Declaration Of Melissa Bradley In Support Of: FOR EQUAL RIGHTS, v. FEARLESS FUND
Summary
Document 59-4 in American Alliance for Equal Rights v. Fearless Fund Management, LLC, et al., Case No. 1:23-CV-3424-TWT, in the U.S. District Court for the Northern District of Georgia, filed August 31, 2023. Behind an Exhibit 3 cover page, it is the Declaration of Melissa Bradley in support of defendants' opposition to the plaintiff's motion for preliminary injunction. The declarant, a Professor of Practice at Georgetown University's McDonough School of Business and founder of 1863 Ventures, states she was asked by defense counsel to give expert testimony on racial disparities in financing for Black-owned businesses and is not being paid. Her opinions cite research data on business ownership, startup capital, bank credit, Fintech and PPP lending, and venture capital, with cited materials attached as Exhibits A–Z. The 801-page filing ends with Census Bureau household income tables.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Declaration Of Melissa Bradley In Support Of: FOR EQUAL RIGHTS, v. FEARLESS FUND.”
Original at storage.courtlistener.com. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/american-alliance-for-equal-rights-v-fearless-fund-management-llc-fili-d7a52fa1c4/
, SHA-256 467c08c2e1d2cb860962afac42638a1c043410bb888b0f2990927471810d8ab5.Full text
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 1 of 801
Exhibit 3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 2 of 801
IN THE UNITED STATES DISTRICT COURT
FOR THE NORTHERN DISTRICT OF GEORGIA
ATLANTA DIVISION
AMERICAN ALLIANCE
FOR EQUAL RIGHTS,
Plaintiff,
v.
Case No. 1:23-CV-3424-TWT
FEARLESS FUND
MANAGEMENT, LLC, et al.,
Defendants.
DECLARATION OF MELISSA BRADLEY IN SUPPORT OF
DEFENDANTS’ OPPOSITION TO PLAINTIFF’S MOTION FOR
PRELIMINARY INJUNCTION
i
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 3 of 801
I, Melissa Bradley, have personal knowledge of the facts in this
declaration, and declare as follows:
I. Qualifications
1. I am a Professor of Practice at the McDonough School of
Business at Georgetown University, where I have taught courses on
impact investing, social entrepreneurship, peer-to-peer economies, and
innovation since August 2015. Attached hereto as Exhibit A is a true
and correct copy of my current resume.
2. I have a Bachelor’s of Science Degree in Finance from
Georgetown University, and a Master’s Degree in Business
Administration from the Kogod School of Business at American
University.
3. I have more than 20 years of experience in the
entrepreneurship, investment, and leadership sectors.
4. Currently, I am the Founder and Managing Partner of 1863
Ventures, a Black-led national business development nonprofit
accelerator and venture capital fund for entrepreneurs. I am also a
General Partner of the 1863 Venture Fund I and the Inclusive Innovation
1
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 4 of 801
Equity Impact Fund.
5. I have many different responsibilities at 1863 Ventures. As
relevant to this declaration, I spearhead overall organizational strategy
and oversee the firm’s fundraising efforts. I also strategize with my
colleagues to develop content for the educational and mentoring
programs we offer to our “founders” (i.e., the entrepreneurs who work
with 1863 Ventures). I am in frequent contact with our founders about
their successes and the challenges facing their businesses, and I conduct
trainings on growing and sustaining their businesses, and offer coaching
to them.
6. Both in my work with Georgetown University and 1863
Ventures, I am heavily involved in applied research in the area of
entrepreneurship. This research focuses on the financial cost of being a
diverse entrepreneur, diverse entrepreneurs’ lack of access to capital, and
the variation in business longevity by race. Part of this work involves
analyzing the performance of our founders over time, including assessing
their performance against other similar businesses to better understand
the longitudinal differences between minority and non-minority business
2
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 5 of 801
owners.
7. In 2018, I co-founded Ureeka, a company that provided
“always-on coaching” to minority entrepreneurs who lack such coaching
and mentoring opportunities, including 24/7 access to peers and
colleagues via an app or website. Ureeka was later purchased by Zen
Business.
8. Earlier in my career, I served as a Presidential Appointee in
two Presidential administrations. I first served as a Financial
Regulatory Affairs Fellow at the Department of Treasury’s Office of
Thrift Supervision, where I researched the impact of welfare reform and
microenterprise lending on the financial industry. Later, I served as the
Acting Director of the Social Innovation Fund, a White House initiative
and program of the Corporation for National and Community Service.
9. From 2010 to 2013, I served as CEO of Tides, Inc., an
organization that engages in social impact investing and fiscal
sponsorship of nonprofits to advance social justice.
10. I am also the former Co-Chair and current member of the
National Advisory Council on Innovation and Entrepreneurship, a
3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 6 of 801
federal advisory committee tasked with identifying and recommending
solutions to drive the innovation economy.
II. Scope of Engagement
11. I have been asked by counsel for defendants Fearless Fund
Management, LLC, Fearless Fund II, GP, LLC, Fearless Fund II, LP, and
Fearless Foundation, Inc. (together, “Fearless Fund”) to provide expert
testimony concerning racial disparities in financing for Black-owned
businesses, and the importance of grant funding for starting and scaling
Black-owned businesses.
12. I am not being paid for my services. I have no financial
interest in the outcome of this case.
13. The materials I cite in this testimony are attached as
Exhibits A–Z to my declaration.
III. Opinions
A. Black Americans Have Historically Been Underrepresented
in Business Ownership as Compared to White Americans.
14. It is well-established in research literature and survey data
that Black Americans have historically been underrepresented in self-
4
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 7 of 801
employment and business ownership.1 A research study published in
2019 reported that while white Americans represented nearly 72.9% of
the U.S. population, white Americans owned 78% percent of all private
U.S. businesses.2 Conversely, Black Americans represented
approximately 13.3% of the U.S. population, but owned only 9% of private
U.S. businesses. This data reflects that Black Americans are
disproportionately underrepresented in domestic business ownership
and white Americans are disproportionately overrepresented.3
15. Black-owned businesses also tend to be smaller than white-
owned businesses and less likely to have any employees.4 According to
the Census Bureau’s 2021 Survey of Business Owners, 96% of Black-
1 Ex. B, David G. Blanchflower & Jon Wainwright, An Analysis of the Impact of Affirmative Action
Programs on Self-Employment in the Construction Industry 1, 17 (Nat’l Bureau of Econ. Rsch.,
Working Paper No. 11793, 2005).
2 Ex. C, M’BALOU CAMARA ET AL., ENTERING ENTREPRENEURSHIP: RACIAL DISPARITIES IN THE PATHWAYS
INTO BUSINESS OWNERSHIP 3 (2019).
3 According to the Brookings Institution, in 2020, Black people represented 14.2% of all Americans but
only 2.4% of all employer-firm owners. Latino or Hispanic people represented 18.7% of the population
and 6.5% of employer-firm owners, while Asian Americans represented 6% of the population and 10.6%
of employer-firm owners. This data shows that Black Americans are most significantly
underrepresented even among minority groups. See Ex. D, Andre M. Perry et al., Who Is Driving
Black Business Growth? Insights from the Latest Data on Black-Owned Businesses, BROOKINGS
INST. (May 24, 2023), available at https://www.brookings.edu/articles/who-is-driving-black-business-
growth-insights-from-the-latest-data-on-black-owned-
businesses/#:~:text=One%20key%20finding%20pertains%20to,seemed%20to%20drive%20overall%20
growth.
4 Ex. E, MCKINSEY & CO., BUILDING SUPPORTIVE ECOSYSTEMS FOR BLACK-OWNED US BUSINESSES 2
(2020).
5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 8 of 801
owned companies have no employees versus 78% of white-owned
businesses.
B. Black Americans Have Faced, and Continue to Face,
Considerable Barriers in Accessing Capital to Start and
Scale Their Businesses as Compared to White Americans.
16. One of the reasons for the disparities in prevalence and size
of Black-owned and white-owned businesses is that Black Americans
have historically faced, and continue to face, significant barriers
accessing capital to start and grow their businesses. Across all
industries, a larger share of Black-owned businesses have inadequate
initial startup capital as compared to white-owned businesses.5 The
average level of startup capital among Black entrepreneurs is $35,205
compared with $106,720 for white entrepreneurs.6 As sources including
the U.S. Census Bureau’s Annual Survey of Entrepreneurs, the Federal
Reserve’s Survey of Small Business Finance, and the Kauffman
Foundation make clear, Black-owned businesses are more reliant on
5 Id. at 10.
6 Ex. F, Robert W. Fairlie et al., Black and White: Access to Capital Among Minority Owned Startups
11 (Nat’l Bureau of Econ. Rsch., Working Paper No. 28154, 2020).
6
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 9 of 801
owner investment, and less able to access bank financing, than white-
owned Businesses.7 This is true even when controlling for differences in
factors such as credit scores, industry, owner experiences,8 and education
levels.9 This disparity is a significant obstacle hindering Black
Americans attempting to start businesses. Without the right inflows of
capital, a business cannot scale in the way that it needs to in order to be
successful in the long term. For example, purchasing inventory, fulfilling
orders, marketing, covering overhead expenses, hiring additional
employees and even conducting research and development to improve
goods and services requires a significant amount of capital especially in
the initial stages of a business.
17. Outside capital such as bank financing and debt financing
more generally, is a fundamentally important source of capital for young
companies.10 The Kauffman Foundation Research Series on Firm
Formation and Economic Growth consists of reports that explore the
7 Ex. G, Timothy Bates & Alicia Robb, Impacts of Owner Race and Geographic Context on Access to
Small Business Financing, 30 ECON. DEV. Q. 159, 159, 161–63 (2016).
8 Id. at 166.
9 Ex. C at 9.
10 Ex. H, William R. Kerr et al., Entrepreneurship as Experimentation, 28 J. ECON. PERSPECTIVES 25,
41 (2014).
7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 10 of 801
relationship between firm formation and economic growth in the United
States from a variety of angles. In one of its reports analyzing nearly
5000 businesses, the Kauffman Firm Survey found that within the first
three years of founding, 40% of the startup funding is typically
constituted by outside debt capital, and only 4% consists of an owner’s
credit cards and personal loans.11 However, studies have shown that
Black entrepreneurs borrow only half as much as they contribute from
their own capital, while white entrepreneurs on average borrow nearly
1.7 times what they contribute from their own capital.12 This data shows
that there’s a significant disparity in the amount of non-personal capital
that Black entrepreneurs are able to rely on in their ventures compared
to white entrepreneurs, which impedes the ability of Black entrepreneurs
to scale and grow their businesses.
18. To understand why Black businesses are more reliant on
owner investment, it is important to understand the funding ecosystem
in which Black entrepreneurs operate. One traditional path for funding
a new business is to rely first on friends and family for startup capital.
11 Id.
12 Ex. F at 12.
8
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 11 of 801
But when Black entrepreneurs seek these funds, they are disadvantaged
compared to their white counterparts, given this country’s well-
documented wealth and income gap. For instance, 2019 Federal Reserve
data reflects that the typical white family has eight times the wealth of
the typical Black family.13 And in 2021, the U.S. Census Bureau reported
that the median household income for Black Americans was just $48,297,
compared to $77,999 for non-Hispanic white Americans.14
19. This wealth and income gap also affects a Black business
owner’s ability to self-fund their venture, which can be exacerbated by
student loan debt. Black Americans, on average, have higher rates of
student loan debt than other racial groups. The 2023 Education Data
Initiative study of student loan debt statistics revealed that Black college
graduates owe $25,000 more in student loan debt, on average, compared
to white college graduates. The study also found that Black students
13 Ex. I, Neil Bhutta et al., Disparities in Wealth by Race and Ethnicity in 2019 Survey of Consumer
Finances, BD. OF GOVERNORS OF THE FED. RESERVE SYS., (Sept. 28, 2020), available at
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-
in-the-2019-survey-of-consumer-finances-20200928.html.
14 Ex. J, JESSICA SEMEGA & MELISSA KOLLAR, U.S. DEP’T OF COMMERCE, U.S. CENSUS BUREAU,
INCOME IN THE UNITED STATES: 2021 at 2 (Sept. 2022).
9
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 12 of 801
borrowed 188% more in student loans than white students.15 Student
loan debt reduces the amount of discretionary income Black
entrepreneurs have to invest in their ventures.
20. With less access to funding from friends and family, and less
discretionary income of their own, Black entrepreneurs must rely on
other private sources of funding—including bank financing and venture
capital—to start and scale their businesses. But here too, Black
entrepreneurs face significant barriers to access. The 2021 Small
Business Credit Survey conducted by the 12 Federal Reserve Banks
found that more than half of Black business owners who sought credit
reported receiving less than the amount they requested. Only a quarter
of white business owners reported the same. Black Americans are also
charged more to borrow money, paying higher interest rates on their
debt.16 Confronted with these hurdles to accessing bank financing,
Black-owned businesses are more likely to not apply for or to receive a
15 Ex. K, Melanie Hanson, Student Loan Debt By Race, EDUC. DATA INITIATIVE (May 17, 2023),
available at https://educationdata.org/student-loan-debt-by-race.
16 Ex. L, David Blanchflower et al., Discrimination in the Small Business Credit Market 20, 26 (Nat’l
Bureau of Econ. Rsch., Working Paper No. 6840, 1998).
10
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 13 of 801
traditional loan than white businesses, losing out on a significant source
of startup capital for their businesses.17
21. In lieu of traditional bank financing that offers lower interest
rates, Black business owners often turn to more costly sources of startup
funding, including credit cards and Fintech financing.18 More costly debt
is a drain on the already limited capital Black businesses have to grow.
Fintech lending is automated and anonymized, and the lender therefore
typically does not know the race of the applicant. Given this anonymity,
Fintech lenders like Paypal can be more attractive to Black-owned
businesses than traditional financial institutions.19 In fact, 25% of Black-
owned businesses applied for PPP loans from Fintech lenders, and Black-
owned businesses turned to Fintech firms at twice the rate of those owned
17 Ex. M, ALICIA ROBB, OFFICE OF ADVOCACY, U.S. SMALL BUSINESS ADMIN., FINANCING PATTERNS AND
CREDIT MARKET EXPERIENCES: A COMPARISON BY RACE AND ETHNICITY FOR U.S. EMPLOYER FIRMS 4, 10,
12, 49, (Feb. 2018).
18 Ex. N, Aaron K. Chatterji & Robert C. Seamans, Entrepreneurial Finance, Credit Cards and Race,
106 J. FIN. ECON. 182, 182–83 (2012).
19 Ex. O, Celine Yue Fei, What Drives Racial Minorities to Use Fintech Lending? Evidence from a
Structural Estimation 4 (Sept. 2022), available at
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949148.
11
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 14 of 801
by white, Hispanic, and Asian individuals.20 However, Fintech funders
such as Lending Club tend to have higher interest rates than traditional
funding sources such as banks, small business loans and credit unions
(often twice as high),21 and as a result, are a more costly source of capital
for the businessowner. For instance, when a majority of a company’s
revenue and profits are allocated towards paying off high interest debt, a
business will inevitably struggle with maintaining its cashflow sufficient
to expand its operations.
22. Another important source of business funding is venture
capital. Venture capital is a type of financing that private investors
provide to startups and small businesses that are believed to have
potential to show long-term growth. But there, too, is a wide racial divide
in the investments to Black-owned versus white-owned businesses. 2022
Crunchbase data on venture capital funding shows that Black startup
20 Ex. P, Jessica Battisto et al., Who Received PPP Loans by Fintech Lenders, FED. RESERVE BANK OF
N.Y. (May 27, 2021), https://libertystreeteconomics.newyorkfed.org/2021/05/who-received-ppp-loans-
by-fintech-lenders/.
21 Ex. Q, Julapa Jagtiani & Catharine Lemieux, Fed. Reserve Bank of Phila., Fintech Lending:
Financial Inclusion, Risk Pricing, and Alternative Information 9–10 (June 16, 2017), available at
https://www.fdic.gov/analysis/cfr/bank-research-conference/annual-17th/papers/14-jagtiani.pdf.
12
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 15 of 801
founders in the United States raised only around $264 million out of the
total $33.6 billion in venture capital allocated in Q4 2022.22 In total, U.S.
Black founders raised only an estimated $2.254 billion out of the $215.9
billion in U.S. venture capital allocated last year, which amounts to only
about 1% of total venture capital funding.23
23. Notably, while there was an uptick in corporate commitment
to funding Black-led startups in 2021 following the murder of George
Floyd and the national attention that followed, that attention to funding
Black businesses was not sustained. In 2020, America’s Fortune 1000
companies committed $66 billion to racial injustice, but so far, few of
these commitments have been met, and the pace of monetary
commitments has slowed year-over-year.24 Compounding this problem,
financing for Black business owners dropped 45% in 2022 due to
22 Ex. R, Dominic-Madori Davis, Black Founders Still Raised Just 1% of All VC Funds in 2022,
TECHCRUNCH (Jan. 6, 2023), available at https://techcrunch.com/2023/01/06/black-founders-still-
raised-just-1-of-all-vc-funds-in-
2022/?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQ
AAADOeO0N4FG9XtUygcJY5N-
nUq5sm_fqNg1IR2MSNrL2LjspqwJZIYI_scQ2TATxWnPF0Gsu9MNN23ITxULN4OW3x5AshW9FF
Vi8P4f-bF7_I5XM_IL1-EQMtqG5ln2Eci-NIZg8AfKQNS15Ci8ixyWRixI5mLWrabO-oWE6J0tQk.
23 Id.
24 Ex. S, MEGAN ARMSTRONG ET AL., MCKINSEY & CO., CORPORATE COMMITMENTS TO RACIAL JUSTICE:
AN UPDATE 2–3 (Feb. 2023)
13
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 16 of 801
changing market conditions.25
C. Black Female Entrepreneurs Are Especially Disadvantaged in
Accessing Capital Markets.
24. Black women are arguably the most economically
disadvantaged demographic in the U.S. In a report by compensation data
and software company PayScale, it was reported that Black women begin
to earn more than white men with an associate degree only when they
obtain a master’s degree.26 In other words, Black women must earn three
times as many degrees as a white man to earn the same pay, let alone
aim for slightly more.
25. Against this backdrop of being highly disadvantaged in the
workforce, Black women have begun to gravitate toward
entrepreneurship as a means toward economic freedom. Black women
are starting businesses at a record pace—between 2014 and 2019, 42% of
25 Ex. T, Gabrielle Fonrouge, Venture Capital for Black Entrepreneurs Plummeted 45% in 2022, Data
Shows, CNBC (Feb. 2, 2023), https://www.cnbc.com/2023/02/02/venture-capital-black-founders-
plummeted.html.
26 Ex. U, Stephen Miller, Black Workers Still Earn Less Than Their White Counterparts, SHRM
(June 11, 2020), https://www.shrm.org/resourcesandtools/hr-topics/compensation/pages/racial-wage-
gaps-persistence-poses-challenge.aspx.
14
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 17 of 801
new businesses were created by Black women.27 However, they are
significantly underrepresented in ownership of mature businesses (i.e., a
business surviving past 5 years) with just 3% of Black women running
mature businesses.28 And although Black women make up roughly 6% of
the U.S. population, they own just 2% of employer businesses (i.e.,
businesses with more than one employee).29
26. What I know from my own experience, and what is supported
by the data, is that challenges accessing capital play a significant role in
limiting the ability of Black women to grow their businesses.30 According
to the Federal Reserve System’s 2016 Small Business Credit Survey, 66%
of Black women business owners identify access to credit and funds for
expansion as one of their greatest financial challenges.31 Only 39% of
their nonminority peers say the same. In fact, despite making up 20.3%
27 Ex. V, Fearless Fund 2023 Impact Report at 11.
28 Ex. W, Donna Kelley et al., Black Women Are More Likely to Start a Business than White Men,
Harvard Business Rev., (May 2021), available at https://hbr.org/2021/05/black-women-are-more-
likely-to-start-a-business-than-white-men.
29 Ex. X, GIZELLE GEORGE-JOSEPH & DANIEL MILO, GOLDMAN SACHS, BLACK WOMENOMICS: EQUALIZING
ENTREPRENEURSHIP 3 (Feb. 9, 2022).
30 Id.
31 Ex. Y, DELL GINES, FED. RESERVE BANK OF KANSAS CITY, BLACK WOMEN BUSINESS STARTUPS 12
(2017).
15
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 18 of 801
of the U.S. population, Black women received less than 0.4% in venture
capital funding in 2022.32 This is wholly consistent with my experience
working in this industry for over twenty years both as a Black female
founder and as Managing Partner at 1863 Ventures, where I regularly
interact with Black entrepreneurs who report to me on the hurdles they
face accessing capital for their businesses.
27. Even within the venture capital industry, Black women face
roadblocks to accessing private capital. For instance, Black female
business owners who apply for funding face a three-times higher rejection
rate than that of white business owners. And although U.S. startups
raised $329.9 billion in venture investments in 2021, nearly doubling the
previous record of $166.6 billion in 2020, Black female founders raised
just 0.13% of the total venture capital spend in 2022.33
28. A biennial study on Black and Latinx woman founders
conducted in January 2021 by digitalundivided found that only 93 Black
women founders had ever raised over $1 million in venture capital for
32 Ex. V at 9.
33 Id. at 13.
16
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 19 of 801
their businesses.34 And among the vast majority of Black women
founders who have not met that milestone, the median amount of funds
raised in the founder’s seed round was only $125,000 compared to the
national average of $2.5 million.35 And prior to 2018, just 34 Black women
founders had raised over $1 million, according to the most recent findings
by Project Diane.36 These numbers further reflect how Black female-led
firms are struggling to access meaningful capital sufficient to sustain and
scale a business.
D. Racial Discrimination is One of the Reasons Black
Entrepreneurs Face Challenges in Accessing Capital
Markets.
29. Research shows that racial discrimination is one reason why
Black entrepreneurs face barriers in accessing the capital they need to
start and grow their businesses.
30. A 2012 study found that Black entrepreneurs were more
34 Ex. Z, Dominic-Madori Davis, The VC Landscape is Dire for Black Women. 7 Black Women Who
Have Raised Millions in Funding Reveal Their Best Fundraising Advice, BUS. INSIDER (Nov. 10, 2021),
available at https://www.businessinsider.com/how-to-raise-venture-capital-as-black-female-founder-
2021-11.
35 Id.
36 Id.
17
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 20 of 801
likely to fear being turned down by lending institutions than white
entrepreneurs in states with higher levels of historical racial inequality.37
In particular, the study revealed that local areas with high levels of
historical racial inequality have much higher levels of fear of denial and
unmet capital needs among Black entrepreneurs relative to white
entrepreneurs than areas with low levels of racial inequality.38
31. Other studies have shown that Black entrepreneurs fare
poorly in raising capital even in areas with a strong local banking
community that is more favorable toward small startups.39 Researchers
attributed this result to discrimination or banks treating Black
entrepreneurs differently because Black small businesses should
presumably fare better with smaller local banks that rely on “soft”
subjective information than larger national banks that tend to rely on
objectively “hard” information like credit scores. But that was not the
case, and the root cause of the disparity appeared to be the negative
attitudes and perceptions about Black borrowers in credit markets.40
37 Ex. N at 184, 190, 193–94.
38 Id. at 184.
39 Ex. F at 5.
40 Id. at 24–26.
18
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 21 of 801
Therefore, while minorities in general face some form of discrimination
in the U.S., the negative attitudes toward Black borrowers in particular
have significantly impacted Black entrepreneurs’ ability to borrow and
scale their businesses in the long term.
