reduction in rejection rates is 9.7%
for the larger sample and 9.9% for the restricted sample. These findings provide further support
for the hypothesis that African American directors help improve mortgage access for African
American borrowers.
*** Table 8 ***
5.4 Potential mechanisms for the lower rejection rate of African American applicants
The above results show that African American mortgage applicants are less likely to be
rejected when an African American director is on the board. We examine two mechanisms that
32
might explain this result. First, an African American board member might influence bank credit
policy by being a key member of the board. Such an argument has been posited by Chidambaran
et al. (2022) who state: “We take the view that the presence of a diverse director, or getting a seat
at the table, although visible, is only a first step toward diversity on the board. The next step is
“being in the game,” or the engagement of diverse directors in the boards they serve on to influence
and shape how the boards discharge their important functions” (page 194). We define key director
positions as chairperson, lead director, or member of the compensation/nominating/audit
committees.
33
We then investigate if banks with African American directors in key positions
exhibit lower rejection rates for African American applicants compared to banks where African
American directors are not in such key roles.
The results are reported in Table 9. We find that 388 out of a total of 430 (or 90.2%) of
African American board members serve in key positions, and only 42 (or 9.8%) of African
American board members do not. This shows that African American directors have the potential
to influence key decisions on the board. Importantly, we find lower rejection rates for African
American mortgage applicants when the African American director serves in a key position v.
when s/he does not (51.3% vs. 55.2%, on average). The difference of 3.9% is statistically
significant at the 10% value (p-value of 0.0587). This pattern largely holds for the entire
distribution of rejection rates for African American mortgage applicants.
*** Table 9 ***
We next examine whether differences in CRA lending might explain the different rejection
rates of banks with and without African American directors. This analysis is motivated by the
33
Our definition of key positions is slightly broader than the definition used in Chidambaran et al. (2022), who classify
directors as serving in key positions if they serve as chairperson, lead director, or chair of the compensation/
nominating/audit committees.
33
intriguing findings of Basu et al. (2022), who demonstrate that banks with high ESG scores are
more likely to reject mortgage loan applicants from poor census tracts compared to banks with low
ESG scores. They interpret the evidence as a type of “social wash,” where some banks aim for
high ESG scores without lending significantly to disadvantaged communities. Appointing an
African American director might be another example of such social wash. Specifically, banks may
appoint an African American director to obscure or draw attention away from a bank's reduced
lending to low- and moderate-income groups, like those covered by the Community Reinvestment
Act. Because African American borrowers account for a disproportionally large fraction of
households in CRA areas
34
, such a strategy may paradoxically manifest in lower rejection rates of
African American borrowers.
We examine this potential mechanism in Table 10, where we report the percentage of loan
applications from CRA areas, for all applicants and for African American applicants. Panel A
shows the results for the full sample. Panel B shows the results for banks without an African
American director, and Panel C shows the results for banks with an African American director. In
the full sample, a third of mortgage loan applications come from CRA areas. Yet, only 1.4% of
African American borrowers' loan applications are from these areas. The frequency is even lower
for banks with an African American director in Panel C, at 1.2%. Therefore, lending to African
American borrowers in CRA areas is too infrequent to explain the documented differences in
rejection rates.
*** Table 10 ***
34
Bostic and Robinson (2003), Dahl et al. (2010), and Agarwal et. al (2015) find that CRA ratings (assigned by federal
regulators) are correlated to increases in targeted lending to borrowers such as minorities.
34
5.5 African American directors and rejection rates for other racial groups
The above results indicate a negative correlation between having an African American
director and the rejection rates for African American borrowers. This leads to a related question:
Do borrowers from other racial groups also experience improved access to mortgage financing? If
African American directors are more sensitive to racial or socioeconomic disparities in the
mortgage market, their appointment may also benefit other ethnic groups. Addressing inequality
may even benefit White borrowers of lower socioeconomic status who traditionally have also had
difficulties obtaining loans. This is because any lending policy that a bank chooses to implement
must apply to all applicants regardless of race.
Table 11 shows the impact of having an African American director on the board on the
rejection rate of White and Other (Alaska Native/Asian/Hispanic/Native American/Pacific
Islander) borrowers. For this analysis, we group the various minority groups into a single category
because together they account for only 12.1% of all mortgage loan applications. As in Table 5,
we estimate the effect using 2SLS and AAExecutives as an instrument for BoardAA. The sample
for this analysis consists of banks with at least 1,000 mortgage loan applications in year t and at
least 50 applications from White or Other Minority borrowers, respectively. Our analysis shows
that boards with at least one African American director are associated with a 15.2%-17.6% lower
rejection rate for White borrowers. Borrowers from other minority groups do not show a similar
benefit. This might be due to the relatively small number of such applications, resulting in a test
with low power, or because combining these diverse groups into a single category is inappropriate.
*** Table 11 ***
35
5.6 Is appointing an African American director costly to banks’ shareholders?
The above results show that having an African American board member lowers the loan
rejection rates of African American and White borrowers. We next examine if appointing an
African American director costly to banks’ shareholders?
We first test whether banks with at least one African American director experience worse
future performance and/or higher risk because of the lower rejection rates for African American
borrowers. Table 12 shows the association between several bank performance and risk metrics and
BoardAA, controlling for other bank, board, and loan pool characteristics. We examine measures
of financial performance and risk for the two years t+1 and t+2. We use four proxy variables for
bank performance, namely, buy and hold returns (FutBHRet), accounting profitability (FutROA),
stock market value (FutTobinQ), and the nonperforming loans as a fraction of the bank’s total
assets (FutBadLoans). We measure risk as the excess stock return volatility from the three-factor
Fama-French model (FutIVOL). We find no relationship between BoardAA and banks’ future
financial performance and risk. The only marginally significant result is for IVOL, where BoardAA
is associated with 0.8% higher future stock return volatility.
*** Table 12 ***
As an additional test, we conduct an event study around the appointment of a new African
American director to the board.
35
As mentioned in Section 5.3, we identify 47 such instances in
our sample. We use Factiva to pinpoint the earliest mention of each appointment in press releases
or news articles, successfully obtaining definitive dates for 45 out of the 47 cases. We then
calculate excess stock returns for these 45 banks over a 20-day period surrounding the
announcement date for the appointments. The results are summarized in Figure 1. We find no
35
We thank Adi Sunderam for proposing this event study.
36
excess stock returns during the 20-day period around the appointment of a new African American
director. Hence, shareholders do not suffer financial losses in result of these appointments. These
results are consistent with the panel results reported in Table 11, showing no difference in future
bank performance and risk. Overall, we find that banks do not incur material financial costs when
they appoint African American directors to the board.
*** Figure 1 ***
6. Conclusions
US stock exchanges, the SEC, state legislatures, pension funds and proxy advisors have
recently intensified their efforts to increase board diversity in the United States. In this paper, we
examine whether banks with African American board members are less likely to reject mortgage
applications of minority borrowers. Using these different methodologies, we find that banks with
at least one African American director on the board have significantly lower rejection rates for
African American borrowers. Approximately 90% of African American directors serve in key
positions, and when they do, the rejection rates are even lower. We find no evidence that banks
with African American directors reduce lending to low- and moderate-income groups covered by
the Community Reinvestment Act.
We also find that large banks have higher rejection rates when compared to medium size
banks, and the smallest banks have the lowest rejection rates. Loan rejection rates are lower for
Whites borrowers too but not for Other Minority borrowers. In examining what determines the
presence of an African American board member, we find that larger bank size, bigger board size,
a higher proportion of female directors, and a greater fraction of African American executives in
the bank's headquarter state are all associated with having an African American on the board.
37
Importantly, we find no evidence that diversifying the board comes at a cost to the bank’s
shareholders.
Our results inform the debate surrounding the costs and benefits of having diverse boards.
Most papers in the literature focus on the link between diverse board and firm performance and
firm value. We focus on the link between diverse boards and access to credit. Considering the large
economic disparities between races in the United States, understanding how to improve lending
outcomes for minorities is important.
38
References
Adams, R. B., & Ferreira, D. (2007). A theory of friendly boards. Journal of finance, 62(1), 217-250.
Adams, R. B., & Ferreira, D. (2009). Women in the boardroom and their impact on governance and performance.
Journal of Financial Economics, 94(2), 291-309.
Adams, R. B., & Mehran, H. (2012). Bank board structure and performance: Evidence for large bank holding
companies. Journal of Financial Intermediation, 21(2), 243-267.
Ambrose, B. W., Conklin, J. N., & Lopez, L. A. (2021). Does borrower and broker race affect the cost of mortgage
credit? Review of Financial Studies, 34(2), 790-826.
Ann Choi, K. H., Olivia Winslow and project editor Arthur Browne. (2019). Longisland Divided [Newspaper
Article]. https://projects.newsday.com/long-island/real-estate-agents-investigation/ (Newsday)
Arrow, K. (1971). Some Models of Racial Discrimination in the Labor Market. RAND Corporation.
https://www.rand.org/pubs/research_memoranda/RM6253.html
Bartlett, R., Morse, A., Stanton, R., & Wallace, N. (2022). Consumer-lending discrimination in the FinTech era.
Journal of Financial Economics, 143(1), 30-56.
Basu, S., Vitanza, J., Wang, W., & Zhu, X. R. (2022). Walking the walk? Bank ESG disclosures and home mortgage
lending. Review of Accounting Studies, 27(3), 779-821.
Bayer, P., Ferreira, F., & Ross, S. L. (2018). What drives racial and ethnic differences in high-cost mortgages? The
role of high-risk lenders. Review of Financial Studies, 31(1), 175-205.
Becker, G. S. (1957). The economics of discrimination. University of Chicago Press, Chicago.
Berger, A. N., Feldman, M., Langford, W. S., & Roman, R. A. (2023). 'Let Us Put Our Moneys Together:'Minority-
Owned Banks and Resilience to Crises. Available at SSRN 4231594.
Bernile, G., Bhagwat, V., & Yonker, S. (2018). Board diversity, firm risk, and corporate policies. Journal of
Financial Economics, 127(3), 588-612.
Bertrand, M., Chugh, D., & Mullainathan, S. (2005). Implicit discrimination. American Economic Review, 95(2),
94-98.
Bhutta, N., & Hizmo, A. (2020). Do Minorities Pay More for Mortgages? Review of Financial Studies, 34(2), 763-
789.
Black, H., Schweitzer, R. L., & Mandell, L. (1978). Discrimination in mortgage lending. American Economic
Review, 68(2), 186-191.