E. Grant Funding Is an Important Source of Capital for Black
Businesses.
32. Confronted with these documented barriers to accessing
capital, Black startups look to grant funding—like the Fearless
Foundation grants at issue in this case—to grow their businesses. I have
been involved in many small-business grant programs in my career. In
my experience, grant funds are particularly important for Black
entrepreneurs who use this capital to fill gaps in funding for their
businesses that they cannot access from traditional sources like family,
friends, private credit lenders, and venture capital. For instance, grant
recipients often apply their awards toward marketing to improve the
visibility of their businesses. Marketing is typically one of the most
expensive parts of a startup venture, but it is also hard to secure
financing for marketing expenses through traditional debt because it is
effectively a sunk cost.
19
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 22 of 801
33. Grant funding also often comes with meaningful intangible
benefits for small businessowners. In my work, I have been involved in
grant programs that provide coaching, mentorship, and financial
guidance to grant recipients. For example, under my leadership, Ureeka
coached and trained grant recipients on customer acquisition and
customer retention in a post-COVID virtual world. These intangible
benefits of grant programs are especially helpful for Black entrepreneurs,
many of whom lack role models, mentors, or business leaders in their
communities who they can turn to for guidance and advice in their
entrepreneurial journeys.
34. Additionally, grant programs often connect Black
businessowners with peers at various stages of business development,
thus helping to foster leadership networks that may not otherwise exist
for many Black entrepreneurs. Black founders rely on these networks
for advice and guidance on scaling their businesses.
20
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 23 of 801
IV. Materials Cited
35. Attached hereto as Exhibit A is a true and correct copy of my
current resume.
36. Attached hereto as Exhibit B is a true and correct copy of
David G. Blanchflower & Jon Wainwright, An Analysis of the Impact of
Affirmative Action Programs on Self-Employment in the Construction
Industry (Nat’l Bureau of Econ. Rsch., Working Paper No. 11793, 2005).
37. Attached hereto as Exhibit C is a true and correct copy of
M’BALOU CAMARA ET AL., ENTERING ENTREPRENEURSHIP: RACIAL
DISPARITIES IN THE PATHWAYS INTO BUSINESS OWNERSHIP (2019).
38. Attached hereto as Exhibit D is a true and correct copy of
Andre M. Perry et al., Who is Driving Black Business Growth? Insights
from the Latest Data on Black-Owned Businesses, BROOKINGS INST. (May
24, 2023), available at https://www.brookings.edu/articles/who-is-
driving-black-business-growth-insights-from-the-latest-data-on-black-
owned-
businesses/#:~:text=One%20key%20finding%20pertains%20to,seemed%
20to%20drive%20overall%20growth.
21
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 24 of 801
39. Attached hereto as Exhibit E is a true and correct copy of
MCKINSEY & CO., BUILDING SUPPORTIVE ECOSYSTEMS FOR BLACK-OWNED
US BUSINESSES (2020).
40. Attached hereto as Exhibit F is a true and correct copy of
Robert W. Fairlie et al., Black and White: Access to Capital Among
Minority-Owned Startups (Nat’l Bureau of Econ. Rsch., Working Paper
No. 28154, 2020).
41. Attached hereto as Exhibit G is a true and correct copy of
Timothy Bates & Alicia Robb, Impacts of Owner Race and Geographic
Context on Access to Small-Business Financing, 30 ECON. DEV. Q. 159
(2016).
42. Attached hereto as Exhibit H is a true and correct copy of
William R. Kerr et al., Entrepreneurship as Experimentation, 28 J. ECON.
PERSPECTIVES 25 (2014).
43. Attached hereto as Exhibit I is a true and correct copy of Neil
Bhutta et al., Disparities in Wealth by Race and Ethnicity in 2019 Survey
of Consumer Finances, BD. OF GOVERNORS OF THE FED. RESERVE SYS.,
(Sept. 28, 2020), available at
22
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 25 of 801
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-
wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-
20200928.html.
44. Attached hereto as Exhibit J is a true and correct copy of
JESSICA SEMEGA & MELISSA KOLLAR, U.S. DEP’T OF COMMERCE, U.S.
CENSUS BUREAU, INCOME IN THE UNITED STATES: 2021 (Sept. 2022).
45. Attached hereto as Exhibit K is a true and correct copy of
Melanie Hanson, Student Loan Debt By Race, EDUC. DATA INITIATIVE
(May 17, 2023), available at https://educationdata.org/student-loan-debt-
by-race.
46. Attached hereto as Exhibit L is a true and correct copy of
David Blanchflower et al., Discrimination in the Small Business Credit
Market (Nat’l Bureau of Econ. Rsch., Working Paper No. 6840, 1998).
47. Attached hereto as Exhibit M is a true and correct copy of
ALICIA ROBB, OFFICE OF ADVOCACY, U.S. SMALL BUSINESS ADMIN.,
FINANCING PATTERNS AND CREDIT MARKET EXPERIENCES: A COMPARISON
BY RACE AND ETHNICITY FOR U.S. EMPLOYER FIRMS (Feb. 2018).
48. Attached hereto as Exhibit N is a true and correct copy of
23
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 26 of 801
Aaron K. Chatterji & Robert C. Seamans, Entrepreneurial Finance,
Credit Cards and Race, 106 J. FIN. ECON. 182 (2012).
49. Attached hereto as Exhibit O is a true and correct copy of
Celine Yue Fei, What Drives Racial Minorities to Use Fintech Lending?
Evidence from a Structural Estimation (Sept. 2022), available at
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3949148.
50. Attached hereto as Exhibit P is a true and correct copy of
Jessica Battisto et al., Who Received PPP Loans by Fintech Lenders, FED.
RESERVE BANK OF N.Y. (May 27, 2021),
https://libertystreeteconomics.newyorkfed.org/2021/05/who-received-
ppp-loans-by-fintech-lenders/.
51. Attached hereto as Exhibit Q is a true and correct copy of
Julapa Jagtiani & Catharine Lemieux, Fintech Lending: Financial
Inclusion, Risk Pricing, and Alternative Information, FED. RESERVE BANK
OF PHILA. (June 16, 2017), available at
https://www.fdic.gov/analysis/cfr/bank-research-conference/annual-
17th/papers/14-jagtiani.pdf.
52. Attached hereto as Exhibit R is a true and correct copy of
24
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 27 of 801
Dominic-Madori Davis, Black Founders Still Raised Just 1% of All VC
Funds in 2022, TECHCRUNCH (Jan. 6, 2023), available at
https://techcrunch.com/2023/01/06/black-founders-still-raised-just-1-of-
all-vc-funds-in-
2022/?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmN
vbS8&guce_referrer_sig=AQAAADOeO0N4FG9XtUygcJY5N-
nUq5sm_fqNg1IR2MSNrL2LjspqwJZIYI_scQ2TATxWnPF0Gsu9MNN2
3ITxULN4OW3x5AshW9FFVi8P4f-bF7_I5XM_IL1-EQMtqG5ln2Eci-
NIZg8AfKQNS15Ci8ixyWRixI5mLWrabO-oWE6J0tQk.
53. Attached hereto as Exhibit S is a true and correct copy of
MEGAN ARMSTRONG ET AL., MCKINSEY & CO., CORPORATE COMMITMENTS
TO RACIAL JUSTICE: AN UPDATE (Feb. 2023).
54. Attached hereto as Exhibit T is a true and correct copy of
Gabrielle Fonrouge, Venture Capital for Black Entrepreneurs Plummeted
45% in 2022, Data Shows, CNBC (Feb. 2, 2023),
https://www.cnbc.com/2023/02/02/venture-capital-black-founders-
plummeted.html.
55. Attached hereto as Exhibit U is a true and correct copy of
25
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 28 of 801
Stephen Miller, Black Workers Still Earn Less Than Their White
Counterparts, SHRM (June 11, 2020),
https://www.shrm.org/resourcesandtools/hr-
topics/compensation/pages/racial-wage-gaps-persistence-poses-
challenge.aspx.
56. Attached hereto as Exhibit V is a true and correct copy of the
2023 Fearless Fund Impact Report.
57. Attached hereto as Exhibit W is a true and correct copy of
Donna Kelley et al., Black Women Are More Likely to Start a Business
than White Men, HARV. BUS. REV., (May 2021), available at
https://hbr.org/2021/05/black-women-are-more-likely-to-start-a-
business-than-white-men.
58. Attached hereto as Exhibit X is a true and correct copy of
GIZELLE GEORGE-JOSEPH & DANIEL MILO, GOLDMAN SACHS, BLACK
WOMENOMICS: EQUALIZING ENTREPRENEURSHIP (Feb. 9, 2022).
59. Attached hereto as Exhibit Y is a true and correct copy of
DELL GINES, FED. RESERVE BANK OF KANSAS CITY, BLACK WOMEN
26
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 29 of 801
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 30 of 801
CERTIFICATE OF SERVICE
I hereby certify that on August 31, 2023, I electronically filed the
foregoing document with the Clerk of Court using the Court’s CM/ECF
system, which will automatically send email notification of such filing to
all counsel of record.
/s/ Mylan L. Denerstein
Mylan L. Denerstein
28
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 31 of 801
Exhibit A
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 32 of 801
MELISSA LYNN BRADLEY
4200 Wisconsin Avenue, NW, #106-123, Washington, DC 20016
melissabradley@mac.com
(917) 549-6092
Education
1991-1993 AMERICAN UNIVERSITY WASHINGTON, DC
Master’s in Business Administration, Marketing
Graduate Fellow
1985-1989 GEORGETOWN UNIVERSITY WASHINGTON, DC
Bachelor of Science, Finance
Academic
Experience
2015-Present Georgetown University McDonough School of Business
Professor of Practice, Management Department
Awards: Entrepreneurship Faculty Excellence Award, the Joseph F. LeMoine Award for
Undergraduate and Graduate Teaching Excellence, the Peter W. Gonzalez, Jr. Award for
Excellence in Adjunct Faculty Teaching, and The Ideas Worth Teaching Award
2016-2018 American University, Kogod School of Business
Professor, Management Department and Founder, AU Center for Innovation
2021-2022 Spelman College & Morehouse University
Founding Director, Center for Black Entrepreneurship
Original
Research The ROI of D&I: The Cost of Being A Diverse Entrepreneur, 2018.
Beyond 5: Small Business Survivability of New Majority Entrepreneurs, 2022.
Publications
Books
Melissa L. Bradley. Introduction to Media Investing For The Creative and Investment
Communities. New York City: New Capitalist, 2004.
Melissa L. Bradley. A Young Entrepreneur’s Casebook. Washington, DC: The
Entrepreneurial Development Institute (TEDI), 1995.
Melissa L. Bradley. Business Basics: A Young Entrepreneur’s Guide. Washington, DC: The
Entrepreneurial Development Institute (TEDI), 1995.
Articles in Journals
Melissa L. Bradley. "Instructions Included: An Open Letter to Future Change
Agents." National Civic Review, 86, 3 (1997): 245-249.
Melissa L. Bradley. "Reducing Recidivism through Entrepreneurship." State of Corrections,
American Correctional Association (1996)
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 33 of 801
Articles in Conference/Workshop Proceedings (1)
Melissa L. Bradley. "Individual Development Accounts (IDAs): Strategy for Asset
Accumulation.”, San Francisco, CA: Office of Thrift Supervision, 1998.
Presentations
“Total Portfolio Activation: A Framework for Creating Social and Environmental Impact
Across Asset Classes.” 2012. Trillium Asset Management.
Other
Melissa L. Bradley, Brad Dudding, Zachary Wenner. ''Thriving in an Outcomes-Based
Market.'' (2015).
Ascend at the Aspen Institute. ''The Bottom Line, Investing for Impact on Economic Mobility
in the U.S.'' (2014).
2
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 34 of 801
Professional UREEKA Inc. WASHINGTON, DC /SAN FRANCISCO, CA
Experience Venture-backed platform to democratize opportunity via coaching and capital to small
business owners
2019-2022 CO-FOUNDER
Built and launched a tech platform to reduce risk and costs for small business
growth
Secured over $12M+ in venture capital
Responsible for organizational strategy, small business recruitment/retention, and
partnerships
Manage a community of over 15k entrepreneurs globally
Provide leadership to 20+ remote staff and serve on the company’s Board of
Directors
Sold the company to Zen Business in July 2022
2016-Present 1863 VENTURES WASHINGTON, DC
Business development program to accelerate New Majority entrepreneurs to create $100B in
new wealth
FOUNDER & MANAGING PARTNER and General Partner
Founded organization to accelerate New Majority entrepreneurs from high potential to
high growth
Responsible for organizational strategy, financial sustainability, governance, and
partnerships
Serve as General Partner for affiliated revenue-based financing fund, 1863 Venture
Fund I and Inclusive Innovation Equity Impact Fund
Manage and raise a budget of $3M+ annually
2013-2015 CORPORATION FOR NATIONAL AND COMMUNITY SERVICE WASHINGTON, DC
Federal agency that promotes and supports AmeriCorps programs across the United States
ACTING DIRECTOR, SOCIAL INNOVATION FUND
White House Initiative and program of the Corporation for National and Community Service
(CNCS)
Led the creation of a departmental strategic plan
Co-developed the FY 2015 NOFA releases for SIF and Pay for Success
Developed SIF Week – a national campaign to promote the SIF and evidence-based
programs
CHIEF STRATEGY OFFICER
Presidential Appointee responsible for the agency’s strategic plan and creating and sustaining
inter-agency and private sector partnerships.
Secured a $10M partnership between CNCS, Citi Foundation, and Points of Light to create
the nation’s largest corporate-sponsored AmeriCorps program in 10 cities nationwide
Serve as the agency representative on the President’s “My Brother’s Keeper” Initiative and
the Interagency Working Group on Opportunity Youth
Co-created an investment mechanism to secure and scale private sector support for the
education awards of AmeriCorps members
Detailed to the Department of Education to support the President’s My Brother’s Keeper
Initiative, designed to bring attention to, and support for decreasing, the disparities facing
boys and young men of color. As part of the launch of the President’s initiative, authored
the chapter on mentoring, launched the national engagement call to action for mentors,
served as the agency lead on the interagency task force, and identified public-private
partnerships to support this important work which included a $13M commitment from
AT&T.
3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 35 of 801
2010-2013 TIDES SF, CA; NYC, NY; WASHINGTON, DC
International social enterprise providing sophisticated financial and management service to
donors/doers
CEO
Managed over 120 staff in four locations with an annual budget over $20MM and $200MM
AUM
Led innovation through new products, services and partnerships that doubled donors in
one year
Initiated customer service and business model improvements for client base of over 500
donors/activists
Initiated Network rebranding campaign as well as a transition to a multi-tenancy IT
platform
Generated highest contributions since 2000 and produced break-even budget after 5 years
of losses
Served as first professional CEO following the founder’s departure after 35 years
1999-Present NEW CAPITALISTTM WASHINGTON, DC
A business development firm specializing in media, technology, and financial services
industries
Founder & President
Facilitated over $20MM in venture capital transactions for seed stage companies with 20
percent ROI
Created proprietary investment vehicles instrumental in capital sourcing for minority-owned
firms
Provide strategic planning, marketing, and business development services to early-stage
companies
Offer advisory services to investors – institutions, individuals, and philanthropic
organizations
Generate and manage over $1MM in business development annually
2008-2010 GREEN FOR ALL OAKLAND, CA
National policy organization committed to an inclusive green economy
Senior Strategist, Capital Access Program (CAP)
Established the Energy Efficiency Opportunity Fund (EEOF) in partnership with Living Cities
Co-created the Green Jobs Award Program with SJF Ventures and Advisory Services
Created online entrepreneurship center and offline business academies to support green
entrepreneurship, particularly for minority and women-owned firms
Researched and developed policy recommendations to support green business
2004-2008 REENTRY STRATEGIES INSTITUTE™ WASHINGTON, DC
National criminal justice intermediary explicitly focused on reentry
Founder & President
Selected as a Fellow by the Open Society Institute and the Draper Richards Foundation
Established partnerships with the North Carolina Department of Corrections, the Office of
the District Attorney of San Francisco, and the Criminal Justice Coordinating Council in
Washington, DC
Supported over 50 grassroots organizations in capacity building and sustainability planning
Hosted four National Dialogues on Reentry and Criminal Justice in New York, Chicago, San
Francisco, and North Carolina and launched the Reentry Donors Network
4
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 36 of 801
2006-2008 CIT GAP FUNDS HERNDON, VIRGINIA
Providing seed-stage equity investments in Virginia-based technology and life science
companies
Director, Investment Services
Responsible for sourcing regionally based deals in sectors including IT, Biotech, Healthcare,
Consumer Services, Mobile, Financial Services, etc.
Conducted all due diligence and completed all investment documentation for investment
committee presentations on a quarterly basis
Led all portfolio management activities for invested companies including board-level
relationships
2004-2006 POSITIVE IMPACT™ WASHINGTON, DC & NEW YORK, NY
Collaborative initiative to promote diverse voices and visions within independent media
Founder & President
Successfully launched initiative at 2004 Sundance Film Festival
Secured 15 project partners including Sundance Film Festival & Sundance Institute, NY
Latino Film Festival, IFP-NY, Screen Actors Guild, OneVibe TV/Film Group, LLC, etc.
Facilitated over 50 investor meetings that supported over $1MM in investments to media
makers of color
Established Media Training Fund (MTF) grant program to assist media makers of color
participate in training and professional development programs which has provided over
$20,000 in grants
2001-2003 UBS NEW YORK, NY
Private Client Services
Vice President
Recruited as a Marketing Associate and promoted to Assistant Vice President within one
year
Selected to participate in internal leadership development training program
Managed the development and re-launch of the private client website for high-net-worth
clients and received internal recognition for leadership in the re-design process
Developed and implemented training for top ranking financial advisors and select private
clients to increase assets under management to $1 trillion
Responsible for site management/maintenance, client interface, content development,
vendor sourcing and relationship management for private client website
1997-1998 DEPARTMENT OF TREASURY WASHINGTON, DC
Office of Thrift Supervision
Financial Regulatory Affairs Fellow, Presidential Appointee
Researched the impact of welfare reform, microenterprise lending and Individual
Development Accounts on the financial services industry
Provided strategic recommendations for action on behalf of the banking sector
Deliverables included: conference lectures; published research papers; informational and
marketing materials on each subject; and industry-wide conference and training modules
5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 37 of 801
Professional Board Member, The Nathan Cummings Foundation
Service Board Member, AEO
Board member, The Motley Fool Foundation
Advisor, Goldman Sachs One Billion Black Women Initiative
Advisor, Goldman Sachs Launch with GS Program
Member, Fast Company Executive Board
Member, Square & Forbes Small Business Advisory Team
Member, Target Accelerators Entrepreneurship Advisory Council
Current Member & Former Co-Chair, National Advisor Council on Innovation and
Entrepreneurship, Department of Commerce
Former Advisor, Credit Suisse New Markets Advisory board
Former Nonresident Senior Fellow in the Metropolitan Policy Program, Brookings Institution
Founding Member, Dell Center for Entrepreneurs Advisory Board
Founder & Former Chair, Georgetown Entrepreneurship Alliance
6
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 38 of 801
Exhibit B
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 39 of 801
NBER WORKING PAPER SERIES
AN ANALYSIS OF THE IMPACT OF AFFIRMATIVE
ACTION PROGRAMS ON SELF-EMPLOYMENT
IN THE CONSTRUCTION INDUSTRY
David G. Blanchflower
Jon Wainwright
Working Paper 11793
http://www.nber.org/papers/w11793
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
November 2005
We thank Barry Hirsch, Colette Holt, John Scott, Chris Snyder, Jon Skinner, Doug Staiger and Jonathan
Zinman for helpful comments and suggestions. All errors are ours The views expressed herein are those of
the author(s) and do not necessarily reflect the views of the National Bureau of Economic Research.
©2005 by David G. Blanchflower and Jon Wainwright. All rights reserved. Short sections of text, not to
exceed two paragraphs, may be quoted without explicit permission provided that full credit, including ©
notice, is given to the source.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 40 of 801
An Analysis of the Impact of Affirmative Action Programs on Self-Employment in the
Construction Industry
David G. Blanchflower and Jon Wainwright
NBER Working Paper No. 11793
November 2005
JEL No. J4
ABSTRACT
The main findings of this paper are that despite the existence of various affirmative action programs
designed to improve the position of women and minorities in public construction, little has changed
in the last twenty five years. We present evidence showing that where race conscious affirmative
action programs exist they appear to generate significant improvements: when these programs are
removed or replaced with race-neutral programs the utilization of minorities and women in public
construction declines rapidly. We show that the programs have not helped minorities to become self-
employed or to raise their earnings over the period 1979-2004, using data from the Current
Population Survey and the Census, but have improved the position of white females. There has been
a growth in incorporated self-employment rates of white women in construction such that currently
their rate is significantly higher than that of white men. The data are suggestive of the possibility that
some of these companies are 'fronts' which are actually run by their white male spouses or sons to
take advantage of the affirmative action programs.
David G. Blanchflower
Department of Economics
6106 Rockefeller Hall
Dartmouth College
Hanover, NH 03755-3514
and NBER
blanchflower@dartmouth.edu
Jon Wainwright
NERA Economic Consulting
jon.wainwright@nera.com
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 41 of 801
1
1. Introduction
Construction is an important sector. In 2003, over 8 million workers, or approximately six per
cent of total employment in the United States, held a construction job.1 This is made up of 6.7
million wage and salary jobs and 1.6 million self-employed and unpaid family non-government
jobs. The construction industry accounts for a disproportionate share of union members and of
the self-employed. The self-employment rate is over 30% for painters, carpenters, roofers and
carpet, floor and tile installers.2 The construction industry, in 2003, accounted for 13.5% of all
private sector union members and 16.7% of all of the unincorporated self-employed and
approximately 14.9% of the incorporated self-employed.3 Employment of production workers in
construction has increased over the last ten years, beginning at 4,113,000 in 1995 and rising to
5,332,000 in 2001. In 2004 employment of production workers averaged 5,300,000.
The construction industry accounts for about 9.7% of all private sector establishments. More
than four out of five establishments in the industry employ fewer than 10 employees. There
were about 698,000 construction establishments in the United States in 2002 using the 2002
NAICS; 216,000 were building construction contractors; 37,000 were heavy and civil
engineering construction or highway contractors and 447,000 were specialty trade contractors.