Black, H. A., Boehm, T. P., & DeGennaro, R. P. (2003). Is there discrimination in mortgage pricing? The case of
overages. Journal of Banking & Finance, 27(6), 1139-1165.
Boone, A. L., Field, L. C., Karpoff, J. M., & Raheja, C. G. (2007). The determinants of corporate board size and
composition: An empirical analysis. Journal of Financial Economics, 85(1), 66-101.
Bordalo, P., Coffman, K., Gennaioli, N., & Shleifer, A. (2016). Stereotypes. The Quarterly Journal of Economics,
131(4), 1753-1794.
Brown, J. A., Buchholtz, A. K., Stewart, M. M., & Dennis, B. (2012). Board Diversity as a Camouflage Signal
[Abstract]. Academy of Management Annual Meeting Proceedings, 2012(1).
Campbell, J. Y. (2006). Household finance. Journal of Finance, 61(4), 1553-1604.
Carter, D. A., Simkins, B. J., & Simpson, W. G. (2003). Corporate governance, board diversity, and firm value.
Financial Review, 38(1), 33-53.
Certo, S. T. (2003). Influencing initial public offering investors with prestige: Signaling with board structures.
Academy of Management Review, 28(3), 432-446.
Chidambaran, N. K., Liu, Y., & Prabhala, N. (2022). Director diversity and inclusion: At the table but in the game?
Financial Management, 51(1), 193-225.
Cochrane, A. (2007). Urban policy. The Blackwell Encyclopedia of Sociology.
Coles, J. L., Daniel, N. D., & Naveen, L. (2008). Boards: Does one size fit all? Journal of Financial Economics,
87(2), 329-356.
Collins, S. M. (1997). Black mobility in white corporations: up the corporate ladder but out on a limb. Social
Problems, 44(1), 55-67.
Cook, A., & Glass, C. (2015). Diversity begets diversity? The effects of board composition on the appointment and
success of women CEOs. Social Science Research, 53, 137-147.
Courchane, M., & Nickerson, D. (1997). Discrimination resulting from overage practices. In Discrimination in
Financial Services: A Special Issue of the Journal of Financial Services Research (pp. 133-151). Springer.
Deutsch, Y., & Ross, T. W. (2003). You are known by the directors you keep: Reputable directors as a signaling
mechanism for young firms. Management Science, 49(8), 1003-1017.
39
Edmans, A. (2011). Does the stock market fully value intangibles? Employee satisfaction and equity prices. Journal
of Financial Economics, 101(3), 621-640.
Fahlenbrach, R., Prilmeier, R., & Stulz, R. M. (2012). This time is the same: Using bank performance in 1998 to
explain bank performance during the recent financial crisis. Journal of finance, 67(6), 2139-2185.
Ferreira, D. (2010). Board diversity. Corporate governance: A synthesis of theory, research, and practice, 8, 225.
Field, L. C., Souther, M. E., & Yore, A. S. (2020). At the table but can not break through the glass ceiling: Board
leadership positions elude diverse directors. Journal of Financial Economics, 137(3), 787-814.
Fisman, R., Paravisini, D., & Vig, V. (2017). Cultural proximity and loan outcomes. American Economic Review,
107(2), 457-492.
Fisman, R., Sarkar, A., Skrastins, J., & Vig, V. (2020). Experience of communal conflicts and intergroup lending.
Journal of Political Economy, 128(9), 3346-3375.
Flavin, M., & Yamashita, T. (2002). Owner-occupied housing and the composition of the household portfolio.
American Economic Review, 92(1), 345-362.
Ghent, A. C., Hernandez-Murillo, R., & Owyang, M. T. (2014). Differences in subprime loan pricing across races
and neighborhoods. Regional Science and Urban Economics, 48, 199-215.
Giannetti, M., & Zhao, M. (2019). Board ancestral diversity and firm-performance volatility. Journal of Financial
and Quantitative Analysis, 54(3), 1117-1155.
Glass, C., & Cook, A. (2017). Appointment of racial/ethnic minority directors: Ethnic matching or visibility threat?
Social Science Research, 61, 1-10.
Gormley, T. A., Gupta, V. K., Matsa, D. A., Mortal, S. C., & Yang, L. (2023). The big three and board gender
diversity: The effectiveness of shareholder voice. Journal of Financial Economics, 149(2), 323-348.
Gorton, G. B., Grennan, J., & Zentefis, A. K. (2022). Corporate culture. Annual Review of Financial Economics, 14,
535-561.
Graham, J. R., Grennan, J., Harvey, C. R., & Rajgopal, S. (2022). Corporate culture: Evidence from the field.
Journal of Financial Economics, 146(2), 552-593.
Guiso, L., Sapienza, P., & Zingales, L. (2015). The value of corporate culture. Journal of Financial Economics,
117(1), 60-76.
Hubbard, R. G. (1985). Personal taxation, pension wealth, and portfolio composition. Review of Economics and
Statistics, 53-60.
Hubbard, R. G., Palia, D., & Yu, W. (2011). Analysis of discrimination in prime and subprime mortgage markets.
Available at SSRN 1975789.
Hurtado, A., & Sakong, J. (2022). The effect of minority bank ownership on minority credit.
Imbens, G. W. (2004). Nonparametric estimation of average treatment effects under exogeneity: A review. Review
of Economics and Statistics, 86(1), 4-29.