Building construction accounted for 24.9% of wage and salary employment in 2002; heavy and
civil for 12.1% and specialty trades for 63.0%.4
According to the 2002 Economic Census Survey of Business Owners, of the 2,770,888 firms in
construction, 2.4% were owned by African Americans; 7.0% by Hispanics; 1.1% by American
Indians or Alaskan natives; 1.4% by Asians and Pacific Islanders and 10.5% by women. The
representation of minorities in particular among the ownership of firms in construction is well
below their representation in the population as a whole. As a proportion of the population,
according to the 2004/5 Statistical Abstract of the United States Table 13, in 2003 African
Americans were 12.7%; Hispanics 13.7%; Asian/Pacific Islanders 4.3%; American
Indians/Alaskan Native 1.0% and two or more races 1.3%.5
1
The main construction occupations in 2003 were, with number in thousands in parentheses, First-line supervisors
(897); Brick masons (218); Carpenters (1,595); Carpet floor finishers (271); Cement masons (120); Construction
laborers (1,151); Operating engineers (376); Drywall installers (205); Electricians (774); Painters (660); Pipe layers
and plumbers (595); Roofers (233); Sheet metal workers (147); Helpers (114) and Inspectors (95). Source: Table
597 Statistical Abstract of the United States 2004-5 downloadable at
www.census.gov/prod/2004pubs/04statab/labor.pdf
2
Source: www.bls.gov/oco/cg/pdf/cgs003.pdf
3
Source: Statistical Abstract of the United States: 2004-2005, U.S. Census Bureau at www.census.gov/statab/www/
Table 586 and Union Membership and Coverage Database http://www.unionstats.com/ and own weighted estimates
from the basic monthly files of the CPS for those aged 16-74.
4
Source: Statistical Abstract of the United States, 2004-5 Table 919
5
Large disparities between blacks and whites, for example, are also found in relation to wealth (Kennickel, 2003);
income (Bound and Freeman, 1992; Chandra, 2003; Heckman, Lyons and Todd, 2000 and Smith and Welch, 1989),
educational achievement (Jencks and Phillips, 1998); out-of-wedlock childbearing (Ventura and Bachrach, 2000),
health (see Chandra and Skinner, 2003), happiness (Blanchflower and Oswald, 2004) crime (Freeman, 2000) and
even names (Fryer and Leavitt, 2003). The degree of residential segregation by race, though lower today than in the
past is still high (Cutler, Glaeser and Vigdor, 1999).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 42 of 801
2
Ray Marshall (2000) has noted that there are several factors that make the construction industry
especially important for minority development because it "provides opportunity for upward
occupational mobility since workers commonly become managers and contractors". Glover
(1977) notes that minorities have a long tradition in this industry as laborers, skilled workers and
contractors and that it is possible to increase minority employment and income more effectively
in construction than is the case with most other minority businesses.
There is a good deal of evidence that the under-representation of women and minorities in
construction especially is due to widespread and pervasive discrimination that has changed little
over time. Not only is the proportion of firms owned by African Americans especially relatively
low, so also are their representation in the construction workforce in general and in self-
employment in particular. Where firms owned by minorities and women do exist in construction
they are more likely than non-minority males to be in special trades rather than heavy and civil.
They are also more likely to be sub-contractors than prime contractors. This does not appear to
be because of a lack of an ability to expand to undertake these activities because it is well-known
that small construction companies can expand rapidly as demand changes by hiring workers and
renting equipment and making use of sub-contractors. A particular concern in construction is
that it is hard for minority and women-owned firms to obtain capital, especially working capital,
and this causes increased difficulties when bonds have to be posted. This is often made more
difficult still when bonding firms are members of local construction associations. Also unions
are pervasive in the sector and these unions have tended to be dominated by white males and
have successfully controlled entry to craft jobs (Ashenfelter, 1972).
Construction projects especially in the public sector are frequently subject to competitive bidding
procedures. There is some evidence to suggest that there is collusion among bidders. Typically
governments award contracts for construction of highways and buildings through low-price
sealed bid auctions. On the day of the auction sealed bids are publicly reported along with
complete information about the bidder's identity and bids and the project is awarded to the lowest
bidder. In a study of bidding in highway construction projects in Florida from 1981-1986 Gupta
(2002) found that bid prices fall as the number of bidders increase. Gupta concluded that
"collusion is commonplace in these markets," (p.22, 2002). Gupta (2001) used the same data on
Florida and showed that bid prices were higher when repeated interaction among firms is
present. "Recurring market contact among rivals can facilitate bid-rigging by discouraging the
ring members to stray from collusive agreement" (p.466, 2001). Porter and Zona (1993)
examined auctions in New York State highway construction projects in Long Island in the early
1980s and found evidence of bid-rigging. Firms had many opportunities to communicate with
each other and some firms even submitted joint bids. They found evidence that several ring
members bid on most jobs but one was a serious bidder and others submitted phoney higher bids.
Porter and Zona argue that "the characteristics of this particular market tend to facilitate
collusion" (p. 524, 1993). Feinsten et al (1985) studied North Carolina highway contracts and
found that non-competitive bidding led to a clustering of bids in an attempt to influence
engineer's estimates of the cost of future lettings.
Ingraham (2005) considered the possibility of bidder-official collusion in contracting and
examined bidder data from New York City School Construction Authority auctions relating to
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 43 of 801
3
school repair and construction for the period 1990-1997. In April 1993 a scandal relating to bid
rigging became public and a number of prosecutions ensued: officials in the School Construction
Authority colluded with contractors by operating a "magic-number" scheme. The corrupt official
who acted as auctioneer would collude with a contractor and at the bid letting meeting would
keep his bid until last and knowing the low bid would read aloud a false bid just below this price.
Then after the bid opening the official would use White-Out to doctor the bid form. Ingraham's
regression analysis suggested that two auctioneers (other than the auctioneer already convicted)
may have been guilty of such magic-number cheating. Furthermore, an analysis of specific
bidders yielded weak evidence that two additional bidders may have been guilty of magic
number cheating. Such collusive arrangements between longstanding industry participants make
it very hard for new firms, especially those owned by women and minorities, to enter and be
successful.
The mechanism by which firms owned by minorities and women are likely excluded is that
without the collusion and rotating of bids, so that prices are above the competitive level, firms
would subcontract to minority owned businesses and use minority employees to lower costs and
undercut the prices of their rivals. But, of course if the firms are dividing up the business, taking
turns winning with higher than competitive bids, they have no need to compete by using low-cost
inputs, so the benefits of competition are not only lost regarding the price that the public pays for
the construction projects, but the benefits of competition for minorities are lost as well. The role
of the bid letting organizations is probably mostly to tacitly, or in many localities even overtly,
go along with or encourage the combination of banking and bonding discrimination on the
financial side with the established construction firms’ discriminatory subcontracting and hiring
decisions. It appears to be a complex confluence of actions by non-minorities that keeps
minority owned businesses from establishing themselves and surviving. The discrimination
likely results from a combination of actions from bid letting organizations, bidding companies,
banks, and bonding companies. The bid letting organizations go along and probably in some
locales even encourage the discrimination, the bidding firms in the club do not have competitive
incentives to hire lower-cost subcontractors and employees, the banks and other financiers and
bonding companies discriminate so they do not qualify as bidders on the financial side whatever
their business qualifications. So, collusion aids discrimination in subcontracting and in
employment.6
Affirmative action programs in construction have been implemented in many jurisdictions to
help overcome some of these problems. To prevent fraud, firms eligible for participation in the
programs have to register and be approved before they can participate. In many jurisdictions the
programs have been challenged in the courts on constitutional grounds and subsequently
removed entirely or replaced in whole or part with much weaker programs that are “race and
gender neutral” or their goals have been lowered. We document below that when such programs
are removed any improvements that have been achieved seem to rapidly disappear. Jurisdictions
continue to have difficulties in ensuring that firms that are fronted by minorities or women but
are really owned by white males are not benefiting from these programs.
6
We thank John Scott for helpful discussion on these points.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 44 of 801
4
In response to legal action and the threat of potential legal action the U.S. Department of
Transportation (USDOT), which operates one of the most significant federal programs for
affirmative action in construction, now allows its grant recipients to set annual goals for
participation by minorities, women, and other disadvantaged businesses with either a race/gender
conscious component, a race/gender neutral component, or both. Since the implementation of
this policy beginning in FY2000, there has been a huge increase in the portion of grant
recipients’ goals that are approved to be met exclusively or predominantly through race/gender
neutral means. These USDOT affirmative action programs are run under the auspices the
Federal Transit Authority (FTA), the Federal Aviation Authority (FAA) and the biggest of them
by the Federal Highways Administration (FHWA)
For example, of 201 FTA grant recipients we reviewed for FY2002, only 61 (30%) had goals
that were predominantly race/gender conscious. By contrast, 93 entities (46%) had FY2002
DBE goals that were completely race/gender neutral and another 47 entities (23%) had goals that
were predominantly race/gender neutral. This phenomenon was even more pronounced for
FY2003. In and of itself, the increasing use of race/gender neutral goals in the FTA’s
implementation of the DBE program is not a cause for concern. Indeed, few if any would argue
that race/gender conscious goals should be used instead of race/gender neutral ones if they are
equally effective at remediating discriminatory problems.
To test this, we obtained information from the FHWA containing the affirmative action goals
established by state departments of transportation for the years 1998-2004 as well as the amount
of disadvantaged business participation achieved in each year. Between 2000 and 2002
approximately 21% of state DOTs adopted DBE goals that were predominantly race/gender
neutral. Another 14% adopted goals that were entirely race/gender neutral. Although Federal
highway construction aid dollars increased from $14.7 billion in 1998 to $24.3 billion in 2002,
and although overall disadvantaged business goals remained virtually constant during this time,
the proportion of federal aid dollars actually awarded to such businesses declined almost 30%
between 1998 and 2002. The vastly increased use of race/gender neutral goals in the
implementation of the USDOT DBE program appears to have been accompanied by greatly
reduced participation by minority and women firms.
Since 1998 the FAA has not been reaching the goals it set for itself for minority-owned
businesses but it has done so for women-owned. Indeed, there is some evidence that the gap
between the goal and achievement has widened over time for minority businesses. For example,
in 2004 the FAA had a goal of 10% for minority-owned businesses but achieved only 7%
whereas it had a goal of 5% for women-owned businesses and accomplished 6%.In 2005 the
minority goal was cut to 5%. We were unable to obtain information on the relative importance
of race conscious versus race neutral goals in the FAA program.7
In the next section we examine the history of affirmative action programs in construction. In
section three we document what happens when race and gender conscious programs are halted.
In section four we perform an econometric analysis of self-employment to determine how
effective the various affirmative action programs have been. Our main findings are that the
programs have not helped minorities to become self-employed or to raise their earnings but have
7
See www.sbo.faa.gov/sbo/goalsprevious.asp.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 45 of 801
5
improved the position of white females. We find that there has been a growth in incorporated
self-employment rates of white women in construction such that currently they are significantly
higher than those for white men. The data are suggestive of the possibility that some of these
companies are 'fronts' which are actually run by their white male spouses to take advantage of
the affirmative action programs.
2. A History of Affirmative Action Programs in Public Construction.
Congressional efforts to promote minority-owned business enterprises have their origins in
Section 8(a) of the Small Business Act of 1958, which originally empowered the Small Business
Administration (SBA) to “set-aside” certain federal contracts for performance by “small business
concerns or others.”8 The SBA initially used its powers under Section 8(a) to hire businesses
agreeing to locate in or near certain low-income areas and provide training and jobs to the
unemployed or underemployed residents of those areas.9
Between 1969 and 1971 the Executive branch encouraged the focus of the 8(a) program to shift
from creation of jobs in low income areas to creation of demand for minority-owned business
enterprises (MBEs). Executive Order 11458 (1969) established the Office of Minority Business
Enterprise (OMBE) within the Department of Commerce and directed the Secretary of
Commerce to develop programs and coordinate interagency activities to promote the growth of
minority business. Executive Order 11625 (1971) further extended the mission of OMBE,
including supply-side measures such as subsidizing the provision of technical and managerial
assistance to MBEs by various public and private organizations and additional demand-side
measures requiring “best efforts” by prime contractors to utilize minority businesses as
subcontractors on large federal contracts.10 In 1973 the Justice Department also promulgated
regulations pursuant to Title VI of the Civil Rights Act of 1964 prohibiting state and local
governments from discriminating against minority contractors in federally-funded programs.11
In 1978 Congress provided a statutory basis for promoting minority business development
through the SBA 8(a) program with the passage of Public Law 95-507. The first specific
numerical goals for minority participation in federal contracts appeared as part of the Public
Works Employment Act of 1977 which directed that 10 percent of the funds provided by the act
be expended with MBEs.12
In 1982, through the U.S. Department of Transportation (USDOT), Congress established the
Disadvantaged Business Enterprise (DBE) program, designed to increase the participation of
MBEs and other socially and economically disadvantaged businesses in federally-assisted
highway, transit, and airport contracts. The Surface Transportation Assistance Act of 1982 and
8
15 USCS § 637(a).
9
Dale, Charles V. “Affirmative Action Revisited: A Legal History and a Prospectus,” Congressional Research
Service, Library of Congress, 2004.
10
48 CFR 19.201, 19.702 (superceding 41 CFR 1-1.1302, 1310-2).
11
28 CFR 42.104(b)(vi).
12
42 U.S.C.A.§ 6705(f)(2).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 46 of 801
6
the Airport and Airway Improvement Act of 1982 contained the first statutory provisions for
DBEs, requiring that a minimum of 10 percent of the funds provided by the acts be expended
with small businesses owned and controlled by socially and economically disadvantaged
individuals. Non-minority women-owned businesses (WBEs) were not included initially as
socially and economically disadvantaged individuals. The Surface Transportation and Uniform
Relocation Assistance Act of 1987 continued the DBE program and included non-minority
women in the statutory definition of socially and economically disadvantaged individuals,
thereby allowing states to use contracts with both minority- and women-owned businesses
(MWBEs) to meet their DBE goals. The Intermodal Surface Transportation Efficiency Act of
1991 (ISTEA), the Transportation Equity Act for the 21st Century of 1998 (TEA-21), and the
Safe, Accountable, Flexible, Efficient Transportation Equity Act: a Legacy for Users
(SAFETEA-LU) of 2005 reauthorized the program, continuing the combined 10 percent
provision for participation by minority-owned and non-minority women-owned DBEs.13
Following the lead of the federal government, policies to promote the inclusion of MWBEs in
government contracting activities spread gradually throughout the 1970s and 1980s to many
states and municipalities throughout the U.S. States such as Texas, Illinois, and Maryland and
cities such as Chicago, Denver, San Francisco, Baltimore, and Atlanta also established their own
demand-side goals programs on their construction spending on projects funded by state or local
dollars. These programs have taken various forms, employed different approaches to goal-
setting, and achieved different results. Some jurisdictions have included supply-side programs as
well, for example to assist small firms in their access to capital, obtaining surety bonds, or
obtaining technical or managerial assistance. Some of the demand-side programs and most if not
all of the supply-side programs are race neutral in nature in that they are designed to help all
small firms not just minority-owned or woman-owned firms. By 1989 more than 35 states and
almost 200 local governments had passed some sort of affirmative action initiative directed at
MWBEs.14
There is a presumption underlying all of this legislation that discrimination against minorities
and women in business, and especially in the construction industries where much of the
legislation has focused, continues to persist. Senator Lautenberg, for example, stated during the
course of Senate hearings in relation to TEA-21.
“Jim Crow laws were wiped off the books over 30 years ago. However, their
pernicious effects on the construction industry remain. Transportation
construction has historically relied on the old boy network which, until the last
decade, was almost exclusively a white, old boy network. … This is an industry
that relies heavily on business friendships and relationships established decades,
sometimes generations, ago — years before minority-owned firms were even
allowed to compete.”15
13
See http://osdbuweb,dot.gov/documents/pdf/legislation/SAFETEA-LU.pdf
14
Wainwright (2000), p. 8.
15
p. 5101 Federal Register / Vol. 64, No. 21 / Tuesday, February 2, 1999 / Rules and Regulations.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 47 of 801
7
The U.S. Supreme Court's decision in City of Richmond v. J.A. Croson Co., 488 U.S. 469 (1989),
established the current constitutional contours of permissible race-based public contracting
programs. Reversing long established law, the U.S. Supreme Court for the first time extended
the highest level of judicial examination to legislation to benefit rather than injure the historic
victims of discrimination. However benign the government’s motive, race is now always so
suspect a classification that its use must now pass the highest constitutional test of “strict
scrutiny.”
In affirming the Court of Appeals’ determination that Richmond’s plan was unconstitutional,
Justice O’Connor’s plurality opinion rejects the extreme positions that local governments either
have carte blanche to enact race-based legislation, or must prove their own illegal conduct
before they can enact race-based legislation.
“[A] state or local subdivision . . . has the authority to eradicate the effects of
private discrimination within its own legislative jurisdiction. . . . [Richmond] can
use its spending powers to remedy private discrimination, if it identifies that
discrimination with the particularity required by the Fourteenth Amendment. . . .
[I]f the City could show that it had essentially become a 'passive participant' in a
system of racial exclusion . . .[it] could take affirmative steps to dismantle such a
system.”16
Apparently recognizing that the opinion might be misconstrued to categorically eliminate all
race-conscious contracting efforts, Justice O’Connor closes with these admonitions:
“Nothing we say today precludes a state or local entity from taking action to
rectify the effects of identified discrimination within its jurisdiction. If the City of
Richmond had evidence before it that non-minority contractors were
systematically excluding minority businesses from subcontracting opportunities it
could take action to end the discriminatory exclusion. Where there is a significant
statistical disparity between the number of qualified minority contractors willing
and able to perform a particular service and the number of such contractors
actually engaged by the locality or the locality’s prime contractors, an inference
of discriminatory exclusion could arise. . . . Under such circumstances, the City
could act to dismantle the closed business system by taking appropriate measures
against those who discriminate on the basis of race or other illegitimate criteria. . .
In the extreme case, some form of narrowly tailored racial preference might be
necessary to break down patterns of deliberate exclusion. . . . Moreover, evidence
of a pattern of individual discriminatory acts can, if supported by appropriate
statistical proof, lend support to a local government’s determination that broader
remedial relief is justified.”17
Thus,
16
Id. at 491-492.
17
Id. at 509 (citations omitted).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 48 of 801
8
“[t]he strict scrutiny test involves two questions. The first is whether the interest
cited by the government as its reason for injecting the consideration of race into
the application of law is sufficiently compelling to overcome the suspicion that
racial characteristics ought to be irrelevant as far as treatment by the government
is concerned. The second is whether the government has narrowly tailored its use
of race, so that race-based classifications are applied only to the extent absolutely
required to reach the proffered interest. . . . The strict scrutiny test is thus a
recognition that while classifications based on race may be appropriate in certain
limited legislative endeavors, such enactments must be carefully justified and
meticulously applied so that race is determinative of the outcome in only the very
narrow circumstances in which it is relevant.” Adarand Constructors, Inc. v.
Peña, 965 F.Supp. 1556, 1569 (D. Colo. 1997) (“Adarand IV”).
In addition to the federal government many cities and states had adopted demand-side goals
programs and supply-side technical assistance programs. After Croson, however, numerous city,
state, and federal programs that had race conscious components were challenged in the courts
and, in most cases, ruled unconstitutional. Important cases include: City of Richmond; City of
Philadelphia; City of Columbus; Fulton County, Georgia (Atlanta); Dade County, Florida
(Miami); King County, Washington (Seattle); Cook County, Illinois (Chicago); Hillsborough
County, Florida (Ft. Lauderdale); the District of Columbia; the State of Ohio, the State of
Florida, and the Michigan DOT. In some jurisdictions a settlement was reached before the
courts ruled and the program was dropped entirely such as in the South Florida Water
Management District, the City of Charlotte and the City of Memphis. The State of Texas DOT
added white males to its program effectively making the program race-neutral, while the State of
Maryland and the City of Jacksonville adopted revised race-conscious programs. In other
jurisdictions such as the City of Columbus and the New Jersey DOT race conscious programs
were replaced with programs that were entirely race blind.18 Other jurisdictions dropped their
programs for fear of being sued.
The challenging party has the ultimate burden of production and persuasion that an affirmative
action program in public contracting is unconstitutional. While the government has the burden
of producing a ‘strong basis in evidence’ in support of its program, it does not bear the burden to
prove it constitutional. When the statistical information is sufficient to support the inference of
discrimination, the plaintiff must prove that the statistics are flawed. A plaintiff cannot rest upon
general criticisms of studies or other evidence; it must carry the case that the government's proof
is inadequate to meet strict scrutiny, rendering the legislation illegal. The determination whether
a plaintiff has met this burden is a question of law, subject to de novo review.
Much of the dissension in the case law has revolved around what type of evidence is sufficiently
'strong' to establish the continuing existence and effects of economic discrimination against
minorities resulting in diminished opportunities to do business with the government.
Discrimination has to be demonstrated through the use of statistics and economic models to
18
Whether affirmative action procurement programs that benefit women are subject to the lesser constitutional
standard of ‘intermediate scrutiny’ has yet to be settled by the Supreme Court.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 49 of 801
9
examine the effects of systems or markets on different groups, as well as by evidence of personal
experiences with discriminatory conduct, policies or systems. Specific evidence of
discrimination or its absence may be direct or circumstantial, and include economic factors and
opportunities in the private sector affecting the success of Minority and Business Owned
Enterprises (MWBEs). Anecdotal evidence of experiences with discrimination in contracting
opportunities, including testimony from other governments’ studies and programs, is relevant
since it goes to the question whether observed statistical disparities are due to discrimination and
not to some other no-discriminatory cause or causes. Discrimination by prime contractors,
unions, bonding companies, suppliers and lenders has also been found relevant to the creation of
barriers both to minority subcontractors’ business formation and to their success on
governmental projects. Race- and gender- neutral approaches have become a necessary
component of a defensible MBE program despite the fact that as noted by Ray Marshall, there
remains little or no evidence that race blind measures have anything other than minimal impacts.
"Because of the difficulties in applying general antidiscrimination measures to
businesses, federal, state and local governments have used their purchasing power
as the most effective way to overcome the effects of, and to prevent,
discrimination against minority businesses. The federal government adopted this
approach after other so-called ‘race neutral’ approaches had failed to overcome
the effects of discrimination in the construction industry".19
Whether every race-neutral approach must be implemented and then proven to be ineffective
before race-conscious remedies may be utilized, remains uncertain.
Recently there have been a number of cases that have reversed the immediate post-Croson trend
and found affirmative action programs to be constitutional.20 These include Sherbrooke Turf v.
Minnesota DOT and Gross Seed v. Nebraska DOT that related to the federal DBE program.21
Here, the courts ruled that the U.S. Congress had established ‘compelling interest’ and that the
state programs were ‘narrowly tailored.’ In both these cases the U.S. Department of Justice and
the U.S. Department of Transportation acted as intervenors, with the U.S. government defending
the programs.22 Recently the U.S. District Court in Illinois, in the case of Northern Contracting
Inc vs. the State of Illinois Department of Transportation (IDOT), ruled in favor of IDOT, citing
the findings in the Sherbrooke Turf case as precedent, finding a compelling need for IDOT’s plan
which was “narrowly tailored to the goal of remedying the effects of racial and gender
19
Ray Marshall (2000), citing Croson, 488 U.S. 469, 550 (Marshall, J., dissenting), noting that while race neutral
measures are “theoretically appealing, [they] have been discredited as ineffectual in eradicating the effects of
discrimination” in the construction industry.)
20
See Holt (2003).
21
Sherbrooke Turf, Inc. v. Minnesota Department of Transportation, et al, No.02-1665 (8th Cir. October 6, 2003)
(Sherbrooke III), and Gross Seed Company v. Nebraska Department of Roads, et al, No. 02-3016 (8th Cir. October 6,
2003).
22
Appellee briefs filed by the U.S. government in the district court in Western States, and in both the district and
the 8th Circuit Court of Appeal in Sherbrooke Turf and Gross Seed, along with the court decisions are available at
http://www.usdoj.gov/crt/app/briefs_aa.htm
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 50 of 801
10
discrimination within the construction industry.”23 In a similar case relating to the federal DBE
program of Western States Paving v. State of Washington Department of Transportation
(WsDOT), the court granted the defendant summary judgment dismissing the case but this was
overturned in part on appeal to the 9th Circuit Court of Appeals because WsDOT did not offer
any evidence of discrimination in the Washington state construction industry and hence their
program was not narrowly tailored. The 9th Circuit did, however, find the DBE Program
constitutional on its face.
Two further cases relate to programs established not by the federal government with the approval
of Congress but by two cities - Denver and Chicago. In Concrete Works v. the City and County
of Denver and in Builders Association of Greater Chicago (BAGC) v. City of Chicago
affirmative action programs for MWBEs in construction were challenged as unconstitutional.
The defendants did not have the benefit of arguing that Congress had established compelling
interest and thus had a higher bar to reach since they had to establish compelling interest
themselves within their own jurisdictions. Despite this, after lengthy trials, defendants in both
cases successfully argued they had established a compelling interest in having their programs
and in the Concrete Works case, after a decision of the 10th Circuit Court of Appeals overturning
the lower court’s decision, that the program was narrowly tailored.24 The U.S. Supreme Court
refused to overturn the 8th and 10th Circuit Appeals Court decisions respectively in Concrete
Works, Sherbrooke Turf and Gross Seed by refusing to grant writs of certiorari. In the Concrete
Works case, however, Justices Scalia and Rehnquist wrote a dissent from denial of certiorari.25
In the BAGC case the district court temporarily enjoined the program but permitted the City of
Chicago to take steps to narrowly tailor its program, which it did a few months later when it
introduced a new program. The authors were defendants’ experts in the Concrete Works;
Sherbrooke Turf, Gross Seed, BAGC and Northern Contracting cases.26
A recent report on affirmative action programs across all federal contracting by the U.S.
Commission on Civil Rights (2005) makes interesting reading and shows the diversity of views
23
Northern Contracting Inc v. the State of Illinois Department of Transportation (N.D., Il.), No. 00C4515
September 8th 2005, p. 52.
24
Initially the trial court in Concrete Works found the program unconstitutional but the district court's decision was
overturned by the 10th Circuit Court of Appeals. In both the Gross Seed and Sherbrooke Turf cases the 8th Circuit
Court of Appeals, in an opinion where it consolidated the two cases, upheld the lower court's decision that Congress
had established compelling interest for the programs and that defendants had established them to be narrowly
tailored.
25
In Concrete Works the U.S. government filed amicus briefs on the side of the City and County of Denver with the
10th Court of Appeals opposing the district court's decision declaring the program unconstitutional. See
http://www.usdoj.gov/crt/briefs/concrete.htm. The 10th circuit court of Appeals decision is available at
http://www.usdoj.gov/crt/app/briefs_aa.htm. The U.S. Supreme Court dissent by Scalia and Rehnquist in Concrete
Works is available at www.supremecourtus.gov/opinions/03pdf/02-1673.pdf.
26
Both authors contributed to expert reports for defendants in all five of these cases, Blanchflower testified at trial in
Concrete Works, Gross Seed and BAGC and Wainwright testified at trial in Northern Contracting. Blanchflower
and Wainwright both gave depositions in Sherbrooke Turf but this case did not go to trial but rather was dismissed
under summary judgment in favor of defendants.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 51 of 801
11
on the issue.27 The majority report extols the virtues of race neutral contracting but does not
present any evidence of discrimination, for example, by race or gender in construction or
elsewhere or whether it has improved or worsened over time, or how it varies by sector and/or
location. There is merely a presumption in the report that race neutral measures work,
unaccompanied by any evidence one way or the other to justify such a conclusion. A
considerable part of the report consists of a dissent by Commissioner Michael Yaki who calls the
majority report 'fatally flawed" and argues that "there is significant data demonstrating that
socially and economically disadvantaged firms continue to lag well behind where they should be
in proportion to the country's demographic composition or labor force representation.” (p86). He
presents statistical evidence to support his arguments in a “Dissent Appendix.”
There is evidence since the 1980s in the United States that race and gender conscious affirmative
action programs in construction and elsewhere can have a substantial impact.
Enchautegui et al (1997) found that there are substantial disparities between the share of public
contract dollars received by minority-owned firms and the share of all firms that they represent.
The difference in these disparities is especially marked between jurisdictions which have
affirmative action programs and those that have never had them or removed them. The
disparities are much less where affirmative action programs exist.
Chay and Fairlie (1998) examined whether the set-aside programs established in many of the
largest U.S. cities during the 1980s had an impact on the self-employment rates of blacks and
whites. Their analysis was based on data from the Current Population Survey for 1979-1988.
They stopped at 1988 since the 1989 Croson decision led to the dismantling of set-aside
programs in many cities. They had some difficulty in ascertaining the exact starting dates of a
number of the programs because there were inconsistencies in the data sources used. The paper
examined what happened to self-employment rates when programs were introduced. They found
that self-employment rates in construction for black men increased from about 12% before the
programs were introduced to about 18% after. On the other hand, rates for white men were
relatively stable before and after the programs were passed. Chay and Fairlie argued that "these
figures provide strong evidence that set-aside programs worked".
As previously indicated, a condition of accepting federal funding for highway construction, state
departments of transportation need to ensure a proportion of those dollars are awarded to DBEs -
a large percentage of which are owned by minorities and women. A number of these states do
not have a goal program when they spend their own dollars. Utilization of minority- and female-
owned firms in such states is much higher where federal dollars are used than where state dollars
are used. Figure 1 provides the details based on data for fiscal year 1996. In all of these ten
states, utilization of DBEs was higher on federally funded projects with goals than on state
funded contracts without them.
New data have recently become available from the Federal Highways Administration on
participation of DBEs on federal DBE programs in seventeen states in 2002 that do not have
such a program for spending on highway construction funded by state dollars. Usefully,
27
www.usccr.gov/pubs/080505_fedprocadarand.pdf
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 52 of 801
12
information is available in these states on firm's DBE status under the auspices of the federal
program. As is apparent from Table 1, of the seventeen states only three – Florida, Mississippi
and Wyoming have higher utilization of DBEs on state projects without goals than is found when
federal money is spent with attached goals. Overall, utilization of DBEs is 10% when goals are
in place and 5.7% when they are not. We now turn to examine what happens when such goals
programs are removed or replaced with race-neutral efforts in response to litigation or the threat
thereof.
3. What happens when race conscious affirmative action programs are halted?
Over the last few years the courts have either permanently or temporarily halted programs that
benefited MWBEs in many jurisdictions across the United States. In some cases the programs
have started up again when the injunctions were lifted or a new program instituted. This
provides an opportunity to examine what happened to utilization of such firms when the program
is removed. This is a good way to get at the scale of the impact of the programs. As can be seen
below the impact was devastating – utilization dropped to zero in at least one case. Examples
include the following.
• After the Supreme Court invalidated Richmond's minority business set-aside program,
minority participation fell from 30 percent to 4 percent, representing a drop of more than 87
percent.
• In Atlanta, minority participation in City contracts declined from about 35 percent in 1988 to
14 percent in 1989 after the Georgia Supreme Court invalidated the city's program.
• In Fulton County, minorities increased their participation in county procurement business
from .004 percent in 1977-79 before the county's program was adopted, but declined from 17
percent to 1.7 percent after that program was temporarily enjoined in 1986.
• In Philadelphia, public works subcontracts awarded to minority-owned firms fell by 97
percent during the first month after the city's program was enjoined in 1990.
• In Hillsborough County, Florida, minority contractor participation fell by 99 percent after
their program was struck down.
• In the City of Elyria, Ohio (a suburb of Cleveland), minority participation in public works
contracting virtually disappeared after their program was permanently enjoined by the court
in 1991. Participation fell from approximately $1 million dollars (25 percent) in 1991 to
$26,000 (0.6 percent) in 1992. In 1993, participation fell further to a mere $19,000 (less than
one-half of one percent of the total).
• In the City of Tampa, black participation in city contracting fell by 99 percent and Hispanic
participation fell by 50 percent.
• In San Jose, California, minority participation in the city's prime contracts fell by more than
80 percent in 1989 when their program was originally suspended.
• In the wake of Proposition 209, the State of California Department of Transportation halved
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 53 of 801
13
its state-funds for Minority and Women-Owned Business Enterprise (MWBE) goal from 20
percent to 10 percent. Subsequently, MWBE participation fell from 25.8 percent in 1995 to
14.7 percent in 1996 to 12.2 percent in 1997.
In a study of disadvantaged business enterprises the General Accounting Office (2001) reported
that MWBE participation in transport contracting declined after Louisiana's nonfederal program
was discontinued. As shown in Figure 2, taken from the GAO report, MWBE participation in
state transportation contracting increased from 1992 to 1995. In 1996, the year the nonfederal
program was discontinued, the participation rate of MWBEs in state transportation contracting
dropped and continued to decline over the next 4 years (Page 40). The GAO reported that an
official from Louisiana attributed the decline in MWBEs' participation in state transportation
contracting to the removal of affirmative action requirements on state funded projects and the
realization by contractors that efforts to include MWBEs were no longer necessary.
The GAO also obtained data on DBE participation in Louisiana's federal DBE program from
1995 to 2000. In their review of the data on DBE and MWBE participation in federally assisted
and state contracting, they did not observe a shift of MWBEs to federally assisted contracts after
the Louisiana's MWBE program was discontinued. This evidence is entirely consistent with the
evidence presented above that the goals programs work: removing them impacts negatively the
usage of minority and women-owned firms.28
On September 2nd, 1998, in the case of Sherbrooke Turf v. the Minnesota Department of
Transportation and Judge Rosenbaum enjoined Minnesota Department of Transportation's
(MnDOT) MWBE Program. In August 1999, Judge Rosenbaum ruled that the injunction only
applied to ISTEA and not to TEA-21. On April 28, 2000, MnDOT let its first contracts using
goals under the TEA-21 MWBE Program. MnDOT did not operate any sort of MWBE program
during the time the program was enjoined. MnDOT also did not certify any new MWBEs during
that time. MnDOT did not monitor the certification status of MWBEs certified prior to the
injunction. Strictly speaking, there were no certified MWBEs during the 1999 construction
season because there was no program under which minority and female owned firms could be
certified. During the period of enjoinment, MnDOT continued to monitor the participation in
federal aid projects by the MWBEs certified under the enjoined program. This monitoring
spanned the 1999 construction season, which is based on the federal fiscal year ("FFY"), October
1st, 1998 through September 30th, 1999.
When monitoring MWBE participation during the period of enjoinment, MnDOT relied upon
prime contractors voluntarily submitting information to MnDOT. MnDOT did not receive a 100
28
During the Senate hearings on TEA-21 Senator Baucus cited the experience of Michigan where participation in
the state-funded portion of the highway program fell to zero in a nine month period after the state terminated its
MBE/WBE program while the federal program was able to maintain 12.7% participation. Senator Kerry also cited
similar decreases in MBE/WBE participation in Louisiana, Hillsborough County, Florida, and San Jose, California.
Senator Moseley-Braun added the examples of Arizona, Arkansas, Rhode Island and Delaware where state-funded
projects without a MBE/WBE program have significantly less MBE/WBE participation than federally funded
projects subject to the highway program. Senator Kennedy added Nebraska, Missouri, Tampa, and Philadelphia to
the list of jurisdictions that experienced precipitous drops in MBE/WBE participation after goals programs ended.
See p 5101 Federal Register / Vol. 64, No. 21 / Tuesday, February 2, 1999 / Rules and Regulations.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 54 of 801
14
percent return rate from the contractors. During the 1999 construction season, MWBEs certified
under the enjoined MWBE Program received approximately 2.6 percent of the contract dollars.
That number was subsequently used by MnDOT to reflect the amount it believed it could
achieve through race and gender neutral means. When MnDOT began implementing its TEA-21
MWBE Program, it required all firms seeking MWBE certification under the TEA-21 MWBE
Program to submit an application. Firms certified under the enjoined MWBE Program were
required to submit a new application.
MnDOT examined their construction expenditure data for the calendar year 1996-1999 and
produced the results reported in Table 3a. The total amount of contract dollars grew from $210
million in FFY96 to $296 million in FFY 1999. In the years 1996-1998 when the program was
in existence the proportion of contract dollars given to MWBEs averaged 10.8 percent, split up
as 2.2 percent for minority men; 0.1 percent for minority women and 8.5 percent for Caucasian
women. For the year 1999 when the program was enjoined, the proportion of contract dollars
paid to MWBEs fell to 2.25 percent. The amount paid to women fell to 1.95 percent but the
most precipitous decline was for minority men who in 1999 received only 0.28 percent of
contract dollars. In 1999 the amount of dollars received by minority males fell from just over
$6.5 million to $841,000. The amount of contract dollars received in 1999 by minority men was
only 12.7 percent of the amount they received when the program was in existence. Michigan and
Minnesota appear to be similar – remove the goals and the participation of MWBEs falls
precipitously. The evidence on the decline in DBE participation when the program was enjoined
was used by MnDOT to determine the proportion of their goals that could be obtained by race
neutral means.
Once Judge Rosenbaum lifted the injunction on the program, utilization rose again rapidly to
levels comparable to those pertaining prior to the temporary restraining order. Table 3b provides
further supporting evidence: it provides quarterly information for the five quarters for which data
are available since the program restarted and MWBEs once again were re-certified. It shows
dramatic increases in the use of MWBEs once the program was reintroduced. MWBEs gained
no contracts at all in Q1- FFY 2000. By the 1st quarter of 2001 participation of MWBEs had
risen to 8.7 percent. Most notable was the increase in the participation of minority men.
The GAO report also examined the evidence for Minnesota and confirmed the findings that after
the discontinuance of Minnesota's federal DBE program in 1998, Disadvantaged Business
Enterprises (DBEs) participation in federal transportation contracting dramatically declined.
This evidence is confirmed in Figure 3 taken from their report, which confirms dramatic declines
post 1998 both in the number of construction contracts and dollars awarded to DBEs.
There is also evidence on the impact of the removal of set-aside programs in the City of Chicago.
First the Chicago Metropolitan Water District had a very low utilization of firms owned by
women or minorities during the 1970s. From 1972-1974 Minority Business Enterprises (MBEs)
utilization in dollars spent was only 0.46 percent. From July 1977- May 1979 it was 2 percent.
Effective from May 1979 a goals program was implemented with goals of 10-15 percent on EPA
funded contracts. From this point utilization jumped rapidly and remained high during the years
the program was in effect, as can be seen below
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 55 of 801
15
May 1979-January 1982 12.7% MBE
1982 12.3% MBE
1983 12.8% MBE
1984 12.5% MBE
1985 10.6% MBE 2.2% WBE
1986 12.3% MBE 6.5% WBE
1987 17.7% MBE 7.7% WBE
1988 16.8% MBE 7.7% WBE
1989 13.9% MBE 9.6% WBE
When the goals program was suspended in 1989, the utilization of MBEs and Women-Owned
Business Enterprises (WBEs) dropped dramatically. During the first half of 1989 $19.6 million
of contracts were awarded with the goals still in place. Utilization of MBEs was 23.6 percent
with a further to 16.3 percent going to WBEs., half 1989 contracts without goals. During the
second half of 1989 when the goals program had gone and contractors were free to determine
their own usage of MWBEs of the $63.6 million awarded only 10.8 percent went to MBEs and a
further 7.4 percent went to WBEs.29
In 2002 the County of Cook's sub-contractor goals program in construction was struck down.30
Prior to the entry of the injunction the Cook County ordinance set – and apparently met - sub-
contractor goals for MBEs of 30 percent and for WBEs of an additional 10 percent making 40
percent in all. In a subsequent case involving the same plaintiff, evidence was presented at trial
from Cook County on the utilization of minority and women-owned firms.31 When contracts
relating to medical equipment and supplies were omitted and restricting the analysis to contracts
involving ‘dirt' upon examining contracts awarded by Cook County post-litigation in 2001 and
thereafter it was found that MBE utilization had fallen dramatically. Out of a total of
$51,045,508.09 in contract dollars spent, $19,741,139.25 was for equipment and was removed
from the calculations. This leaves $31,304,368.83 of which $4,016,641.58 went to MBEs (12.83
percent) and $1,579,768.97 went to WBEs (5.05 percent). Overall then $5,596,410.55 out of
$31,304,368.83 went to MWBEs (17.88 percent). This is a dramatic drop from the pre-
injunction levels at a time when these contractors were likely aware that they were under
scrutiny and likely to be on their best behavior as they were plaintiffs in ongoing litigation on the
same issue with an overlapping jurisdiction.
A race and gender conscious MWBE program appears to be needed not just to overcome the
effects of past discrimination, but to prevent future discrimination as well. Several state
Departments of Transportation such as Arizona, Arkansas and Connecticut have quite high
utilization of MWBEs when they spend federal dollars that require their use, but the same
29
These numbers are taken from the Chicago Metropolitan Water District Study, 1990
30
Builders Association of Greater Chicago vs. County of Cook, in the U.S. District Court for the Northern District
of Illinois Eastern Division, Case # 96 c 1121, November 2, 2000, Decided
31
Builders Association of Greater Chicago vs. City of Chicago in the U.S. District Court for the Northern District of
Illinois Eastern Division, Case No. 96 C, 1122, December 29th 2003, Decided.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 56 of 801
16
organizations make use of hardly any of these firms when they spend their state monies, which
do not have goals programs. The gaps are not small – in Rhode Island for example they were
12.5 percent and zero respectively. Evidence from many state and municipal jurisdictions
suggests that once these programs are removed the utilization of MWBEs drops precipitously.
Analogously, when they are put in place, utilization of minority and women-owned firms
increases. There is no evidence that we are aware of that suggests that when programs are
removed utilization of minorities and women goes up or remains constant. The evidence is
universally that utilization drops when these programs are enjoined or dropped. Goals programs
are effective in promoting the utilization of MWBEs in construction. We are aware of no
evidence that suggests that race neutral programs improve the relative position of minorities. The
big question is how effective have these programs been in improving the position of minorities
and women in the construction labor market? Race conscious affirmative action programs
appear to work when they are in place but legal action has removed and/or severely restricted
their presence. To what extent has this impacted the position of minorities and women in
construction? This is what we now examine.
4. Analysis of the Impact of Affirmative Action Programs in Construction.
We now turn to examine the construction labor market and see what has happened to the self-
employment rate of women and minorities over time. Given the existence of Prevailing Wage
Laws32 and widespread affirmative action programs it is of interest to determine whether the
position of women and minorities in the construction sector has improved or worsened over time.
In particular it is of interest to examine the changes post-Croson, so we compare self-
employment rates pre-1991 to the post-1991 rates. We show that the position of white females
appears to have improved but there is little or no evidence to support that conclusion in the case
of African Americans or Hispanics. The evidence is more mixed for other racial groups such as
Asians and Native Americans.
Here we concentrate on publicly available nationally representative data on individuals. We look
here at both the owners of firms as well as their employees. At the level of the individual, of
course, we have very good data on the individual's education, race and gender. It is not
unreasonable to look at the self-employed given our finding above from the 2002 Economic
Census Survey of Minority-owned Business Enterprises that approximately 74% of construction
firms have no employees. We have data both on the unincorporated self-employed as well as the
incorporated self-employed who receive wages and salaries but are not counted as part of the
official self-employment rate in the U.S. government counts.
A continuing puzzle in the literature in the U.S. has been why the self-employment rate of black
males is one third of that of white males and has remained roughly constant since 1910.
Moreover, minorities start businesses at much lower rates than non-Hispanic whites. Using the
Public Use Microdata Samples (PUMS) data from the 1990 Census, Wainwright (2000)
demonstrated that these disparities persist even when factors such as geography, industry,
occupation, age, education and assets are held constant.
32
Prevailing Wage Laws (PWLS) exist at the federal and in addition the majority of states have their own PWLs.
The intent of such legislation is primarily to protect the local wage rates in construction. These laws appear to have
only small impacts on construction labor markets (Kessler and Katz, 2001).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 57 of 801
17
Fairlie and Meyer (2000), rule out a number of explanations for the difference in the self-
employment rates of white and black males. They found that trends in demographic factors,
including the Great Migration and the racial convergence in education levels “did not have large
effects on the trend in the racial gap in self-employment” (p. 662). They also found that an initial
lack of business experience “cannot explain the current low levels of black self-employment.”
Further they found that “the lack of traditions in business enterprise among blacks that resulted
from slavery cannot explain a substantial part of the current racial gap in self-employment” (p.
664 Fairlie (1999) and Wainwright (2000) have shown that a considerable part of the explanation
of the differences between the African American and white self-employment rate can be
attributed to discrimination. Bates (1989) finds strong supporting evidence that racial differences
in levels of financial capital have significant effects upon racial patterns in business failure rates.
Fairlie (1999) also found that the black exit rate from self-employment is twice as high as that of
whites.
The gap between the black and white male self-employment rates has persisted for the best part
of a century. The white male self-employment rate decreased from its highest point in 1910
(16.0 percent) to its lowest point in 1970 (10.0 percent). In 1970 the downward trend in the self-
employment rate ended and began a climb to 11.4 percent in 1990. Since that time the rate has
declined. The similar trends in the white and black rates resulted in a roughly constant
black/white ratio during the past 80 years. As Fairlie and Meyer (2000) note:
“The constancy of the black/white ratio is surprising in light of the substantial gains
blacks have made in education, earnings and civil rights during the twentieth
century…and the numerous government programs created to promote minority
business ownership” (p. 656)
In contrast, there has been a striking growth over time in the self-employment rate of females.
Devine (1994) showed, using the March Current Population Survey data that the number of self-
employed females aged 18-64 in the non-agricultural sector increased by 2.2 million or 145
percent between 1975 and 1990. This represented an increase in the self-employment rate from
4 percent to 6.6 percent.
If the various affirmative action programs had been successful we might expect to see a growth
in the number of firms owned by women and minorities, an increase in the self-employment rate,
an increase in the earnings of women and minority business owners. We do not; discrimination
against minorities in general, and African-Americans and Hispanics in particular, continues to
persist despite efforts on the part of state, federal and local governments to speed its demise.33
33
One thing that makes construction an attractive sector for affirmative action is the relatively large importance of
the government in that sector – hence more leverage. Total spending on construction in 2004 was $1,027,736
million. Of this 77.7% was private, made up of 54.8% private residential and 22.9% private non-residential; of the
remainder 20.6% state and local and 1.7% federal (Source: Statistical Abstract of the United States, 2004-5, Table
642). Even in construction though, the private sector is far larger than the public sector. So even with affirmative
action in place the public sector is still only a relatively small slice of the overall economy.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 58 of 801
18
Table 3 presents new evidence on self-employment rates by race and gender for the period 1979-
2004 using data from the Merged Outgoing Rotation Group (MORG) files of the Current
Population Survey (CPS). Weighted estimates were obtained for each year and then averaged
over groupings of years for all industries as well as for construction for white males, white
females, African-Americans and Hispanics. Additional data are available for Asians/Pacific
Islanders and Native Americans since 1992 but they are excluded from this table as there is no
long run consistent time series. We examine them in detail below. Data are presented for total
self-employment which is the sum of incorporated and unincorporated as well as separately. The
figures in the table are percentages so 3.6 in the top left of the table should be interpreted as
3.6% of employed white males on average from 1979-1983 were classified as incorporated self-
employed, compared with 11.5% who were unincorporated in part B of the Table and hence
15.1% overall in part C and so on. The main findings from the table are as follows.