Jiang, E. X., Lee, Y., & Liu, W. S. (2021). Reducing racial disparities in consumer credit: the role of minority loan
officers in the era of algorithmic underwriting. USC Marshall School of Business Research Paper
Sponsored by iORB.
Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive
Psychology, 3(3), 430-454.
Kanter, R. M. (1977). Men and women of the corporation. Basic, New York.
Kim, D., & Starks, L. T. (2016). Gender diversity on corporate boards: Do women contribute unique skills?
American Economic Review, 106(5), 267-271.
King, A. T. (1980). Mortgage Lending, Social Responsibility, and Public Policy: Some Perspectives on HMDA and
CRA. Real Estate Economics, 8(1), 77-90.
King, A. T. (1981). Discrimination in mortgage lending: A study of three cities (Vol. 4). New York University,
Graduate School of Business Administration, Salomon ….
Knyazeva, A., Knyazeva, D., & Masulis, R. W. (2013). The supply of corporate directors and board independence.
Review of Financial Studies, 26(6), 1561-1605.
Lamkin Broome, L., & Krawiec, K. D. (2008). Signaling through board diversity: Is anyone listening. U. Cin. L.
Rev., 77, 431.
Lemayian, Z., Pownall, G., & Short, J. (2020). Why Are US Corporate Boards Under-diversified Among Genders
and Races? Available at SSRN 3567420.
Linck, J. S., Netter, J. M., & Yang, T. (2008). The determinants of board structure. Journal of Financial Economics,
87(2), 308-328.
Martinez, E., & Glantz, A. (2018). How reveal identified lending disparities in federal mortgage data. Center for
Investigative Reporting.
40
Miller, T., & del Carmen Triana, M. (2009). Demographic diversity in the boardroom: Mediators of the board
diversity–firm performance relationship. Journal of Management Studies, 46(5), 755-786.
Munnell, A. H., Tootell, G. M., Browne, L. E., & McEneaney, J. (1996). Mortgage lending in Boston: Interpreting
HMDA data. American Economic Review, 25-53.
Palia, D., Qi, Y., & Wu, Y. (2014). Heterogeneous background risks and portfolio choice: Evidence from micro‐
level data. Journal of Money, Credit and Banking, 46(8), 1687-1720.
Peterson, C. A., Philpot, J., & O'Shaughnessy, K. (2007). African‐American Diversity in the Boardrooms of the
US Fortune 500: director presence, expertise and committee membership. Corporate Governance: An
International Review, 15(4), 558-575.
Phelps, E. S. (1972). The statistical theory of racism and sexism. American Economic Review, 62(4), 659-661.
Post, C., Rahman, N., & Rubow, E. (2011). Green governance: Boards of directors’ composition and environmental
corporate social responsibility. Business & Society, 50(1), 189-223.
Quillian, L., Lee, J. J., & Honoré, B. (2020). Racial Discrimination in the U.S. Housing and Mortgage Lending
Markets: A Quantitative Review of Trends, 1976–2016. Race and Social Problems, 12(1), 13-28.
Reid, C. K., Bocian, D., Li, W., & Quercia, R. G. (2017). Revisiting the subprime crisis: The dual mortgage market
and mortgage defaults by race and ethnicity. Journal of Urban Affairs, 39(4), 469-487.
Robinson, G., & Dechant, K. (1997). Building a business case for diversity. Academy of Management Perspectives,
11(3), 21-31.
Rose, C. S., & Bielby, W. T. (2011). Race at the top: How companies shape the inclusion of African Americans on
their boards in response to institutional pressures. Social Science Research, 40(3), 841-859.
Rosenbaum, P. R. (2002). Overt bias in observational studies. In Observational studies (pp. 71-104). Springer.
Ross, S. L., Turner, M. A., Godfrey, E., & Smith, R. R. (2008). Mortgage lending in Chicago and Los Angeles: A
paired testing study of the pre-application process. Journal of Urban Economics, 63(3), 902-919.
Rubin, D. B. (2006). Matched sampling for causal effects. Cambridge University Press, Cambridge.
Schafer, R., & Ladd, H. F. (1981). Discrimination in mortgage lending. MIT Press, Boston.
Smith, E. L. B., & Nkomo, S. M. (2003). Our separate ways: Black and white women and the struggle for
professional identity. Harvard Business Press.
Stuart, E. A. (2010). Matching methods for causal inference: A review and a look forward. Statistical Science: A
Review Journal of the Institute of Mathematical Statistics, 25(1), 1.
Wang, J., & Coffey, B. S. (1992). Board composition and corporate philanthropy. Journal of Business Ethics,
11(10), 771-778.
Willen, P., & Zhang, D. (2022). Testing for discrimination in menus.
41
Figure 1: Cumulative excess returns around the appointment of an African American director to the board of directors
This figure shows cumulative excess returns around the announcement of the appointment (day 0) of an African American director to a bank’s
board. The sample consists of 45 banks that appoint a new African American director to the board from 2010 to 2021. Excess returns are
computed relative to the Fama-French three factor model, with factor loadings estimated over the 100 days ending 50 days prior to the event date.
The chart is generated using the CRSP U.S. Daily Event Study tool accessed through http://wrds-www.wharton.upenn.edu.
42
Table 1: Variable definitions and data sources
Variable name
Variable definition
Source*
AfricanAmericanRR
Percentage of mortgage applications from African American
borrowers that are rejected year t.