1) There has been a growth in the proportion of workers who are classified as incorporated self-
employed that the Census Bureau counts as wage and salary workers.
2) The self-employment rate of white males, whether incorporated or unincorporated, is
markedly higher than is the case of any other group.
3) There has been a decline in the unincorporated and the overall self-employment rate of white
males but increases for the other groups.
4) Construction self-employment rates, incorporated and unincorporated, have declined for
white males but increased for white females, blacks and Hispanics.
5) Incorporated self-employment rates for white women in construction are now higher than
they are for white males.
6) The differential between the overall self-employment rates of white males and white females
in construction has narrowed dramatically over time. The narrowing is much more apparent than
is found for ‘all Industries'.
7) The differential between the overall self-employment rates in construction of white males
and blacks has narrowed but less than it has for white women.
8) The differential between the overall self-employment rates in construction of white males
and Hispanics has widened over time.
We now move on to examine in more detail the disparity in business formation/self-employment
rates between racial and ethnic minorities and non-minorities.34 We examine the private sector in
general as well as construction in particular. Blanchflower (2000, 2004) provides summaries of
the literature, and compares the determinants of self-employment in the U.S. with that in other
OECD countries. Probit regression is used to examine the relationship between whether a person
is self-employed (yes or no) and a set of socioeconomic characteristics of that person. The data
used first are from the MORG files of the CPS. Data are available in a comparable form from
34
We use the phrases “business formation rates” and “self-employment rates” interchangeably in this chapter.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 59 of 801
19
1979 and the most recent comparable data are from 2004. Results are also presented using data
from the 2000 Census. The sample used in these regressions includes men and women who were
working at the time they were interviewed either as self-employed (dependent variable=1) or as a
private sector employee (dependent variable=0). Controls for age, location, industry, and
education are included along with indicators for race and gender. Coefficients represent
percentage point differences between a minority group and comparable whites—comparable in
age, location, industry, and education. The dependent variable is set to zero if the individual is a
private sector employee and 1 if self-employed, either incorporated or unincorporated. These
equations are estimating the probability of being self-employed, holding constant the named
characteristics.
In 2002 the industry classification system changed from the Standard Industrial Classification
(SIC) to using the 1997 North American Industry Classification System (NAICS). While many
of the individual SIC industries correspond directly to industries as defined under NAICS, most
of the higher level groupings do not. Particular care has to be taken in comparing data for retail
trade, wholesale trade and manufacturing which are sector titles used in both the SIC and NAICS
but cover somewhat different groups of industries. In the case of construction some changes
affecting it are within construction itself but in the NAICS it now includes industries that were
previously classified in other sectors. For this reason we only make use of data for 2003 and
2004 in columns 5-7 which are for construction because of the difficulty in obtaining consistent
industry codes for the All Industries grouping over these years.
Columns 1, 3 and 4 of Table 4 are for the pre-Croson period 1979-1991 and columns 2 and 5-7
for the subsequent post-Croson period 1992-2002. Columns 1 and 2 are for all industries while
columns 3-7 are restricted to the construction industry. Different education controls are used in
the two periods because that is what is available in our data. Columns 4 and 7 are for
construction where the dependent variable is set to one if the respondent was incorporated self-
employed, zero otherwise. Column 6 is also for construction where the dependent variable is set
to one if the respondent was unincorporated self-employed, zero otherwise.
The main findings from Table 4 are as follows:
1) Holding constant an array of worker and workplace characteristics, white males have highly
statistically significantly higher self-employment rates than do any other groups. This is evident
in the U.S. as a whole, as well as in Construction.
2) Blacks and Hispanics have especially low self-employment rates in construction.
3) Disparities by race and gender are considerably higher in construction than overall.
4) The magnitude of the difference in self-employment rates between white females and white
males has declined over time. Post-Croson the position of white females has improved relative
to that of white males. This is true for the private sector as a whole but especially so in
construction.
5) The magnitude of the difference in self-employment rates between both blacks and Hispanics
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 60 of 801
20
and white males has narrowed post-Croson overall but remained roughly constant in
construction.
6) White females in construction have a significantly higher incorporated self-employment rate
than is found for white males in both periods, but the gap has increased post-Croson.
7) Asians/Pacific Islanders and American Indians have significantly lower self-employment rates
than white males or white females in construction, driven by their much lower incorporated rates.
Table 5 provides supporting evidence for the findings for construction when a much larger
sample from the 2000 census is used. Once again we estimate probit equations which model the
probability that an individual is self-employed, holding constant characteristics. We include the
same set of controls as used in Table 4 using the MORG data. We exclude any case where one
or more of the variables we use have data that are imputed.35 Column 1 is for total self-
employment, column 2 for the unincorporated and column 3 for the incorporated. Overall there
are significant differences between white males and all race and gender groups overall and for
the unincorporated. In column 3 there is a positive and significant coefficient on white females
confirming the results in column of Table 4 using the MORG data. The coefficient on the
Asian/Pacific Islander variable is insignificant in column 3 and smaller than for other groups in
column 1.
The question then is why has there been an increase in incorporated self-employment among
white females in construction and not in other sectors. One possibility is that this has arisen to
take advantage of the various affirmative action programs that exist in construction and what we
are observing is that firms owned by white males are bringing their spouses into the business to
take advantage of the programs. Who are these women? In the 2000 Census there are 3,950
white female incorporated business owners whose status was not imputed. Of these 86% were
married compared with 63% of all white female workers in the construction sector and 68% of
white female unincorporated self-employed in the sector. In terms of family types 71% of the
white female incorporated self-employed were a married couple where both the husband and
wife were in the labor force compared with 51% for all white females in construction and 48% of
the unincorporated self-employed white females.
We examined the occupational mix of both the incorporated and unincorporated self-employed
in construction once again using data from the 2000 Census. We omitted any case where the
class of worker was imputed. White females who are incorporated self-employed are more
likely to be in bookkeeping and accounting and especially likely to be secretaries or
administrative assistants. 43% of white females who were incorporated self-employed were
bookkeepers, secretaries or administrative assistants compared with 0.2% of white men and 23%
of minority women. There is much less difference in the occupation distribution for the
unincorporated self-employed between white women and minority women than is the case for
the incorporated self-employed. It is likely that what is going on here is that some white females
are fronting firms that are actually being run by their white male spouses. They are doing so to
35
For a discussion of the problems of using imputed data in the MORG files of the CPS see Blanchflower and
Bryson (2003, 2004 and Hirsch and Schumacher (2002).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 61 of 801
21
take advantage of the benefits of the affirmative action programs. A number of prosecutions
around the country have occurred because contracts were fraudulently obtained by companies
that were 'fronted' by women and minorities but who were not true owners of the firm and played
no active part in its operations.36
We now move on to examine whether self-employed minority and white female entrepreneurs
receive lower earnings from the businesses they start than white males with similar
characteristics receive from their businesses. The ORG files do not contain data on the earnings
of the self-employed. Annual earnings, whether from wage or self-employment, are available
from the March CPS, also known as the Annual Demographic File. This file also contains the
basic monthly demographic and labor force data. In the March CPS, data on employment,
earnings, and income refer to the preceding year, although demographic data refer to the time of
the survey. Sample sizes are much lower than previously with around 194,000 observations per
annum. Information on self-employed earnings is available for approximately 6,000 cases per
annum. We supplement this information for construction with data on annual earnings in the
year preceding interview (1999) from the 2000 Census.
Table 6 reports estimates of the log of annual self-employment earnings differentials between
white females and minority groups and comparable white males. Columns 1 and 2 are for all
industries for the periods 1978-1990 and 1991-2001 respectively. Columns 3 and 4 are for
construction over the same periods while column 5 is for construction in 1999.
White females, blacks and other minority groups have significantly lower self-employment
earnings than white males with similar personal characteristics. This is true in the private sector
as a whole and in construction using both the March CPS files and the Census. Interestingly,
controlling for occupation in the final column reduces in absolute size the coefficient on the
white female dummy – in part their lower earnings are due to their occupations. This is not the
case for the other minority groups whose negative differentials are essentially unchanged in
columns 5 and 6. Their lower earnings are not because of their occupations.
One important impediment to minority entrepreneurship is lack of capital. As noted by Nobel
Laureate Kenneth Arrow.
"Racial discrimination pervades every aspect of a society in which it is found. It
is found above all in attitudes of both races, but also in social relations, in
intermarriage, in residential location, and frequently in legal barriers. It is also
found in levels of economic accomplishment; that is income, wages, prices paid,
and credit extended" (1998, p.91 italics added)
Blanchflower, Levine and Zimmerman (2003) examined the availability of credit to minority and
female-owned small businesses using data from the 1993 and 1998 National Surveys of Small
36
For example, James Duff was fined more than $22 million and sentenced to nearly 10 years in prison in May 2005
after he pleaded guilty to falsely portraying Windy City Maintenance as woman-controlled and Remedial
Environmental Manpower as minority-controlled and for defrauding the City of Chicago by taking $120 million in
city contracts reserved for minority- and women-owned companies. Other associates were also fined and received
jail time. See the Chicago Sun-Times, May 19th, 2005 and May 21st, 2005 and Chicago Tribune, October 21st, 2005.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 62 of 801
22
Business Finances conducted by the Reserve Board of Governors. They demonstrated that loan
denial probabilities for African-American owned firms are approximately double those for
comparable white-owned firms. Even when African-Americans were able to obtain loans they
have to pay higher interest rates. Comparable but smaller effects are found for Hispanics. These
differences were not explained by differences in creditworthiness or other observables. Such
differences disappeared when the use of credit cards was examined, where the banks were
unaware of the race of the applicant. The authors found that firms owned by minorities are
discriminated against in the credit market. Similar results were found by Cavalluzzo, Cavalluzzo
and Wolken (2002).
Table 7 replicates the finding from Table 3 of Blanchflower, Levine and Zimmerman (2003)
using data from both the 1993 and 1998 surveys to estimate a series of equations estimating the
probability of a loan application being denied. Variables are included to control for region,
industry, type of organization and creditworthiness. It is clear from Table 3 of Blanchflower et
al. (2003) paper that adding further controls for characteristics of the loan, other firm
characteristics has little substantive impact on the results so for ease of exposition we use this
limited set of controls here. In both 1993 and 1998 firms owned by blacks have dramatically
higher probabilities of having their loans denied, holding constant their characteristics. Firms
owned by Asians/Pacific Islanders also have higher denial rates although the size of the impact,
although large, is smaller than for black-owned firms. There is evidence of a large positive
effect n 1998 for Hispanics but not in 1993. We find no evidence of any impact for white
women. In columns 2 and 4 we include a series of interaction terms between the race and gender
variables and a construction dummy to test if the results are different in the Construction
industry. They are not: all of the interaction terms are insignificant. Construction firms owned
by minorities are discriminated against in the credit market.
In the two surveys of small business finances used above, business owners were asked variously
about the main problems that they faced. In the 1993 survey owners were asked about a variety
of problems that they had experienced during the preceding 12 months. They were also asked to
report the most important issues facing the firm over the following 12 months. In the 1998
survey respondents were asked to report the "single most important problem facing your business
today". Credit availability turned out to be the most important problem in all three cases for firms
owned by African-Americans and to a lesser extent Hispanics. The problem appeared to be even
more serious in construction than elsewhere.
a) Problems firms experienced during preceding 12 months - credit
market conditions. (% reporting serious problem)
All Black Hispanic
All industries 14 31 23
Construction 14 34 23
Source: Authors' calculations from 1993 NSSBF
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 63 of 801
23
b) Percentage of firms reporting most important issues affecting them
over the next 12 months – credit availability
All Black Hispanic
All industries 6 21 5
Construction 7 26 3
Source: Authors' calculations from 1993 NSSBF.
c) What is the single most important problem facing your business today?
% reporting "financing and interest rates"
All Black Hispanic
All industries 7 18 10
Construction 10 36 11
Source: Authors' calculations from 1998 NSSBF.
In the 1992 Characteristics of Business Owners Survey conducted by the Bureau of the Census
firms were asked to report the impact of various kinds of costs upon their profitability (Table 1,
p.21). Black and Hispanic-owned firms reported stronger negative impacts of credit market
conditions and a lack of financial capital; there are no strong race or gender effects for the
various other reasons given. The survey also reported on reasons why a discontinued business
was unsuccessful. Black-owned and to a lesser degree Hispanic-owned firms were much more
likely to report that the reason was due to lack of access to business or personal loans or credit
than was true for other races.37
A recent study published by the U.S. Chamber of Commerce (2005) confirms the findings in
Blanchflower, Levine and Zimmerman (2003). The survey was conducted in March and April
2005 and detailed the financing problems experienced by small business owners, 95% of whom
had less than 100 employees: 1080 business owners were interviewed and reported that minority
businesses rely heavily on credit cards to fund their businesses, often do not apply for credit,
even though they need it, for fear of being denied and were especially likely to need working
capital. In particular they report that availability of credit is their top problem, exactly as
reported by Blanchflower, Levine and Zimmerman. The biggest difference in responses between
minorities and Caucasian men and women was availability of credit: 19% of Caucasian males
report credit as their top problem compared with 54% for minority males – a 35 percentage point
difference. There was a 15 percentage point difference for women. In no other category is there
more than a 10 percentage point difference for men or women.
37
When asked if lack of financial capital was a serious problem affecting business profitability 29% of firms owned
by white males in the CBO survey answered in the affirmative compared with 46% owned by blacks and 38% by
Hispanics. For firms that were discontinued 7% of firms owned by white males reported it was due to lack of access
to business capital compared to 16% for firms owned by blacks and 9% for Hispanics. A further 3% of white males
said it was due to lack of personal capital compared to a further 8% for firms owned by blacks and 6% for
Hispanics. See tables 3a and 3b, Blanchflower, Levine and Zimmerman (1998)
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 64 of 801
24
Overall, we have shown that there has been no significant change over time in the relative
position of African-Americans compared to white males in construction. The probability that
they are self-employed is markedly lower than white males and is unchanged over time. Their
self-employment earnings remain markedly lower. The probability of being self-employed for
Hispanics compared to white males has widened over time although the magnitude of the effects
are smaller than for African Americans. The position of white females has improved over time,
especially in terms of self-employment probabilities, although it is unclear whether this is simply
because firms owned by their white male spouses are transferring ownership to take advantage of
the affirmative action programs. Credit market discrimination appears to be important in
construction and especially so for firms owned by African Americans and to a lesser extent for
Hispanics.
7. Conclusion
The main findings of this paper are that despite the existence of various programs designed to
improve the position of women and minorities in construction, little has changed in the last
twenty five years. We presented evidence that showed that where race conscious affirmative
action programs exist they appear to generate significant improvements. For example, there is
evidence that when they are put in place the self-employment rate of minorities rises. We also
showed that when these programs are removed or replaced with race-neutral programs the
utilization of minorities and women in public construction declines rapidly.
The Croson decision in 1989, however, made it very difficult to maintain affirmative action
programs and many were subsequently struck down by the courts. Where they did remain in
existence many were watered down to the extent that many only had race-neutral components
left. The evidence we have available to us suggests that very rapidly after the race and gender
conscious programs were removed the utilization of firms owned by women and minorities
collapsed. The position is particularly serious for African-Americans and less so for Hispanics.
They have lower business formation rates, lower self-employment rates and earnings than white
males. The position of white females, in contrast, appears to have improved, but some of this
improvement is likely via 'front' companies where the white female is reported as the owner but
the firm is actually run by their white male spouse.
Several recent cases such as Concrete Works, Gross Seed, Sherbrooke Turf, BAGC, Northern
Contracting appear to have stemmed, at least temporarily, the judicial tide pushing against race
conscious affirmative action programs in contracting. However, such affirmative action
programs have not achieved their objectives of improving the position of minority firms in
construction, not because they don't work, as it appears that they do, but because they have not
been allowed to work by non-minority contractors and by the courts. Discrimination against
minorities in general and African-Americans and Hispanics in particular continues to persist in
construction. Plus ca change, plus c'est la meme chose.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 65 of 801
Table 1. Participation of Disadvantaged Business Enterprises on Federal construction projects
with goals and state construction projects without goals, 2002.
State funded Federal funds
Alabama 3.46% 7.23%
Arizona 2.80% 7.22%
Connecticut 6.79% 15.86%
Delaware 3.95% 8.16%
Florida 8.30% 5.59%
Idaho 3.77% 9.81%
Kansas 1.93% 8.26%
Kentucky 3.87% 10.17%
Michigan 5.06% 12.65%
Minnesota 4.42% 6.63%
Mississippi 12.11% 10.68%
Nebraska 3.68% 7.24%
Nevada 1.11% 6.61%
New Hampshire 14.27% 22.11%
Oregon 1.68% 7.88%
Utah 2.57% 5.81%
Vermont 6.31% 12.85%
Wyoming 16.32% 14.40%
Average 5.7% 10.0%
Source: Federal Highways Administration
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 66 of 801
Table 2a. Summary of MWBE Participation in Contracts Awarded by MnDOT 1996-1999
Minority Minority Caucasian Total MWBE
Year Total Federal $ men $ % Women $ % Women $ % $ awarded %
FFY96 $209,964,333 $3,467,523 1.65 $176,480 0.08 $20,056,922 9.55 $23,700,825 11.29
FFY97 $241,031,074 $6,002,497 2.49 $108,519 0.05 $19,441,750 8.07 $25,552,766 10.80
FFY98 $285,081,558 $6,589,434 2.31 $129,647 0.04 $22,515,424 7.89 $29,234,500 10.25
FFY99 $295,835,471 $841,239 0.28 $40,028 0.01 $5,761,276 1.95 $6,642,543 2.25
Table 2b. Summary of MWBE Participation in Contracts Awarded by MnDOT 2000Q1-2001Q1
Total MWBE
Year Total Federal $ Minority Men $ % Women $ % Total MWBE $ %
Q1-FFY00 $22,099,798.27 $0 0 $0 0 $0 0
Q2-FFY00 $55,219,499.00 $0 0 $126,224.00 0.23 $126,224.00 0.2
Q3-FFY00 $172,252,885.00 $353,805.00 0.21 $2,005,974.00 1.16 $2,359,779.00 1.4
Q4-FFY00 $79,898,976.00 $1,128,366.51 1.41 $3,364,234.50 4.21 $4,492,601.02 5.6
FFY00 $329,471,158.27 $1,482,171.51 0.45 $5,496,432.50 1.67 $6,978,604.03 2.1
Q1-FFY01 $14,184,474.65 $1,003,401.33 7.07 $234,880.11 1.66 $1,238,281.51 8.7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 67 of 801
Table 3. Self-employment rates 1979-2005 by race and gender
All industries Construction industry
White White Black Hispanic White White Black Hispanic
male female male female
A) Incorporated
1979-83 3.6 1.0 0.3 0.8 4.8 4.9 1.1 1.2
1984-88 4.2 1.3 0.4 1.0 5.3 6.7 1.2 1.3
1989-93 4.5 1.5 0.5 1.1 5.9 8.5 1.6 2.2
1994-98 5.0 2.1 0.8 1.2 6.7 11.2 1.9 2.2
1999-04 5.2 2.1 1.1 1.3 7.6 9.8 2.3 1.8
B) Unincorporated
1979-83 11.5 5.6 3.2 4.7 19.4 7.4 10.0 10.9
1984-88 11.4 6.4 3.4 5.4 19.9 10.3 11.3 11.6
1989-93 11.3 6.6 3.7 5.7 21.1 12.0 14.6 12.6
1994-98 10.4 7.2 3.7 5.2 19.6 14.3 12.2 9.8
1999-04 9.4 6.6 3.9 5.2 18.9 11.5 13.2 10.4
C) All Self-employed
1979-83 15.1 6.6 3.5 5.5 24.2 12.4 11.1 12.1
1984-88 15.6 7.6 3.8 6.5 25.2 17.0 12.5 12.8
1989-93 15.8 8.1 4.2 6.8 27.0 20.5 16.2 14.8
1994-98 15.4 9.3 4.4 6.4 26.3 25.5 14.1 12.0
1999-04 14.6 8.6 4.9 6.5 26.5 21.2 15.5 12.2
Source: MORG files of the CPS (weighted). For years 1979-1982 the 1970 SIC is used to define industry. 1980 SIC is used until
2002 which is then replaced with the 2000 NAICS
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 68 of 801
Table 4. Private Sector Self-employment Probabilities, 1979-2004.
(1) (2) (3) (4) (5) (6) (7)
1979-1991 1992-2002 1979-1991 1979-1991 1992-2004 1992-2004 1992-2004
All industries All industries Construction Construction Construction Construction Construction
Self-employed Self-employed Self-employed Incorporated Self-employed Unincorporated Incorporated
White Female -.0458 -.0271 -.1083 .0029 -.0492 -.0689 .0158
(142.24) (66.35) (32.95) (2.25) (13.89) (22.26) (9.53)
Black -.0556 -.0440 -.1174 -.0259 -.1103 -.0615 -.0372
(116.14) (65.57) (29.46) (16.37) (24.43) (15.23) (18.22)
Hispanic -.0399 -.0374 -.0817 -.0204 -.0898 -.0649 -.0228
(67.40) (52.93) (17.65) (10.26) (21.12) (17.56) (10.84)
Other races -.0256 n/a -.1192 -.0160 n/a n/a n/a
(33.69) (17.94) (6.11)
Asian -.0060 -.0793 -.0520 -.0193
(5.86) (8.74) (6.41) (4.88)
American Indian -.0213 -.0945 -.0654 -.0242
(11.70) (10.83) (8.65) (5.68)
Construction -.0506 -.0481 n/a n/a n/a n/a
(59.21) (44.62)
Years of education .0101 No .0172 .0084 No No No
(171.23) (41.36) (49.81)
Schooling dummies No Yes No No Yes Yes Yes
Year dummies Yes Yes Yes Yes Yes Yes Yes
Industry Dummies Yes Yes No No No No No
State Dummies Yes Yes Yes Yes Yes Yes Yes
N 2,808,248 1,994,519 196,951 196,951 158,939 158,939 158,939
Pseudo R2 .2537 .2210 .0886 .1130 .0718 -70608 -33957
Log Likelihood -819449 -609,176 -100,987 -36171 -81564 .0525 .0940
Notes: Probit analysis is performed using the dProbit command in STATA on a sample of workers. The percentages reported here are the coefficients from the Probit analysis and
indicate the percentage point differences in self-employment rates between the indicated group and Whites. Equations also include age and its square as controls. Self-employment
includes both the incorporated and unincorporated. T-statistics in parentheses. Source: Outgoing Rotation Group files of the Current Population Survey, 1979-2004.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 69 of 801
Table 5. Private Sector Self-employment Construction Industry Dprobits, 2000.