HMDA
OtherMinorityRR
Percentage of mortgage applications from other minority
borrowers (Alaskan Native/Asian/Hispanic/Native American/
Pacific Islander) that are rejected in year t.
HMDA
WhiteRR
Percentage of mortgage applications from White borrowers that
are rejected in year t.
HMDA
BoardAA
Indicator variable that equals one when the board has at least one
African American director, and zero otherwise, in year t.
ISS
BoardAAP
The percentage of African American directors on the board in year
t.
ISS
SmallSize
Indicator variable that equals one when the total assets in year t-1
is less than $10 million.
Compustat
MediumSize
Indicator variable that equals one when the total assets in year t-1
is between $10 million and $100 million.
Compustat
BoardSize
Natural logarithm of the number of directors on the board in year
t-1.
ISS
BoardFemale
Fraction of female directors on the board in year t-1.
ISS
BoardOtherMinority
Percentage of directors from other minority groups on the board in
year t-1.
ISS
Big3
Percentage share ownership of Blackrock, State Street, and
Vanguard in year t-1.
ISS
AAExecutives
Ratio of African American executives to total executives in the
state where bank is headquartered in year t-1.
EEOC
ROA
Ratio of net income to total assets in year t-1.
Compustat
IVOL
Sqrt(12) multiplied by the standard deviation of monthly excess
stock returns. Excess stocks returns are defined using the Fama-
French three-factor market model estimated over year t-1.
CRSP
Capital
Equity capital plus minority interest less portion of perpetual
preferred stock and goodwill as a percent of adjusted risk-
weighted assets in year t-1.
Compustat
Deposits
Ratio of deposits to total assets in year t-1.
Compustat
RealEstateLoans
Ratio of real estate loans to total assets in year t-1.
FR-Y9C
43
IncDisparity
Ratio of the median household income in the past 12 months of
White households to that of African American households in the
bank’s headquartered state, measured in year t-1.
ACS
LogStateGDP
Natural logarithm of GDP per capita for the bank’s headquarter
state (2012 US dollars) in year t-1.
BEA
FutIVOL
Square root of 12 multiplied by the standard deviation of monthly
excess stock returns. Excess return is defined using the Fama-
French 3-factor market model estimated for years t+1 and t+2.
CRSP
FutROA
Average net income for year t+1 and t+2 divided by total assets in
year t.
Compustat
FutTobinQ
Average market value of equity plus book value of total liabilities
for years t+1 and t+2 divided by total assets in year t.
Compustat
FutBadLoans
Average nonperforming loans for years t+1 and t+2 divided by
total assets at year t.
FR-Y9C
FutBHRet
Buy and holding returns for years t+1 and t+2
CRSP
* Abbreviations: Home Mortgage Disclosure Act (HMDA); Institutional Shareholder Services (ISS);
U.S. Equal Employment Opportunity Commission (EEOC); Consolidated Financial Statements for
Holding Companies Report filed with the U.S. Federal Reserve (FR-Y9C); American Community Survey
of the U.S. Census Bureau (ACS); Bureau of Economic Analysis (BEA); Center for Research in Security
Prices at the University of Chicago (CRSP).
44
Table 2: Sample selection
The initial sample comprises of S&P 1500 financial companies with director racial information from ISS
for the years 2010 to 2021.
Sample Construction
# of bank-
year
observations
# of loan
applications
Financial companies (SIC 6000-6900) with director racial information
for the years 2010 to 2021
3,740
n/a
Dropped if no RSSD is assigned by the Federal Reserve
(2,400)
n/a
Dropped if no mortgage loan application records in HMDA
(266)
n/a
Bank-years with loan data
1,173
44,600,096
Dropped if other control variables are missing from CRSP,
Compustat, Y9C
(228)
(1,940,736)
Dropped if the number of loan applications is less than 1,000
(84)
(40,712)
Resulting sample in bank-years
861
42,618,648
Bank-years with at least 50 African American loan applications
681
41,990,504
Unique bank holding companies (BHCs)
99
n/m
45
Table 3: Summary statistics
The sample comprises of S&P 1500 US banks with available board of directors data (from ISS) and
mortgage loan application data (from HMDA). Data are for the years 2010 to 2021. All variables are defined
in Table 1.
Variable
N
Mean
p25
Median
p75
min
max
Std dev
AfricanAmericanRR
681
.518
.417
.519
.625
.117
1
.146
OtherMinorityRR
681
.44
.341
.437
.531
.123
.997
.136
WhiteRR
681
.346
.27
.341
.408
.09
.991
.114
BoardAA
681
.631
0
1
1
0
1
.483
BoardAAP
681
.067
0
.071
.111
0
.25
.061
SmallSize
681
.266
0
0
1
0
1
.442
MediumSize
681
.512
0
1
1
0
1
.5
BoardSize
681
2.598
2.485
2.639
2.773
1.946
3.296
.22
BoardFemale
681
.174
.111
.167
.231
0
.455
.091
BoardOtherMinority
681
.045
0
0
.071
0
.667
.092
Big3
681
.193
.161
.193
.231
.003
.352
.056
AAExecutives
681
.034
.024
.03
.038
0
.115
.019
ROA
681
.009
.007
.009
.012
-.06
.037
.007
IVOL
681
.152
.097
.127
.172
.043
.698
.094
Capital
681
12.53
10.81
12
13.43
6.75
29.1
2.808
Deposits
681
.734
.706
.76
.804
0
.875
.118
RealEstateLoans
681
.401
.275
.413
.525
.001
.804
.174
IncDisparity
632
.617
.568
.625
.661
.394
.802
.063
LogStateGDP
681
-2.988
-3.125
-2.968
-2.782
-3.668
-2.545
.272
FutIVOL
587
.143
.113
.138
.167
.064
.574
.044
FutROA
529
.012
.009
.011
.014
-.022
.03
.005
FutTobinQ
505
1.193
1.068
1.141
1.245
.84
2.446
.209
FutBadLoans
515
.008
.004
.007
.01
0
.041
.006
FutBHRet
587
.263
.007
.212
.474
-.371
1.958
.334
46
Table 4: Determinants of African American board representation
Table 4 report OLS regression estimates where the dependent variable is BoardAA. All models include year
dummies. Numbers in parentheses are t-statistics based on standard errors clustered at the bank level. ***,
**, and * correspond to 1%, 5%, and 10% significance levels, respectively. All variables are defined in
Table 1.
Model
(1)
(2)
(3)
SmallSize
-0.333***
-0.465***
-0.331**
(-2.650)
(-3.771)
(-2.604)
MediumSize
-0.062
-0.152*
-0.060
(-0.731)
(-1.801)
(-0.703)
BoardSize
0.342**
0.319*
(2.156)
(1.947)
BoardFemale
1.063***
1.078***
(2.733)
(2.717)
BoardOtherMinority
-0.356
-0.276
(-0.652)
(-0.496)
Big3
-0.285
-0.129
-0.347
(-0.331)
(-0.143)
(-0.389)
AAExecutives
5.004**
3.646*
4.666**
(2.286)
(1.878)
(2.108)
ROA
0.098
-3.119
-0.192
(0.035)
(-0.966)
(-0.065)
IVOL
0.196
0.080
0.199
(0.641)
(0.247)
(0.656)
Capital
-0.007
-0.012
-0.008
(-0.394)
(-0.623)
(-0.498)
Deposits
-0.158
-0.210
-0.240
(-0.529)
(-0.675)
(-0.804)
RealEstateLoans
-0.414
-0.492*
-0.402
(-1.617)
(-1.795)
(-1.516)
IncDisparity
-0.177
(-0.317)
LogStateGDP
-0.090
(-0.544)
Constant
-0.231
1.109***
-0.251
(-0.421)
(2.936)
(-0.347)
Observations
681
681
681
Adj. R
2
0.339
0.289
0.341
Year dummies
YES
YES
YES
Clustered se by bank
YES
YES
YES
t-statistic for the difference between
SmallSize & MediumSize coefficients
8.21***
9.30***
8.15***
47
Table 5: African American board representation and mortgage loan rejection rates for African
American borrowers: 2SLS
This table presents 2SLS second stage estimates, where BoardAA is instrumented using AAExecutives,
following Model (1) in Table 4. The dependent variable is AfricanAmericanRR. In column (1)
AfricanAmericanRR is computed based on loan applications for properties outside the bank’s headquarter
state. In column (2), AfricanAmericanRR is computed based on all loan applications. All models include
year dummies. Numbers in parentheses are t-statistics based on standard errors clustered at the bank level.
***, **, and * correspond to 1%, 5%, and 10% significance levels, respectively. All variables are defined
in Table 1.
AfricanAmericanRR
t
(1)
(2)
Sample
Loans outside HQ state
All loans
IVBoardAA
-0.248***
-0.236***
(-2.770)
(-2.743)
SmallSize
-0.150***
-0.149***
(-2.647)
(-3.059)
MediumSize
-0.069*
-0.060*
(-1.888)
(-1.674)
BoardSize
0.034
0.038
(0.530)
(0.662)
BoardFemale
0.351**
0.402***
(2.129)
(2.880)
BoardOtherMinority
0.144
-0.084
(1.354)
(-0.560)
Big3
0.022
-0.039
(0.096)
(-0.175)
ROA
2.506
1.680
(1.521)
(1.086)
IVOL
0.005
-0.087
(0.033)
(-0.760)
Capital
-0.016***
-0.015***
(-3.292)
(-2.893)
Deposits
0.150
0.175*
(1.499)
(1.811)
RealEstateLoans
-0.353***
-0.368***
(-2.789)
(-3.249)
Constant
0.838***
0.861***
(3.649)
(4.128)
Observations
645
681
Adj. R
2
0.128
0.089
Year dummies
YES
YES
Clustered se by bank
YES
YES
F-Stat of IV
5.065
5.224
48
Table 6: African American board representation and mortgage loan rejection rates for African
Americans: Matching methods
This table reports treatment effects using two different matching methods: Kernel Matching and Nearest
Neighbor Matching. The treatment group contains bank-years with at least one African American director
on the board and the control group contains bank-years without any African American directors on the
board. All matching is done with replacement using Model (1) in Table 4. AfricanAmericanRR is calculated
using loan applications from either outside the bank’s headquarter state or from all states, as indicated in
the respective column. Numbers in parentheses are z-statistics based on standard errors adjusted using
10,000 bootstrap replications. ***, **, and * correspond to 1%, 5%, and 10% significance levels,
respectively. All variables are defined in Table 1.