(1) (2) (3)
2000 2000 2000
Self-employed Unincorporated Incorporated
White Female -.0859 -.0923 .0085
(39.55) (49.90) (6.97)
Black -.1064 -.0693 -.0305
(36.95) (28.02) (19.27)
Hispanic -.0788 -.0605 -.0151
(26.14) (23.72) (8.67)
Other races -.0844 -.0635 -.0183
(31.12) (27.75) (11.49)
Asian/Pacific Islander -.0395 -.0287 -.0066
(6.06) (5.14) (1.87)
American Indian -.0847 -.0576 -.0230
(14.77) (11.84) (7.02)
Schooling dummies Yes Yes Yes
State Dummies Yes Yes Yes
N 436,437 436,437 436,437
Pseudo R2 .0707 .0572 .0582
Log Likelihood -229178 -242383 -108475
Notes: Probit analysis is performed using the dProbit command in STATA on a sample of workers.
The percentages reported here are the coefficients from the Probit analysis and indicate the
percentage point differences in self-employment rates between the indicated group and Whites.
Equations also include age and its square as controls. Self-employed includes both the
incorporated and unincorporated. State dummies include Puerto Rico. All cases with imputed data
to one or more variables are deleted.
Source: 2000 Census. T-statistics in parentheses.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 70 of 801
Table 6. Log Self Employment Earnings Equations, 1979–2002
(1) (2) (3) (4) (5) (6)
All All Construction Construction Construction Construction
1978-1990 1991-2001 1978-1990 1991-2001 1999 1999
White female -.729 -.617 -.835 -.839 -.702 -.513
(68.07) (31.34) (21.63) (15.73) (28.67) (17.81)
Black -.500 -.591 -.428 -.323 -.387 -.350
(15.64) (14.85) (5.73) (2.40) (12.09) (11.03)
Hispanic -.278 -.390 -.252 -.145 -.146 -.118
(9.46) (9.8) (3.96) (1.38) (4.96) (4.05)
Other races -.328 -.221 -.208 -.180 -.160 -.122
(8.29) (3.41) (1.79) (0.84) (5.08) (4.52)
Asian/Pacific Islanders .004 .045
(0.08) (0.85)
Native Americans -.397 -.361
(7.15) (6.60)
Occupation dummies (284) No No No No No Yes
Education controls (16) No Yes No Yes Yes Yes
Year dummies (13) Yes Yes Yes Yes No No
State dummies (51) Yes Yes Yes Yes Yes Yes
Industry dummies (88) Yes Yes No No No No
N 82,094 55,639 12,577 8,446 64,188 64,188
Adjusted R2 .177 .128 .077 .064 .064 .093
Notes: The percents indicate the percent difference in self-employment earnings between the indicated group and Whites. Controls also include age, age squared
and a full set of year dummies.
Source: Annual Demographic (March) files of the Current Population Survey for 1979-2002 and 2000 Census.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 71 of 801
31
Table 7. Determinants of Loan Denial Rates
1993 1993 1998 1998
(1) (2) (3) (4)
.3132 .3338 .2923 .2471
Black
(7.16) (6.73) (4.40) (3.45)
.1553 .1646 .1822 .1437
Asian/Pacific islander
(2.84) (2.84) (2.66) (2.01)
.0522 .0113 .2528 .2446
Hispanic
(1.05) (0.21) (3.45) (3.20)
.0373 .0305 .0581 .0411
Female-Owned
(1.26) (0.94) (1.38) (0.94)
-.0713 .2537
Black*Construction
(0.89) (1.43)
-.0701 .3306
Asian/Pacific Islander*Construction
(0.48) (1.44)
.1908 .0373
Hispanic*Construction
(1.51) (0.18)
.0113 .1806
Female-Owned*Construction
(0.21) (1.15)
N 1986 1986 962 962
Pseudo R2 .1651 .1667 .2633 .2671
Chi2 387.3 391.1 289.7 293.9
Log likelihood -979.1 -977.2 -405.3 -403.1
Source: Authors’ calculations from 1993 and 1998 NSSBF. Notes: controls for 1993 are 8 region
dummies, 51 industry dummies, 3 types of organization dummies and 4 creditworthiness
variables. Controls for 1998 are 8 region dummies, 37 industry dummies, 43 type of organization
dummies, 4 Dunn and Bradstreet credit rating variables and 4 creditworthiness variables.
Reported estimates are derivatives from Probit models, t-statistics are in parentheses)
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 72 of 801
Figure 1. Disadvantaged Business Enterprise (DBE) Participation Levels in States
That Use DOT's DBE Program But Have no State DBE Program
DOT DBE Program (Federal-Aid Highways) No DBE Program ( State-Funded Highways)
18.0%
15.7% 16.2%
15.5% 15.1%
16.0%
Percent of Total Dollars Spent
14.0% 12.7%
11.9% 12.4% 12.1%
12.0% 10.5%
10.0% 8.9%
8.0%
6.0% 5.2%
3.8% 3.6% 3.8%
4.0% 2.9%
1.4% 1.7%
2.0% 0.9% 0.4% 0.0%
0.0%
Ar as ut are an
a
ga
n ur i sk go
n nd
iz ns ic so a la
on ka
ct aw si hi is ra re Is
a Ar
ne el Lo
ui
M
ic M eb O de
on D N ho
C R
Fiscal Year 1996
Case 1:23-cv-03424-TWT Document
3359-4 Filed 08/31/23 Page 73 of 801
Figure 2: MWBEs' Participation in Louisiana's State-Funded Transportation Contracting, 1992-2000
(Source: GAO's analysis of data from Louisiana's Department of Transportation)
Case 1:23-cv-03424-TWT Document
3459-4 Filed 08/31/23 Page 74 of 801
Figure 3: DBEs' Participation in Federal Transportation Contracting in Minnesota, 1995-2000
(Source: GAO's analysis of data from Minnesota's Department of Transportation).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 75 of 801
35
References
Arrow, K.J. (1998), 'What has economics to say about racial discrimination?', Journal of
Economic Perspectives, 12(2), pp. 91-100.
Ashenfelter, O. (1972), 'Race discrimination and trade unionism', Journal of Political Economy,
Volume 80, Issue 3, Part 1, May-June pp. pp. 435-464.
Blanchflower, D.G., (2000), 'Self-employment in OECD countries', Labour Economics, 7,
September, pp. 471-505.
Blanchflower, D.G., (2004), 'Self-employment: more may not be better', Swedish Economic
Policy Review, 11(2), Fall, pp.15-74.
Blanchflower, D.G. and A. Bryson (2003), 'Changes over time in union relative wage effects in
the UK and the US revisited', International Handbook of Trade Unions, Edward Elgar, edited by
John Addison and Claus Schnabel.
Blanchflower, D.G. and A. Bryson (2004), 'What effect do unions have on wages now and would
Freeman and Medoff be surprised?', Journal of Labor Research, 25(3), Summer, pp. 383-414 and
in What do unions do?: the evidence twenty years later, edited by James T. Bennett and Bruce E.
Kaufman, forthcoming.
Blanchflower, P. Levine and D. Zimmerman (1998), 'Discrimination in the small business credit
market', National Bureau of Economic Research Working Paper #W6840, Cambridge, MA.
Blanchflower, P. Levine and D. Zimmerman (2003), 'Discrimination in the small business credit
market', Review of Economics and Statistics, November, 85(4), pp. 930-943.
Blanchflower D.G. and A.J. Oswald (2004), 'Wellbeing over time in Britain and the United
States', Journal of Public Economics, 88(7-8), pp.1359-1386.
Cavalluzzo, K., L. Cavalluzzo and J. Wolken (2002), 'Competition, small business financing and
discrimination: evidence from a new survey', Journal of Business, 75, October, pp.641-679.
Chandra, A. and J. Skinner (2003), 'Geography and racial health disparities,' in Critical
perspectives on racial and ethnic differences in health in late life, edited by N. B. Anderson, R.A.
Bulatao, and B. Cohen, Panel on race, ethnicity, and health in later life, National Research
Council, pp.604-639.
Chay, K.Y. and R. W. Fairlie (1998), 'Minority business set-asides and black self-employment,
Working Paper, December.
Cutler, D.M. Glaeser, E.L. and J.L. Vigdor (1999), 'The rise and decline of the American ghetto',
Journal of Political Economy, 107(3), pp. 455-506.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 76 of 801
36
Enchautegui, M.E., M. Fix, P. Loprest, S. von der Lippe, and D. Wissoker (1997), Do Minority-
Owned Businesses Get a Fair Share of Government Contracts? Washington, D.C.: Urban
Institute.
Fairlie, R. W. (1999), 'The absence of the African-American owned business: an analysis of the
dynamics of self-employment', Journal of Labor Economics, 17(1), pp. 80-108.
Fairlie, R.W. and B.D. Meyer (2000), 'Trends in self-employment among white and black men
during the twentieth century', Journal of Human Resources, XXXV(4), pp. 643-669.
Feinstein, J.S. M.K. Block and F.C. Nold (1985), 'Asymmetric information and collusive
behavior in auction markets', American Economic Review, 75, June, pp. 441-460.
Freeman, R.B (2000), 'Disadvantaged young men and crime' in Youth employment and
joblessness in advanced countries, edited by D.G. Blanchflower and R.B. Freeman, University of
Chicago Press and NBER.
Fryer, R.G. and S. Levitt (2003), 'The causes and consequences of distinctively black names',
NBER Working Paper #9938.
General Accounting Office (2001), Disadvantaged business enterprises, Report to Congressional
Committees, GAO-01-586, June, Washington DC.
Glover, R.W (1977), Minority enterprise in construction, Praeger Publications.
Gupta, S. (2001), 'The effect of bid rigging on prices: a study of the highway construction
industry' , Review of Industrial Organization, December, 19(4), pp. 453-467
Gupta, S. (2002), 'Competition and collusion in government procurement auction market',
Atlantic Economic Journal, 30(1), March, pp. 13-25
Heckman, J., T. Lyons and P. Todd (2000), 'Understanding black-white wage differentials, 1960-
1990,' American Economic Review, 90(2), pp. 344-49.
Hirsch, B.T., and E.J. Schumacher (2004), 'Match bias in wage gap estimates due to earnings
imputation', Journal of Labor Economics, Vol. 22, No. 3, July, pp. 689-722.
Holt, C. (2003) 'Strict constitutional scrutiny is not fatal in fact: federal courts uphold affirmative
action programs in public contracting,' Labor Law Journal, Winter, 54(4), pp. 248-262.
Ingraham, Allan, T. (2005), 'A test for collusion between a bidder and an auctioneer in sealed-bid
auctions', Contributions to Economic Analysis & Policy, Volume 4, Issue 1, pp. 1-32
Jencks, C., and M. Phillips (1998), ‘The black-white test score gap’, Washington DC: The
Brookings Institute.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 77 of 801
37
Kessler, D.P. and L.F. Katz (2001), 'Prevailing Wage Laws and construction labor markets',
Industrial and Labor Relations Review, 54(2), January, pp. 259-274.
Marshall, R. (2000), 'Minority and female business development after Croson', University of
Texas Working Paper.
Porter, R. H. and D. J. Zona (1993), 'Detection of bid-rigging in procurement auctions,' Journal
of Political Economy, 101(3), pp. 518-538.
Smith, J.P. and F.R. Welch (1989), 'Black economic progress after Myrdal', Journal of Economic
Literature, 27(2), June, pp.519-64.
U.S. Chamber of Commerce (2005), Access to capital: what funding sources work for you?, U.S.
Chamber of Commerce, May, Washington DC.
U.S. Commission on Civil Rights (2005), Federal procurement after Adarand, September,
Washington DC.
Ventura, S. and C. Bachrach (2000), 'Non-marital childbearing in the United States, 1940-1999,'
National Vital Statistics Reports, 48(16), Hyattsville, MD: National Center for Health Statistics.
Wainwright, J. (2000), Racial discrimination and minority business enterprise: evidence from the
1990 census, Studies in Entrepreneurship Series, New York and London: Garland Publishing.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 78 of 801
Exhibit C
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 79 of 801
Entering Entrepreneurship: Racial Disparities in the Pathways into Business Ownership
Authors: M’Balou Camara, Khaing Zaw, Darrick Hamilton, and William Darity Jr.
The Samuel DuBois Cook Center on Social Equity, Duke University
The Institute on Assets and Social Policy, Brandeis University
The Kirwan Institute for the Study of Race and Ethnicity, The Ohio State University
October 2019
_____________________________________________________________________________
1. Introduction
The current landscape of entrepreneurship in the United States is radically different than it was a
century ago, but the connections between entrepreneurship and the capital needed to finance it still greatly
contribute to today’s racially-embedded disparities. Research on entrepreneurship often highlights that
business ownership is a key element for building wealth (AEO, 2017). Moreover, a substantial literature
suggests that the story of business ownership does not look the same for blacks as it does for whites.
While black Americans are almost twice as likely to start a business, black-owned businesses persistently
lag behind their white counterparts, appear disproportionately less in high-revenue-earning industries, and
fail at higher rates. In fact, even when comparing the same industries, black-owned businesses that do
survive have fewer employees and smaller revenues than white-owned businesses (Kollinger & Minniti,
2006; Fairlie & Krashinsky, 2012; Bates, 1998).
Despite past studies, this body of work is silent on the role of prior wealth in shaping the
pathways into business ownership and the corresponding racial disparities associated with entering
entrepreneurship. On this account, this report aims to address the following question: How do
characteristics of black business owners (or black entrepreneurs) and black-owned businesses
differ from characteristics of white business owners and white-owned businesses at the time of
entry into business ownership? This study aims to contextualize patterns of initial ownership and to
shed some light on the mechanisms and processes by which racial stratification in business ownership
persists. Insight into racial disparities in business will move us one step closer to understanding how we
can tackle structural inequality in contemporary America.
2. Description of Racial Disparities in the Rate of Business Ownership
2.1 Employer Businesses vs Nonemployer Businesses, by Race
1
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 80 of 801
Based on the US Census’ ASE and SBO, “business ownership” by race, gender or ethnicity is
defined as having 51 percent or more of the stock or equity in a business. Therefore, a “black-owned
business” is one in which 51 percent or more of the stock or equity is owned by black Americans. But one
cannot fully understand how racial disparities in business ownership operate without considering the
difference between nonemployer and employer businesses. All privately-owned U.S. businesses can be
defined as the sum of nonemployer businesses—those that have no paid employees other than the
owner—and employer businesses—those with at least one other paid employee. While nonemployer
businesses outnumber employer businesses in the US, the latter generate a significantly larger proportion
of total annual revenue as measured by sales (AEO, 2017). Employer businesses receive nearly 97 percent
of the total annual revenue but comprise less than 20 percent of all privately-owned businesses. A
nonemployer business often requires less start-up capital than does an employer business to acquire a
business (SBCS, 2018). According to the 2012 Survey of Business Owners (SBO), 35.4 percent of all
nonemployer businesses used less than $5,000 in start-up capital; only 17.9 percent of all employer
businesses did likewise. Conversely, only 4.5 percent of all nonemployer used more than $100,000 of
capital to start or acquire the business, compared to 16.7 percent of all employer businesses1.
Across race, the dynamics distinguishing employer and nonemployer business ownership are
much more pronounced. Roughly 79% of all white-owned businesses and 96% of black-owned
businesses are nonemployer, respectively (SBO, 2012). Nevertheless, while 7.7% of sales revenue of all
white-owned businesses comes from those that are nonemployer, 31% of all sales revenue of black-
1
20.5 percent of employer businesses and 10.4 percent of nonemployer businesses do not know how much capital
was used to start or acquire the corresponding business.
2
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 81 of 801
owned businesses are nonemployer-generated (SBO, 2012). In sum, nonemployer businesses (despite
bringing in relatively less revenue) play a larger role amongst black-owned businesses than they do
amongst white-owned businesses. Understanding these trends is important for conceptualizing how the
implications of the racial disparities may vary depending on the types of businesses we observe.
2.2 Racial Disparities in the Rate of Business Ownership Across the United States
While white Americans represent nearly 72.9 percent of the U.S. population2, 78 percent of all
private U.S. businesses are white-owned, according to the 2012 Survey of Business Owners (SBO). On
the other hand, black Americans represent approximately 13.3 percent of the U.S. population, but own
just 9 percent of all the nation’s private businesses. Put simply, black Americans across the country are
disproportionately underrepresented as business owners while white Americans are disproportionately
overrepresented. The portrait is even more unequal when comparing the average annual business revenues
and average salaries of employees of black-owned and white-owned employer businesses (See Appendix
B).
As displayed in Table 2.1 the racial composition across all fifty states varies dramatically from a
1 percent working-age black population in Idaho to a 38 percent working-age black population in
Mississippi. Analogously, the black-owned business composition varies from 0 percent in Montana to 28
percent in Georgia. The discrepancy between the racial composition of the entire U.S. population versus
the racial composition of privately-owned U.S. businesses illustrates one key aspect of racial disparities in
business ownership. Across every single state the rate of business ownership for black Americans falls
below that group’s prevalence in the overall population. In other words, in no single state are black
Americans as likely to own a business as the average resident of that state. On the other hand, white
Americans are overrepresented in business ownership in every state except for California, where their true
working age proportion equally represents their business ownership composition (See Appendix A).
Table 2.1: Black American Share of Population and Business Ownership (2012)
Share of total Share of Share of
Share of all
State working age employer nonemployer
businesses
population businesses businesses
Alabama 0.27 0.20 0.03 0.23
Alaska 0.05 0.02 0.01 0.02
Arizona 0.05 0.03 0.01 0.04
Arkansas 0.16 0.09 0.02 0.11
California 0.07 0.06 0.02 0.07
Colorado 0.05 0.02 0.01 0.03
2
2012 American Community Survey Population Estimates (ACS)
3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 82 of 801
Connecticut 0.12 0.06 0.01 0.07
Delaware 0.23 0.11 0.03 0.14
Florida 0.17 0.12 0.03 0.15
Georgia 0.32 0.28 0.06 0.33
Hawaii 0.03 0.01 0.01 0.01
Idaho 0.01 0.00 0.00 0.00
Illinois 0.15 0.13 0.02 0.16
Indiana 0.10 0.07 0.01 0.09
Iowa 0.04 0.02 0.00 0.02
Kansas 0.07 0.03 0.01 0.04
Kentucky 0.09 0.04 0.01 0.05
Louisiana 0.33 0.23 0.04 0.27
Maine 0.02 0.01 0.00 0.01
Maryland 0.31 0.24 0.06 0.28
Massachusetts 0.09 0.04 0.01 0.05
Michigan 0.15 0.13 0.02 0.15
Minnesota 0.06 0.04 0.01 0.05
Mississippi 0.38 0.28 0.04 0.33
Missouri 0.12 0.08 0.03 0.09
Montana 0.01 0.00 0.00 0.00
Nebraska 0.06 0.03 0.01 0.04
Nevada 0.10 0.07 0.01 0.09
New Hampshire 0.02 0.01 0.00 0.01
New Jersey 0.15 0.08 0.02 0.10
New Mexico 0.03 0.02 0.01 0.02
New York 0.18 0.12 0.02 0.14
North Carolina 0.23 0.14 0.04 0.17
North Dakota 0.02 0.01 0.00 0.01
Ohio 0.13 0.09 0.02 0.11
Oklahoma 0.09 0.04 0.01 0.05
Oregon 0.03 0.02 0.01 0.02
Pennsylvania 0.12 0.06 0.01 0.07
Rhode Island 0.09 0.04 0.01 0.05
South Carolina 0.28 0.17 0.04 0.20
South Dakota 0.02 0.01 0.00 0.01
Tennessee 0.18 0.14 0.03 0.16
Texas 0.13 0.10 0.03 0.11
Utah 0.02 0.01 0.00 0.01
Vermont 0.02 0.01 0.00 0.01
Virginia 0.2 0.13 0.04 0.15
Washington 0.05 0.03 0.01 0.03
West Virginia 0.04 0.02 0.01 0.02
Wisconsin 0.07 0.05 0.01 0.06
Wyoming 0.02 0.01 0 0.01
United States 0.13 0.09 0.02 0.11
Source: 2012 American Community Survey Population Estimates and 2012 Survey of Business Owners (SBO)
When observing the rate of business ownership across states, black-owned businesses are also
disproportionately underrepresented in the employer businesses. To understand the scale of this disparity,
we also observe that average annual sales of black-owned businesses and average annual salaries of
4
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 83 of 801
employees from black owned businesses significantly lag behind those of white-owned businesses. For
example, in North Carolina, the average black-owned employer business generates $666,168 in annual
revenue sales while the average white-owned employer business generates $2,037,372, nearly twice as
much. The average annual salary of an employee of a black-owned business in North Carolina makes
$23,544. For a white-owned business, the average annual salary is $33,690 (See Appendix B).
This disproportionality, however, is not as pronounced when only looking at nonemployer businesses. In
fact, the nonemployer black-owned business composition in California, Georgia, Illinois, and Michigan is
either an equal or overrepresentation of the black working-age proportion in those respective states. The
relatively high nonemployer entrepreneurial activity (as opposed to employer activity) among black-
owned businesses suggests that they are lacking in the number of rapidly-growing, “commercially-
oriented keystone organizations” (McKoy and Johnson, 2018). Why are there are fewer black-owned
businesses (and at varying magnitudes) than we might expect given the population size across different
states? In this report, we aim to shed the light on how racial differences in prior wealth help explain the
racial differences in four dimensions of entry into business ownership: (1) the method of acquiring a
business, (2) the source and amount of start-up capital used for the business, (3) the duration or age of the
business, and (4) the business performance.
3. Racial Disparities in How Owners Enter Entrepreneurship
What is the portrait of entry into business ownership by race? Based on the latest data available
on business owners, the 2012 and 2015 Survey of Business Owners, the most common method of
acquiring ownership of a business is by founding or starting one, as summarized in tables 3.1 and 3.2.
However, quantitative racial disparities exist in the likelihood of entering entrepreneurship by each of the
5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 84 of 801
four possible methods: (1) founding or starting the business, (2) purchasing the business, (3) inheriting the
business, or (4) receiving a transfer of ownership of the business3. Each method of entry into
entrepreneurship suggests a very different story about the initial experiences, barriers, and choice sets that
business owners must navigate. Purchasing a firm typically requires more capital than starting or
founding one, and establishing a new firm often entails a tougher and longer road to startup. Those
businesses that are purchased, inherited, or transferred are typically bigger and more mature than those
that are newly established (SBCS, 2017).
Table 3.1. How Owners Initially Acquired Employer businesses, by Race (2016)
How the owner Number of black owners of Number of white owners of
initially acquired the business respondent employer-firms respondent employer-firms
Founded or started 72,364 3,183,089
Purchased 11,460 951,604
Inherited 2,432 203,525
Transfer of ownership or gift 3,767 346,640
Total reporting 88,202 4,569,600
Table 3.2. How Owners Initially Acquired Nonemployer businesses, by Race (2012)4
How the owner Number of black owners of Number of white owners of
initially acquired the business respondent nonemployer businesses respondent nonemployer businesses
Founded or started 781,464 12,707,791
Purchased 55,609 1,105,094
Inherited 11,953 373,395
Transfer of ownership or gift 24,206 512,580
Total reporting 862,833 14,510,034
As shown in Figure 3.1, 82 percent of black owners of employer businesses founded or started
their businesses, whereas the corresponding figure for white owners is 69.7 percent. This finding means
that black owners of employer businesses are 17.6 percent more likely than white owners to enter
business ownership by founding or starting a new enterprise. Conversely, white owners of employer
businesses are 60 percent more likely to enter business ownership via acquisition by purchase, 60.7
3
Note that respondents could select more than one answer to the question regarding initial mode of acquisition.
Therefore, the cumulative amount of responses across all four categories may exceed the total reporting amount.
This applies to Table 3.1, Table 3.2, Figure 3.1, Figure 3.2, and Table 3.3.
4
The latest survey data on owners of nonemployer businesses is from 2012.
6
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 85 of 801
percent more likely to acquire by way of inheritance, and nearly 76.7 percent more likely to acquire
through a transfer of ownership or gift than are black business owners of employer businesses.
Figure 3.1. How Black and White Owners Initially Acquired their Employer Businesses (2016)
As expected, it is even more common for owners of nonemployer businesses than owners of
employer businesses to acquire a business by founding or starting it, regardless of the owner’s race.
Nearly 91 percent of black owners and 88 percent of white owners of nonemployer businesses founded or
started their businesses. Figure 3.2 illustrates that, like owners of employer businesses, white owners of
nonemployer businesses are more likely to enter business ownership through purchase, through
inheritance, and through a transfer of ownership or gift than are black owners of nonemployer businesses.
Figure 3.2. How Black and White Owners Initially Acquired their Nonemployer Businesses (2012)
7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 86 of 801
We observe similar patterns of initially acquiring employer businesses across nine different
sectoral categories. As displayed in Table 3.3, black business owners are more likely than white business
owners to found or start their own business in each of these sectors except for the information sector.
However, white business owners are more likely than black business to receive business ownership as a
transfer or gift in all of nine industry categories. Furthermore, white business owners are more likely than
black business owners to initially acquire a business by purchasing it in every category except for the
information sector. This gap widens within higher-end sectors. In fact, more than half (66 percent) of
white-owned businesses in the comparatively more lucrative management sector (bolded in the table)
were acquired either by purchase, inheritance, or transfer, while only about 33 percent of black-owned
businesses in that sector were acquired in one of those three ways.
Table 3.3. How Owners Initially Acquired Employer Businesses, by Industry (2016)
Transfer of
Founded or
Industry Inherited Purchased ownership or
started
gift
Black White Black White Black White Black White
Accommodation and food services 0.66 0.60 0.03 0.04 0.33 0.34 0.03 0.06
Administrative and support and
waste management and remediation 0.87 0.76 0.03 0.03 0.08 0.16 0.03 0.07
services
Agriculture, forestry, fishing and
0.75 0.73 0.25 0.06 0.01 0.14 0% 0.10
hunting
Construction 0.88 0.77 0.04 0.04 0.04 0.12 0.05 0.09
Finance and insurance 0.86 0.71 0.03 0.05 0.08 0.19 0.05 0.08
Information 0.69 0.73 0.03 0.04 0.25 0.18 0.04 0.08
Management of companies and
0.73 0.44 0.11 0.16 0.15 0.29 0.07 0.21
enterprises
Manufacturing 0.72 0.54 0.08 0.09 0.19 0.27 0.09 0.14
Other services (except public
0.75 0.66 0.06 0.04 0.16 0.25 0.06 0.07
administration)
In short, black business owners are far more likely to enter into business ownership by starting a
new enterprise than white business owners, which coincides well with the proposition that black
Americans have historically been undercapitalized because of barriers in wealth accumulation, credit and
loan discrimination, and lack of intergenerational transfers (Herring & Loren, 2016). Compared to white
owners, black owners likely enter into businesses with less, if any, financial and institutional support and
mentorship networks in their respective locations.
8
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 87 of 801
4. Racial Disparities in the Source and Amount of Start-Up Capital Used
Depending on the type and amount of capital needed by potential business owners, different
sources of funding may be optimal for supporting them at the start-up stage of ownership. Independent of
scale and sector, an aspiring entrepreneur and even an already established business entrepreneur require
some degree of monetary support. Along with the figures presented in Table 4.1, evidence suggests that
even after controlling for education levels, blacks are less likely than whites to have their business loans
approved “and the loans they do receive from financial institutions are much smaller than those flowing to
white business borrowers” (Bates, 1998). Given these trends, some would argue that black Americans are
not given the opportunity to realize their full potential as entrepreneurs.
Table 4.1. The Amount of Start-Up Capital Used (2016)
Amount of Capital to Start or
Percent of Black-Owned Percent of White-Owned
Acquire the Employer Firm
Employer Businesses Employer Businesses
(2016)
$3,000,000 or more 0.2 0.5
$1,000,000 to $2,999,999 0.7 1.5
$250,000 to $999,999 4.5 6.5
$100,000 to $249,999 7.6 9.6
$50,000 to $99,999 10.8 9.4
$25,000 to $49,999 9.6 9.0
$10,000 to $24,999 15.5 11.8
$5,000 to $9,999 10.8 8.8
Less than $5,000 19.0 16.3
Don’t know 12.8 17.2
Not applicable 8.4 9.3
Total reporting (number) 62,670 2,966,595
9
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 88 of 801
As displayed in Table 4.1, 45.3% of black-owned employer businesses compared with 36.9% of
white-owned employer businesses used less than $25,000 of capital to start-up. In contrast, only 13% of
black-owned compared with 18.1% of white-owned businesses used more than $100,000 in capital to start
up. Clearly, black-owned businesses are starting off at a relatively smaller scale.
Like the amount of start-up capital, the source start-up capital has implications for what kind of
support the business owners have. The source can be in the form of a personal account, a loan from a
financial institution, an investment by a venture capitalist, or a government grant. As Table 4.2 shows,
21.8 percent of white-owned employer businesses receive business loans from banks, financial
institutions, or families and friends; only 15.5 percent of similar black-owned businesses received start-up
capital from equivalent sources. Furthermore, black-owned employer business are 59 percent more likely
to use start-up capital in the form of personal credit cards than are white-owned employer businesses.
Table 4.2. The Source of Capital Used to Start Business (2016)
Sources of Capital Used to Start or Percent of Black-Owned Percent of White-Owned
Acquire the Business Employer Firms (%) Employer Firms (%)
Personal/family savings of owner(s) 69.6 64.8
Personal/family assets other than savings
9.0 8.7
of owner(s)
Personal/family home equity loan 5.2 6.3
Personal credit card(s) carrying balances 14.7 9.2
Business credit card(s) carrying balances 6.3 5.0
Business loan from federal, state, or local
0.5 0.4
government
Government-guaranteed business loan
1.9 1.8
from a bank or financial institution
Business loan from a bank or financial
12.6 17.2
institution
Business loan/investment from
2.9 4.6
family/friends
Investment by venture capitalist(s) 0.5 0.5
Grants 0.5 0.2
Other source(s) of capital 4.4 2.9
Don’t know 8.1 10.0
None needed 8.4 9.3
Total Reporting (number) 62,657 2,966,829
All told, black American business owners have a limited access to capital relative to their white
counterparts. In fact, financial banks are rarely a source of credit for black-owned businesses and when
they are approved for credit, they often receive less than what they applied for. 20% of black households
are un-banked, while only 4% of white non-Hispanic households remain unbanked (AEO, 2017). Historic
10
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 89 of 801
racial bias and discrimination play roles at nearly every step of the process of obtaining a loan and this
leads to a somewhat dangerous trend and interplay between the amount and type of financial capital black
Americans typically seek to the financial capital they get. Black Americans are often forced to rely more
heavily on personal credit cards for financing a business. Consequently, this can deplete their assets even
further by damaging their credit scores further and suppressing their business prospects.
5. Racial Disparities in the Age of Employer Businesses
How long ago were existing employer and nonemployer businesses established? The duration or
“age” of a business is the difference between the year when the ASE survey was conducted—2012 for
nonemployer businesses; 2016 for employer businesses—and the year the business was originally
established. This corresponding duration of an employer-business has implications for business
stabilization, quantity of sales, and whether the business can withstand changing market conditions over
time. Older firms, by virtue of their longevity, tend to be more stable and more experienced. Moreover,
they typically generate more annual revenue from sales and have more market-navigation resources.
Thus, one would expect that older firms are not only larger in size but are also in stronger and more
successful positions than firms that were established more recently.
11
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 90 of 801
Figure 5.1. Age of Employer Business, by Race (2016)5
Based on the 2012 and 2016 ASE, white-owned employer businesses are older than black-owned
employer businesses. As displayed in Figure 5.1, 43.6 percent of all existing white-owned firms are 16
years or older (established prior to the 200s), while only 27.1 percent of black-owned firms are 16 years
or older6. In contrast, 35.2 percent of black-owned firms are 7 years or younger, whereas 21.6 percent of
white-owned firms are 7 years or younger. Therefore, close to half of all white-owned firms that exist
today were established before the turn of the twenty-first century, whereas most existing black-owned
firms were originally started in the 2000s.
For nonemployer businesses existing as of 2012, we observe similar patterns. While the most
common age of nonemployer businesses in 2012 was between 5 and 12 years old regardless of race,
white-owned nonemployer businesses were generally older than black-owned nonemployer businesses. In
fact, 62.1 percent of existing black-owned nonemployer firms in 2012 were originally established after
2000 while less than one third of all white-owned nonemployer businesses were originally established
between 2008 and 2012. Moreover, the proportion of white-owned nonemployer businesses that were
older than the age of 32 as of 2012 was roughly four times the proportion of black-owned nonemployer
businesses older than 32 years old. Over one-fourth of black-owned nonemployer businesses existing in
2012 were only one year old or younger.
5
The age/duration of a business is calculated from “year business was started”. For example, “Older than 35” refers
to businesses that were started before 1980.
6
13.1 percent and 10.4 percent of white-owned and black-owned businesses, respectively, do not know the year
their businesses were originally established.
12
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 91 of 801
Figure 5-2. Age of Nonemployer Businesses, by Race (2012)7
Given that more black-owned firms are younger, black business owners of those firms are likely less
experienced and lack the capacity to offer mentorship to other black American entrepreneurs who are just
starting off. One can imagine that this can create a serious disadvantage for one who is looking to receive
guidance on how to overcome historically-embedded barriers and how to navigate discriminatory
environments as they pertain to the business ownership experience for black Americans.
6. Racial Disparities in Characteristics of Employer Businesses, by Age
For white-owned businesses, older firms tend to generate more revenue in sales than younger
firms. This, however, does not hold true for black-owned businesses. As shown in Table 6.1, 50 percent
of sales from black-owned employer businesses comes those that are 9 to 26 years old. 15 percent of sales
from black-owned employer businesses come from those 27 years or older, while the proportion of sales
coming from for white-owned employer businesses that are 27 years or older is 51 percent. In other
words, a majority of the total revenue from black-owned firms are earned by firms that are relatively
younger than white-owned firms (ASE, 2016).
7
Note that data on nonemployer firms ended in 2012.
13
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 92 of 801
Table 6-1. Wage and Sales of Black-owned and White-owned Employer businesses, by Age (2016)
Black-Owned Employer Businesses White-Owned Employer Businesses
Duration Average Average
of Percent of salary per Percent Average Percent of salary per Percent Average
Business businesses employee of Sales Sales businesses employee of Sales Sales
as of 2016 (%) ($1000) (%) ($1,000) (%) ($1000) (%) ($1,000)
Older
0.04 31 0.05 1242 0.12 47 0.35 6699
than 36
27 to 35 0.07 32 0.10 1355 0.12 45 0.16 3025
17 to 26 0.14 33 0.20 1343 0.17 43 0.16 2215
9 to 16 0.26 30 0.30 1115 0.20 41 0.13 1466
4 to 8 0.22 29 0.17 755 0.14 36 0.07 1105
3 0.05 36 0.03 556 0.03 35 0.01 905
2 0.06 21 0.02 340 0.04 31 0.01 750
1 0.05 21 0.03 487 0.03 31 0.01 728
Don't
0.11 31 0.11 968 0.14 37 0.11 1828
Know
All told, the association between sales for older firms and larger firms does not hold across race.
Older black-owned firms are on average smaller than white-owned firms that are younger. For example,
black-owned businesses that are older than 36 years old generate less in average revenue sales
($1,242,000) as do white-owned businesses ($1,466,000) that are between 9 and 16 years old. At every
age category, the average salary per employee—measured by dividing annual payroll by the number of
employees in each age category—is greater for employees of white-owned employer firms than for
employees of black-owned employer firms8. In fact, the average employee from a black-owned employer
business that is older than 36 years earns ($31,000) just as much as the average employee of a white-
owned employer business that is only two years old. Most strikingly, across every age category, black-
owned employer businesses bring in less average revenue than white-owned firms originally established
around the same time. Much of this disparity in the performance of different aged black- and white-
owned businesses can be attributed to aspects related to the initial stage of ownership; particularly related
to racial differentials in prior wealth and access to capital. many differences in economic outcomes across
individuals operate through lingering effects of racially-motivated historical injustices of violence and
hostility.
8
This holds with the exception of employer firms whose duration is only three years as of 2015.
14
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 93 of 801
7. Wealth as an Input for the Pathway into Business Ownership
Empirically, black Americans are more likely than white Americans to start a business, but white-
owned businesses are more likely to grow and succeed. Similarly, white American adults have 13 times
the wealth than black American adults, and on average “black business owners have fewer assets, lower
wealth, and less disposable income to invest into the business than white business owners” (AEO, 2017).
The prevalence of racial wealth inequality in the U.S. has long been examined and well documented by
past scholarship. Wealth is arguably the foremost indicator of economic well-being, signifying what
people own and whether a household has access to life-enhancing and income-producing assets. Herring
and Hynes (2017) state that “racial disparities in wealth provide material goods, opportunities, resources,
services, and psychological satisfactions for Whites, and they serve to penalize African Americans by
decreasing their quality of life.” Moreover, wealth has a tendency to aggregate in the hands of very few
and accumulate intergenerationally over time (Chiteji and Hamilton, 2002; Addo, Houle, & Simon, 2016;
Brown, 2016).
This suggests that prior wealth endowments (and the various economic, social, and political
benefits that come with them) may influence the capability of an individual to become a business owner.
Lack of wealth and access to capital serves as a current barrier that governs the pathway into business
ownership for black Americans. In order to support an ecosystem for black business owners, it is critical
that more attention is given to wealth as an input into the business ownership experience.
15
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 94 of 801
8. Policy Implications and Conclusion
Virtually all commentaries on the relationship between entrepreneurship and wealth emphasize
the contribution of business ownership to wealth accumulation; very little consideration is given to the
role of prior wealth in shaping pathways of business ownership. Our study exploring racial differences in
patterns of entry into entrepreneurship points us toward the latter, less recognized relationship—the
influence of initial asset endowments on the capability of individuals to become business owners.
Racial differences in prior wealth are implicated in racial differences in the ways in which people
acquire businesses. Adding the dimension of multiple pathways toward business ownership enrichens the
entry story. For both employer and nonemployer businesses, whites are far more likely than blacks to
become owners via routes that require direct (or network) access to wealth.
A body of literature establishes how racial disparities are not merely a function of the political,
social, and economic processes as we know them today (Archarya, Blackwell, & Sen, 2018). Rather,
many differences in economic outcomes across individuals operate through lingering effects of racially-
motivated historical injustices and hostilities. While the construction of freeways to cut through Hayti and
Black Wall Street in Durham, North Carolina, reveals one of the historical methods used to disrupt and
block black business ownership, the 1921 white massacre in Tulsa, Oklahoma, exemplifies the explicit
acts of violence mounted against the economic prosperity of black Americans.
These injustices were legitimated through “official narratives that discredit(ed), vilifi(ed),” and
targeted black communities and infrastructures that were prosperous or thriving (Messer et al., 2018).
Social and economic structures of historical racial violence not only have laid the foundation for
maintenance of white supremacy in the past, but arguably also have had lasting effects on the business
opportunities, economic prospects, and corresponding potential support networks at present. Much of the
black-white business ownership disparity observed today may be linked to how conducive or damaging
social structures have been in supporting black wealth and opportunity. Future research should explore
how localities with different histories of racial hostility make black business ownership more or less
likely today.
This study aimed to contextualize patterns of initial ownership and to shed some light on the
mechanisms and processes by which racial stratification in business ownership persists. Our results
suggest that tackling the racial business ownership disparity will require tackling the racial wealth
disparity prior to increasing business ownership among blacks. It points us toward public targeting of
seed capital for black entrepreneurs or major redistributive measures to build black wealth in advance of
accumulation that takes place due to business ownership. This could mean, in turn, adoption of general
policies aimed at closing the racial wealth gap, which might include “baby bonds” (young adult trust
16
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 95 of 801
accounts) and/or, more direct and substantial, a program of reparations for black Americans. In sum,
policies focused on pathways into business ownership should consider how critical prior wealth can be for
the fostering a more equitable ecosystem for business.
17
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 96 of 801
9. References
Addo, F. R., et al. (2016). Young, Black, and (still) in the red: Parental wealth, race, and student loan
debt. Race and Social Problems, 8, 64–76.
Annual Survey of Entrepreneurs (2016). United States Census Bureau.
Archarya, A., Blackwell, M., & Sen, M. (2018). Deep Roots How Slavery Still Shapes Southern Politics.
Princeton University Press.
Association for Enterprise Opportunity (2017). The Tapestry of Black Business Ownership in America:
Untapped Opportunities for Success, The Association for Enterprise Opportunity (AEO).
Bates, T. (1998), Race, Self-Employment and Upward Mobility: An Illusive American Dream.
Washington, DC; The Woodrow Wilson Center Press and Baltimore: The Johns Hopkins
University Press.
Brown, T. H. (2016). Diverging fortunes: Racial/ethnic inequality in wealth trajectories in middle and late
life. Race and Social Problems, 8, 29–41.
Chiteji, N, and D. Hamilton (2002). “Family Connections and the Black-White Wealth Gap among the
Middle Class.” Review of Black Political Economy, 30(1): 9-27.
Darity, W. et al. (2017). The Tapestry of Black Business Ownership in America: Untapped Opportunities
for Success. Association for Enterprise Opportunity (AEO).
Fairlie, Robert W. “The Absence of the African-American Owned Business: An Analysis of the
Dynamics of Self-Employment.” Journal of Labor Economics, 17(1), January 1999: 80-108.
Fairlie, Robert. (2012). Immigrant Entrepreneurs and Small Business Owners, and Their Access to
Financial Capital. Topics in Entrepreneurship: Select Research.
Herring, C., & Loren, H. (2016). Wealth inequality in Black and White: Cultural and structural sources of
the racial wealth gap. Race and Social Problems, 8, 4–17.
Hurst, E., & Lusardi, A. (2004). Liquidity Constraints, Household Wealth, and Entrepreneurship. Journal
of Political Economy, 112(2), 319-347. doi:10.1086/381478
Köllinger, P. and Minniti, M. Not Lack of Trying: American Entrepreneurship in Black and White.
Small Business Economics (2006) 27: 59-79. https://doi.org/10.1007/s11187-006-0019-6
McKoy, H. C. and Johnson J. H. (2018). Do Business Ecosystems See Color? International Journal
of Social Ecology and Sustainable Development.
Meschede, T., et al. (2016a). Inequality in the “cradle of liberty”: Race/ethnicity and wealth in greater
Boston. Race and Social Problems, 8, 18–28.
Meschede, T., et al. (2016b). Wealth mobility of families raising children in the twenty-first century. Race
and Social Problems, 8, 77–92.
Messer, C. M., Shriver, T. E., & Beamon, K. K. (2018). Official Frames and the Tulsa Race Riot of 1921:
The Struggle for Reparations. Sociology of Race and Ethnicity, 4(3), 386–399.
Small Business Credit Survey (2017). Report on Employer Firms. Federal Reserve Banks.
Small Business Credit Survey (2018). Report on Nonemployer Firms. Federal Reserve Banks.
Survey of Business Owners (2012). United States Census Bureau.
Survey of Business Owners (2015). United States Census Bureau.