Matching method
Outcome variable
Loans outside HQ state
All loans
Kernel
Mean difference in AfricanAmericanRR
-0.063***
-0.034***
(-3.583)
(-2.660)
Nearest Neighbor
Mean difference in AfricanAmericanRR
-0.063***
-0.044***
(-3.085)
(-2.754)
Number of observations
545
581
49
Table 7: Increase in African American Board representation and mortgage loan rejection rates of African American borrowers:
Univariate analysis
This table reports rejection rates for African American borrowers at banks appointing an African American director. Panel A reports rejection rates
for all banks that appoint an African American director in year 0. Panel B reports rejection rates for banks without any African American directors
that appoint an African American director to the board in year 0. Numbers in parentheses are t-statistics based on t-test of the mean rejection rates
in pre-5 years and post-5 years. ***, **, and * correspond to 1%, 5%, and 10% significance levels, respectively.
Panel A: All banks that appoint an African American director to the board in year 0
Event year
-5
-4
-3
-2
-1
0
1
2
3
4
Mean AfricanAmericanRR
0.60
0.61
0.57
0.56
0.53
0.50
0.50
0.49
0.51
0.49
Number of observations
25
26
34
36
41
47
40
32
27
21
Mean pre & post AfricanAmericanRR
0.57
0.5
Post – Pre AfricanAmericanRR
0.07***
(t-statistic for difference)
(4.47)
Panel B: Banks without any African American directors that appoint an African American director to the board in year 0
Event year
-5
-4
-3
-2
-1
0
1
2
3
4
Mean AfricanAmericanRR rates
0.57
0.57
0.54
0.56
0.51
0.47
0.45
0.48
0.52
0.48
Number of observations
10
11
12
13
18
23
18
14
10
8
Mean pre & post AfricanAmericanRR
0.55
0.48
Post – Pre AfricanAmericanRR
0.07***
(t-statistic for difference)
(3.62)
50
Table 8: Increase in African American Board representation and mortgage loan rejection rates of
African American borrowers: OLS
This tables shows OLS estimates for models with AfricanAmericanRR as the dependent variable, focusing
on sample banks that appoint an African American director. The sample in column (1) includes all banks
that appoint an African American director in year 0. The sample in column (2) includes banks without any
African American directors that appoint an African American director in year 0. For each bank, data are
for the period from year -5 to year +4 relative to the year of appointment (year 0). AfterBoardAA is an
indicator variable that equals one for bank observations for years 0 to 4. ***, **, and * correspond to 1%,
5%, and 10% significance levels, respectively. All variables are defined in Table 1.
(1)
(2)
Sample
All banks that appoint an African
American director
Banks without any African American
directors that appoint an African
American director
AfterBoardAA
-0.097***
-0.099***
(-4.471)
(-3.726)
SmallSize
-0.047
0.146*
(-0.747)
(2.044)
MediumSize
0.025
0.147**
(0.643)
(2.503)
BoardSize
-0.060
-0.016
(-0.916)
(-0.169)
BoardFemale
0.289*
0.423**
(1.836)
(2.162)
BoardOtherMinority
-0.103
0.551
(-0.560)
(0.756)
Big3
-0.111
0.143
(-0.313)
(0.252)
ROA
1.731
-2.407
(1.045)
(-1.094)
IVOL
-0.074
-0.478**
(-0.452)
(-2.806)
Capital
-0.002
0.005
(-0.471)
(0.721)
Deposits
0.197*
0.068
(1.993)
(0.331)
RealEstateLoans
-0.529***
-0.484***
(-6.021)
(-4.737)
Constant
0.806***
0.623*
(3.419)
(1.937)
Observations
329
137
R-squared
0.511
0.533
YEAR FE
Yes
Yes
Cluster BANK
Yes
Yes
51
Table 9: African American loan rejection rates for banks with an African American director, based
on the director’s role on the board.
This table displays rejection rates for African American borrowers at banks with an African American
director, based on the director's role on the board. Key positions include chairman, lead director, or member
of the compensation, nominating, or audit committees. Variables are defined in Table 1.
AfricanAmericanRR
Subsample
N
Mean
Min
p25
p50
p75
Max
Banks with African American
directors in key positions
388
.513
.117
.407
.513
.621
1
Banks with African American
directors not in key position
42
.552
.216
.432
.513
.701
.898
52
Table 10: Mortgage loan applications from CRA areas
The table shows summary statistics for the frequency of loan applications from CRA areas. Panel A shows
the results for the full sample. Panel B shows the results for banks without an African American director.
Panel C shows the results for banks with an African American director.
Panel A: Full Sample
N
Mean
Min
p25
Median
p75
Max
Percentage of loan applications
from CRA areas
681
.32
0
0
0
.845
.988
Percentage of loan applications
of African American borrowers
from CRA areas
681
.014
0
0
0
.019
.155
Panel B: Banks without an African American director
N
Mean
Min
p25
Median
p75
Max
Percentage of loan applications
from CRA areas
251
.545
0
0
.801
.931
.988
Percentage of loan applications
of African American borrowers
from CRA areas
251
.018
0
0
.014
.027
.112
Panel C: Banks with an African American director
N
Mean
Min
p25
Median
p75
Max
Percentage of loan applications
from CRA areas
430
.189
0
0
0
.123
.978
Percentage of loan applications
of African American borrowers
from CRA areas
430
.012
0
0
0
.003
.155
53
Table 11: African American board representation and mortgage loan rejection rates for White
borrowers and other racial minority borrowers
This table presents 2SLS second stage estimates, where BoardAA is instrumented using AAExecutives,
following Model (1) in Table 4. The dependent variable in columns (1) and (2) is WhiteRR, computed using
loan applications of White borrowers from either outside the bank’s headquarter state or from all states, as
indicated in the respective column. The dependent variable in columns (3) and (4) is OtherRR, computed
using loan applications of Other minority borrowers from either outside the bank’s headquarter state of
from all states, as indicated in the respective column. Other minority borrowers are Alaska Native, Asian,
Hispanic, Native American, or Pacific Islanders. All models include year dummies. Numbers in parentheses
are t-statistics based on standard errors clustered at the bank level. ***, **, and * correspond to 1%, 5%,
and 10% significance levels, respectively. All variables are defined in Table 1.
WhiteRR
OtherMinorityRR
(1)
(2)
(3)
(4)
Sample
Loans outside HQ state
All loans
Loans outside HQ State
All loans
IVBoardAA
-0.152**
-0.176***
-0.067
-0.088
(-1.983)
(-2.590)
(-0.685)
(-0.964)
SmallSize
-0.088*
-0.112***
-0.098*
-0.122**
(-1.831)
(-2.690)
(-1.706)
(-2.288)
MediumSize
-0.048
-0.055*
-0.082***
-0.074**
(-1.479)
(-1.820)
(-2.631)
(-2.324)
BoardSize
0.029
0.004
-0.106*
-0.070
(0.567)
(0.090)
(-1.670)
(-1.207)
BoardFemale
0.319**
0.365***
0.265*
0.299*
(2.457)
(3.021)
(1.657)
(1.955)
BoardOtherMinority
0.188*
-0.008
0.258
0.073
(1.683)
(-0.055)
(1.417)
(0.686)
Big3
-0.133
-0.088
0.007
0.060
(-0.780)
(-0.548)
(0.037)
(0.360)
ROA
1.683
1.789
0.304
1.195
(0.884)
(1.004)
(0.158)
(0.767)
IVOL
0.163
0.085
0.142
-0.116
(1.310)
(0.761)
(1.031)
(-1.008)
Capital
-0.008*
-0.012***
-0.012**
-0.010**
(-1.705)
(-2.681)
(-2.013)
(-1.980)
Deposits
0.070
0.018
0.116
0.167*
(0.746)
(0.222)
(0.961)
(1.702)
RealEstateLoans
-0.219*
-0.259**
-0.218*
-0.258**
(-1.865)
(-2.492)
(-1.805)
(-2.255)
Constant
0.471**
0.675***
0.983***
0.929***
(2.423)
(3.892)
(4.301)
(4.563)
Observations
674
681
654
681
Adj. R
2
0.047
0.070
0.188
0.249
Year dummies
YES
YES
YES
YES
Clustered se by bank
YES
YES
YES
YES
54
Table 12: African American board representation and future bank performance and risk
This table reports the estimates of OLS regressions where the dependent variable is the metric identified in
each column. All metrics are computed over the two years following the year in which BoardAA is measured.
Numbers in parentheses are t-statistics based on standard errors clustered at the bank level. ***, **, and *
correspond to 1%, 5%, and 10% significance levels, respectively. All variables are defined in Table 1.
(1)
(2)
(3)
(4)
(5)
Model
FutBHRet
FutROA
FutTobinQ
FutBadLoans
FutIVOL
BoardAA
0.015
0.000
0.037
0.001
0.008*
(0.617)
(0.270)
(1.211)
(0.935)
(1.821)
SmallSize
-0.021
0.001
0.286***
-0.006***
0.020**
(-0.480)
(0.813)
(5.212)
(-3.530)
(2.316)
MediumSize
-0.053*
-0.001
0.112***
-0.005***
0.010
(-1.722)
(-0.713)
(3.594)
(-4.079)
(1.223)
BoardSize
-0.112**
-0.000
0.126*
0.002
0.005
(-2.037)
(-0.065)
(1.765)
(0.993)
(0.440)
BoardFemale
-0.126
-0.010**
-0.362***
-0.001
-0.047**
(-1.030)
(-2.316)
(-2.733)
(-0.261)
(-2.144)
BoardOtherMinority
-0.059
0.002
-0.008
-0.002
0.023
(-0.719)
(0.607)
(-0.087)
(-0.491)
(1.391)
Big3
-0.376
-0.003
-0.130
-0.003
-0.044
(-1.082)
(-0.436)
(-0.360)
(-0.351)
(-0.718)
Capital
0.006
0.000*
0.001
-0.000
0.001
(1.073)
(1.969)
(0.145)
(-0.204)
(0.896)
Deposits
0.118
0.009**
0.245***
0.009**
-0.027
(0.820)
(2.383)
(2.963)
(2.088)
(-1.292)
RealEstateLoans
-0.090
-0.006
-0.312***
0.007*
-0.001
(-1.214)
(-1.473)
(-3.740)
(1.962)
(-0.048)
Constant
0.277
0.003
0.601***
0.006
0.150***
(1.308)
(0.488)
(2.935)
(1.000)
(3.492)
Observations
587
529
505
515
587
R-squared
0.597
0.222
0.287
0.437
0.259
YEAR FE
YES
YES
YES
YES
YES
Cluster BANK
YES
YES
YES
YES
YES