18
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 97 of 801
10. Appendix
Appendix A. White American Share of Population and Business Ownership (2012)
White American
State
Population Share Share of All Businesses
Alabama 0.70 0.76
Alaska 0.68 0.82
Arizona 0.84 0.89
Arkansas 0.80 0.87
California 0.74 0.74
Colorado 0.88 0.93
Connecticut 0.82 0.89
Delaware 0.72 0.83
Florida 0.78 0.83
Georgia 0.63 0.64
Hawaii 0.33 0.4
Idaho 0.94 0.97
Illinois 0.78 0.79
Indiana 0.87 0.9
Iowa 0.93 0.96
Kansas 0.87 0.93
Kentucky 0.89 0.93
Louisiana 0.64 0.72
Maine 0.95 0.97
Maryland 0.61 0.67
Massachusetts 0.84 0.9
Michigan 0.80 0.83
Minnesota 0.86 0.92
Mississippi 0.60 0.69
Missouri 0.84 0.89
Montana 0.90 0.96
Nebraska 0.90 0.94
Nevada 0.77 0.81
New Hampshire 0.94 0.97
New Jersey 0.74 0.8
New Mexico 0.83 0.89
New York 0.71 0.74
North Carolina 0.72 0.81
North Dakota 0.90 0.96
Ohio 0.83 0.88
Oklahoma 0.76 0.84
Oregon 0.88 0.92
Pennsylvania 0.83 0.89
Rhode Island 0.86 0.92
South Carolina 0.69 0.79
South Dakota 0.86 0.96
Tennessee 0.79 0.83
Texas 0.81 0.82
Utah 0.92 0.95
Vermont 0.95 0.98
Virginia 0.71 0.77
Washington 0.82 0.87
West Virginia 0.94 0.96
Wisconsin 0.88 0.92
Wyoming 0.93 0.97
Source: Survey of Business Owners (2012)
19
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 98 of 801
Appendix B: White-Owned vs. Black-Owned Sales and Salaries of Employer Businesses (2012)
White-Owned Businesses Black-Owned Businesses
State
Average Average Annual Average Average Annual
Annual Sales in Employee Salary in USD Annual Sales in Employee Salary In
USD USD USD
Alabama 2,510,078 34,310 42,288 1,025,705
Alaska 2,275,223 46,222 118,656 834,808
Arizona 1,849,125 33,335 68,080 879,470
Arkansas 2,372,633 30,339 39,255 591,900
California 2,376,776 45,483 84,177 1,128,511
Colorado 1,598,580 38,143 97,087 1,010,609
Connecticut 2,492,964 43,938 47,552 653,300
Delaware 2,100,689 38,724 44,151 380,435
District of Columbia 2,744,202 56,913 113,922 1,437,584
Florida 1,532,142 35,001 44,645 718,506
Georgia 2,338,896 36,320 37,324 646,629
Hawaii 1,951,869 34,466 121,644 797,335
Idaho 1,526,704 29,618 108,884 912,906
Illinois 2,353,729 41,505 51,596 1,246,749
Indiana 2,485,203 34,131 111,085 2,483,369
Iowa 2,362,056 33,500 100,293 1,703,721
Kansas 2,372,540 35,247 79,667 931,713
Kentucky 2,240,284 31,976 82,870 1,109,809
Louisiana 2,581,325 38,153 35,592 657,779
Maine 1,588,498 32,557 119,729 N/A
Maryland 2,328,619 42,507 68,671 1,039,304
Massachusetts 2,448,372 45,168 83,172 1,049,822
Michigan 2,198,558 35,844 48,254 1,349,167
Minnesota 2,342,317 37,291 80,484 1,066,135
Mississippi 2,280,364 31,309 31,933 617,004
Missouri 2,194,089 35,434 66,986 564,842
Montana 1,244,546 30,194 81,260 597,034
Nebraska 2,460,743 35,189 65,000 N/A
Nevada 2,419,465 36,932 55,290 1,058,853
New Hampshire 1,855,538 37,439 94,049 784,671
New Jersey 2,440,137 41,505 65,587 843,437
New Mexico 1,653,151 32,220 121,866 843,824
New York 2,123,946 46,132 48,406 744,787
North Carolina 2,037,372 33,690 53,674 666,169
North Dakota 2,849,648 38,689 82,714 1,155,440
Ohio 2,551,218 34,908 112,621 2,300,496
Oklahoma 2,313,248 35,968 54,115 635,379
Oregon 1,746,953 36,801 107,718 1,056,919
Pennsylvania 2,313,371 37,138 54,333 724,157
Rhode Island 1,758,156 37,533 55,747 696,400
South Carolina 1,929,716 31,788 54,134 913,623
South Dakota 2,217,621 31,754 76,941 1,093,000
Tennessee 2,585,799 33,768 40,274 780,106
Texas 549,280 37,576 54,795 839,817
Utah 467,462 34,556 209,924 1,938,787
Vermont 354,558 34,640 26,570 191,000
Virginia 490,426 40,337 98,233 1,117,020
Washington 552,652 35,034 86,838 743,617
West Virginia 463,807 30,106 76,079 746,247
Wisconsin 615,201 34,514 76,270 1,086,545
20
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 99 of 801
Wyoming 514,931 38,162 86,006 N/A
Source: Survey of Business Owners (2012)
21
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 100 of 801
Exhibit D
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 101
on Black-owned of 801
businesses | Brookings
COMMENTARY
Who is driving Black business growth? Insights from the latest
data on Black-owned businesses
Andre M. Perry, Manann Donoghoe, Hannah Stephens
May 24, 2023
Follow the authors
@andreperryedu
@ManannanAD
EXECUTIVE SUMMARY
Using the Census Bureau’s Annual Business Survey (ABS) data on employer firms
(those that have at least one employee), this third installment of Brookings Metro’s
series of reports on Black-owned businesses focuses on the overall state of those
firms across U.S. metropolitan areas. These reports are in support of the Path to 15|55
initiative (https://www.pathto1555.org/) , which endeavors to transition 15% of Black-
owned sole proprietorships into employer firms, adding $55 billion to the U.S.
economy. The ABS data, collected in 2020, allows for an in-depth analysis of employer
firms during the initial stages of the pandemic. In addition, this report places a
spotlight on employer firms owned by Black women. We find that:
In 2020, Black people represented 14.2% of all Americans but only 2.4% of all
employer-firm owners. Latino or Hispanic people represented 18.7% of the
population and 6.5% of employer-firm owners, while Asian Americans represented
6% of the population and 10.6% of employer-firm owners. 1 From 2019 to 2020,
Black-owned businesses grew by 6,351 firms, or 4.72%—behind Latino or Hispanic
(8.19%) and Asian American (5.33%) firms, but above white-owned businesses, which
shrunk by 0.9%.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 1/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 102
on Black-owned of 801
businesses | Brookings
Black business growth was trending upward before the COVID-19 pandemic. From
2017 to 2020, the number of Black-owned businesses across the country increased
by 13.64%—larger than all businesses in general, which increased by 0.53% over
the same period. Black-owned firms brought in an estimated $141.1 billion in gross
revenue in 2020—an 11% increase since 2017.
More so than other racial groups, Black-owned businesses had pronounced increases
in revenue, employees, and payroll. In 2020, Black business owners employed 1.321
million people, and created 48,549 new jobs—adding an additional $1.7 billion in
aggregate payroll to the U.S. economy.
If Black business ownership continues to grow at its current rate (4.72%), it will take
256 years to reach parity with the share of Black people in America—a timeline that
leaves racial wealth gaps entrenched. Reaching parity in 15 years would take a 74.4%
growth rate; for 25 years, a 44.6% growth rate; and for 50 years, a 22.3% growth rate.
The 10 metro areas with the highest proportional representation of Black-owned
businesses are all in the Southeast, including three in North Carolina. However, in
2020, as in previous survey years, no metro area had a share of Black employer firms
equal to or greater than its share of Black residents.
Despite the impacts of COVID-19, from 2017 to 2020, the number of employer
businesses owned by Black women rose to 52,374—encompassing 37.2% of all
Black-owned businesses and representing an increase of 1.41 percentage points.
However, though Black women accounted for about 7% of the total U.S. population
and 13.9% of American women in 2020, they only owned 0.91% of all businesses and
4.23% of women-owned businesses.
A comparatively substantial portion (37.9%) of all Black women business owners
have an advanced degree, including 19.1% with master’s degrees. This is more than
10 percentage points higher than the national average of 23.5%.
Though far from equitable, the rate of business ownership for Black women is
growing rapidly. Black-women-owned employer businesses increased by 18.14%
between 2017 and 2020—outpacing women-owned businesses (9.06%) and Black-
owned businesses (13.64%). These trends parallel the growth in small sole-proprietor
businesses owned by Black and Latino or Hispanic women. However, the extent to
which this growth rate is the result of structural changes in the business and lending
sectors that have made room for more inclusive growth is questionable. A closer
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 2/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 103
on Black-owned of 801
businesses | Brookings
evaluation of these trends could help develop more intentional strategies to uplift
these businesses and promote further growth.
INTRODUCTION
The COVID-19 pandemic and protests stemming from the murder of George Floyd put
a spotlight on Black wealth, health, and overall well-being. The social crises illuminated
connections between attitudes that lead to a disproportionate number of stops,
arrests, and convictions among Black Americans, as well as the underinvestment in
Black businesses. Amid nationwide shutdowns, Black businesses felt significant pain,
as they were less likely to obtain federal COVID-19 relief loans
(https://www.nber.org/digest/202112/racial-disparities-paycheck-protection-program-
lending) .
After Floyd’s death, America’s 50 largest public companies collectively pledged $49.5
billion to address racial inequality. (According to a Washington Post analysis
(https://www.washingtonpost.com/business/interactive/2021/george-floyd-corporate-
america-racial-justice/) , 90% of those commitments were loans or investments the
corporations could profit from.) While many of those dollars focused on saving and
developing Black-owned businesses, broader federal relief funds seemed to enhance
Black business ownership. The National Bureau of Economic Research
(https://www.nber.org/papers/w28787) found correlations between a surge of new
businesses in Black neighborhoods from 2019 to 2020 and the first two rounds of
pandemic relief checks.
Similarly, federal and state bodies are investing in businesses owned by people of
color. The Biden administration has signed executive orders to strengthen racial equity
and support underserved communities (https://www.whitehouse.gov/briefing-
room/statements-releases/2023/02/16/fact-sheet-president-biden-signs-executive-
order-to-strengthen-racial-equity-and-support-for-underserved-communities-across-
the-federal-government/) , with a focus on economic justice. This has included
initiatives to ensure 50% or more of federal contracting dollars are awarded to small,
disadvantaged businesses by 2025, as well as the establishment of racial equity
committees to address disparities in financial inclusion and access to capital, among
other areas.
Intentional and passive investments in Black businesses during the pandemic might
forge a pathway for future development efforts—or it may prove to be an anomaly that
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 3/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 104
on Black-owned of 801
businesses | Brookings
cannot be replicated. There are strong signs that racialized gaps are closing; for
example, unemployment disparities between Black and white Americans have
lessened since 2010, while the aggregate unemployment rate for Black Americans
(5%) (https://www.bls.gov/charts/employment-situation/civilian-unemployment-
rate.htm) is now as low as its pre-pandemic rate, and down from 9.9% in April 2021.
Still, Black unemployment is higher than the national rate (3.5% as of March 2023).
Given the extent of racial wealth and prosperity disparities, the rate of progress is
slow. While racial wealth gaps have narrowed slightly since 2013
(https://www.federalreserve.gov/econres/notes/feds-notes/recent-trends-in-wealth-
holding-by-race-and-ethnicity-evidence-from-the-survey-of-consumer-finances-
20170927.html) , income inequality has ballooned, with increasing gaps between the
poorest and wealthiest households. For example, the highest-income white
households have seen the largest growth in wealth
(https://www.pewresearch.org/fact-tank/2017/11/01/how-wealth-inequality-has-
changed-in-the-u-s-since-the-great-recession-by-race-ethnicity-and-income/) —75
times that of lower-income families and 195 times the wealth of the median low-
income Black family. These wealth inequalities have implications for rates of racial and
ethnic representation in business ownership, and threaten to undermine progress on
Black-owned businesses.
This report focuses on the state of businesses owned and run by Black people
between 2017 and 2020. One key finding pertains to Black women, who are one of the
nation’s most underserved groups (https://www.brookings.edu/essay/to-protect-
black-women-and-save-america-from-itself-elect-black-women/) , with the lowest
proportional democratic representation (https://www.pewresearch.org/fact-
tank/2023/02/16/22-states-have-ever-elected-a-black-woman-to-
congress/#:~:text=Two%20Black%20women%20have%20served,are%20both%20Black%20and%20
and the largest median income gap for full-time workers
(https://www.nationalpartnership.org/our-work/resources/economic-justice/fair-
pay/african-american-women-wage-gap.pdf) . However, amid a rise in Black-owned
businesses, increases in Black women founders seemed to drive overall growth. This
report provides an analysis of this rise in business ownership among Black women and
what it means for equity in business ownership in America.
RECENT TRENDS
Recent trends show a rise in Black-owned businesses
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 4/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 105
on Black-owned of 801
businesses | Brookings
In 2020, Black people represented 14.2% of all Americans but accounted for 2.4% of all
employer-firm owners. Table 1 shows this is in comparison to Latino or Hispanic
Americans (18.7% of the population and 6.5% of employer-firm owners), Asian
Americans (6% of the population, 10.6% of employer-firm owners), and white
Americans (61.6% of the population, 82.67% of employer-firm owners).
(https://www.brookings.edu/wp-content/uploads/2023/05/Table.png)
The underrepresentation of Black and brown business ownership represents a
substantial obstacle for progress toward developing a racially inclusive economy in the
U.S. Black businesses are important—not only to restore the value that racism has
extracted from Black communities and households, but also for the strength of our
national and local economies. A lack of Black-owned employer businesses constrains
economic growth (https://www.brookings.edu/essay/to-expand-the-economy-invest-
in-black-businesses/) and stymies opportunities for investment and inclusive
development, especially in our most diverse cities and regions.
Between 2010 and 2020, the share of the minority population
(https://www.census.gov/library/stories/2021/08/improved-race-ethnicity-measures-
reveal-united-states-population-much-more-multiracial.html) (those not identifying as
white alone) has grown from 36.3% to 42.4%, with the number of Americans
identifying as two or more races more than tripling
(https://www.brookings.edu/research/why-the-federal-government-needs-to-change-
how-it-collects-data-on-native-americans/) . Yet the rate of minority-owned
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 5/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 106
on Black-owned of 801
businesses | Brookings
businesses has grown at a much slower pace, meaning there is a widening gap
between population share and business representation. This gap has real implications
for the state of the U.S. economy and the extent to which all people can benefit from
it. However, there are signs of progress.
Between 2017 and 2020, the number of Black-owned businesses increased by 13.64%
—larger than the increase in all businesses combined, which was 0.53% over the same
period. The average yearly growth rate in the number of Black-owned businesses
during this time was 4.2%, which is more than 20 times the overall business growth
rate of 0.18%. Black-owned business growth was greatest between 2018 and 2019,
when it rose by 8%; it declined to 4.7% between 2019 and 2020, likely representing the
pandemic’s early impacts on foreclosure rates and investment decisions. Still, Black-
owned firms brought in an estimated $141.1 billion in gross revenue in 2020—an 11%
increase since 2017.
If Black business ownership continues to grow at a rate of 4.2%, it will take 256 years
to reach parity with the percentage of Black people in America—a timeline that leaves
racial wealth gaps entrenched. Reaching parity in 15 years would take a 74.4% growth
rate; for 25 years, a 44.6% growth rate; and for 50 years, a 22.3% growth rate.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 6/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 107
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure1-2.png)
Black-owned businesses were resilient to the pandemic’s impact on revenue and
closure rates. Figure 2 shows that from 2019 to 2020, the number of Black owned
businesses is estimated to have grown by 6,351, or 4.7%—well above white-owned
employer firms (which shrunk by 0.9%), but behind Latino or Hispanic-owned
employer firms (which grew by 8.2%) and Asian American-owned employer firms
(5.3%). Over the same period, Black-owned businesses increased their proportional
share of employees and revenue by a greater extent: Between 2019 and 2020, Black
business owners employed more than 1.32 million people, adding 48,549 new jobs and
an additional $1.7 billion in aggregate payroll to the U.S. economy.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 7/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 108
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure2-1.png)
GEOGRAPHIC PATTERNS (AND INTERACTIVE)
Black business ownership continued to grow in the
South, with weaker trends in other metro areas
This report updates the 2022 report’s metro-level data visualization, estimating the
impacts to the U.S. economy if the proportion of Black-owned employer business in a
metro area matched the Black population share. The ABS cannot accurately capture
information on race and ethnicity in each survey year, so to provide a more complete
picture of metro areas, this interactive includes data from across all survey years as
well as the most recent 2020 data.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 8/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 109
on Black-owned of 801
businesses | Brookings
What if Black businesses reached parity with all businesses?
Select a metro area
Huntsville, AL Metro Area
…
…
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key f… 9/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 110
on Black-owned of 801
businesses | Brookings
There are 295 Black businesses, accounting for 3.7% of employer businesses. If Black businesses accounted for 23.6% of
employer firms (equivalent to the Black population), there would be 2,166 more Black businesses.
Black businesses create an average of 7 jobs per firm, compared to 21 for all businesses. If the average employees per Black
business reached parity, it would create approximately 4,272 new jobs.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 10/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 111
on Black-owned of 801
businesses | Brookings
Black businesses pay their employees an average of $57,721, compared to $61,717 for all businesses. If Black businesses
paid this much, then those employees would see an increase in total pay by approximately $8,115,633.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 11/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 112
on Black-owned of 801
businesses | Brookings
If the number of Black businesses matched the population size and the employees per firm matched the average business, it
would create 50,552 jobs.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 12/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 113
on Black-owned of 801
businesses | Brookings
Table 2 shows that the top 10 metro areas with the highest proportional
representation of Black-owned firms are all in the Southeast, including three in North
Carolina alone. Atlanta had the highest representation in 2020 at 7.42%, followed by
Washington, D.C., Virginia Beach, Va., Columbus, Ga., Memphis, Tenn., and Richmond,
Va. As noted above, the most recent data excludes some key metro areas; looking
across all data from 2017 to 2020, Fayetteville, N.C. continues to have the highest
representation of Black business ownership of all metro areas, at over 11%.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 13/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 114
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table2-2.png)
Table 3 shows that larger cities—including New York, Washington, D.C., Chicago, and
Los Angeles—had higher absolute numbers of Black-owned businesses. Several of
these, including New York, Chicago, and Dallas, are the nation’s largest Black and
Latino-majority cities (https://www.brookings.edu/research/recognizing-black-and-
latino-majority-cities-is-the-first-step-to-finding-a-real-world-wakanda/) , with nearly
equal proportions of Black and Latino or Hispanic residents.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 14/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 115
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table3-1.png) Table 4
lists the top 10 metro areas by the ratio of Black businesses to Black residents. It
shows that most of the areas that are closest to achieving parity (a Black business
ownership rate equal to the share of Black residents in the metro area) tend to have
smaller numbers of Black-owned businesses. The top three—Portland, Maine; Fresno,
Calif.; and Santa Rosa, Calif.—all have fewer than 300 Black-owned employer firms
across their entire metro areas. Metro areas with larger Black populations—such as
Washington, D.C. and St. Louis—are outliers, having high numbers of both Black
residents and Black-owned employer businesses.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 15/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 116
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table4.png)
Despite substantial growth in the number of Black-owned businesses, their share is
still far behind parity with the percentage of Black residents across all metro areas.
While philanthropic and policy attention is increasingly focused on racial equity, this
has not always translated to more equitable rates of business ownership. For example,
innovation economies in Pennsylvania
(https://www.brookings.edu/research/commonwealth-of-innovation-a-policy-agenda-
for-revitalizing-pennsylvanias-economic-dynamism/) and Pittsburgh
(https://www.brookings.edu/research/centering-racial-equity-and-inclusion-in-
pittsburghs-innovation-economy/) have struggled to make diversity and inclusion a
core component of investments in the business environment.
In 2020, as in previous survey years, no metro area had proportional representation in
Black business ownership. Figure 3 shows that areas with smaller Black populations
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 16/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 117
on Black-owned of 801
businesses | Brookings
tended to have better proportional representation; for example, Portland, Maine has a
2.85% Black population share and a 1.09% Black business ownership rate.
As the number of Black residents in a metro area increases, the business ownership
gap also tends to increase. Figure 3 orders all metro areas surveyed in the ABS from
2017 to 2020 by the percentage share of the Black population, beginning with Albany,
Ga. at 54% and ending with Boulder, Colo. at 1.8%. The graph demonstrates that metro
areas with the highest shares of Black residents also tend to be the most unequal in
terms of business ownership representation. In other words, despite a better
opportunity for investment in Black communities, regions with a higher percentage of
Black residents do not see those opportunities realized as employer businesses.
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure3_v2-01.png)
Figure 4 shows this same data as a scatter plot. The orange, yellow, and red lines
indicate how far away a city is from parity. Fayetteville, N.C., and St. Louis, for
example, have high Black representation and some of the most equal Black business
representation rates of all metro areas. Most metro areas, including Memphis, Tenn.,
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 17/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 118
on Black-owned of 801
businesses | Brookings
and Birmingham, Ala., are at least four to eight times away from parity, while cities
such as Montgomery, Ala., and Hinesville, Ga. have rates over 12 times off parity.
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure4_v2-01.png)
Figure 5 shows that the average number of Black-owned businesses sees a drastic
decline for metro areas with Black population shares greater than 40%. This includes
cities such as Jackson, Miss., Montgomery, Ala., and Albany, Ga. Moreover, as the
share of Black residents increases, the average number of employees per business
decreases—from an average of 16 employees in metro areas with a Black population
share lower than 10%, to seven for metro areas with a Black population share greater
than 40%.
These figures provide another indication that Black-owned businesses are not
representative of their population size, especially in metro areas with the highest
shares of Black residents. Moreover, the economic impact of the average Black-owned
business tends to be smaller in metro areas with more Black residents. Taken together,
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 18/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 119
on Black-owned of 801
businesses | Brookings
these statistics highlight the continued underinvestment in Black-majority
communities and regions—paralleling similar disparities we have documented in the
absence of premium grocers (https://www.brookings.edu/research/what-the-lack-of-
premium-grocery-stores-says-about-disinvestment-in-black-neighborhoods/) and the
lower value of commercial real estate (https://www.brookings.edu/research/the-
devaluation-of-assets-in-black-neighborhoods-the-case-of-commercial-property/) in
Black neighborhoods, even controlling for differences in income and wealth. This has
implications for the capacity of locally owned Black businesses to generate inclusive
growth and employment opportunities in their communities. Black-owned businesses
tend to create more opportunities for people of color
(https://www.irp.wisc.edu/publications/dps/pdfs/dp123601.pdf) , and contribute to
building more equitable regional economies.
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure5.png)
Looking across all survey years, the largest trends toward parity in business ownership
occurred in Columbus, Ga., where the percentage of Black business ownership
increased 2.35 percentage points. Albany, Ga. has seen the greatest decline, at 2.42
percentage points. Metro areas such as Columbus, Ga. and Virginia Beach, Va. have
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 19/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 120
on Black-owned of 801
businesses | Brookings
wealthy and stable markets and are located close to military bases. And the growth in
areas such as Atlanta and Richmond, Va. align with other indicators of development in
these areas (https://www.brookings.edu/research/metro-monitor-2023-examining-
pandemic-era-patterns-of-inclusive-growth-across-the-u-s/) , which show high rates
of population growth and business development.
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure6.png)
Progress in Black business ownership is variable between metro areas, with booming
growth in upward trending areas and higher-than-average closure rates in declining
areas. Figure 7 shows the fastest-growing and declining metro areas for the total
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 20/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 121
on Black-owned of 801
businesses | Brookings
number of Black-owned businesses, and compares growth between Black-owned and
non-Black-owned businesses.
Most metro areas with high Black-owned business growth also have high total
business growth, except Reading, Penn., New Haven, Conn., and Trenton, N.J., where
Black business ownership has increased despite overall business declines. Similarly,
areas with the largest declines in Black-owned businesses also have declines across
all businesses. However, Black-owned businesses tend to be disproportionately
impacted in these places, with much larger decreases in the number of Black
businesses compared to the region’s average.
Map 1 shows an overall positive picture of growth for Black-owned businesses across
the U.S., with high-growth areas in the Southeast and Southwest, but some
pronounced declines in other areas, particularly the Midwest.
(https://www.brookings.edu/wp-content/uploads/2023/05/Map1.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 21/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 122
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure7_v2-01.png)
Figure 8 provides more detail on the variability Black-owned businesses experience in
different metro areas. In some areas, Black-owned businesses are responsible for
markedly high growth in overall business ownership. For example, in Washington, D.C.
and Charlotte, N.C., Black-owned businesses accounted for 41.5% and 31.9% of total
business growth, respectively. Conversely, in declining metro areas such as St. Louis
and Milwaukee, Black-owned businesses have been highly impacted, representing
43.2% and 55.9% of the loss in regional businesses, respectively.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 22/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 123
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure8.png)
PANDEMIC REBOUND
Black-owned businesses rebounded strongly from the
pandemic
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 23/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 124
on Black-owned of 801
businesses | Brookings
This report includes the first year of ABS data that captures the impacts of the COVID-
19 pandemic on employer businesses. Our previous report used data from the Small
Business Credit Survey
(https://www.newyorkfed.org/medialibrary/FedSmallBusiness/files/2021/sbcs-report-
on-firms-owned-by-people-of-color) (conducted between September and October
2020), which showed that the pandemic was disproportionately impacting
(https://www.nber.org/papers/w27462) Black-owned businesses. However, the 2021
ABS data on employer businesses—which captures the state of businesses over 2020
—reveals strong rebounds for businesses overall, and Black-owned businesses in
particular.
The pandemic initially hit new businesses hard, drastically reducing the rate of new
business applications from January to April 2020. But by July, application rates had
more than recovered
(https://www.census.gov/econ/bfs/visualizations/geovisualization.html) , reaching
numbers above those prior to the pandemic. From March 2020 to March 2021, new
business applications rose by 73.9%, driven by increases in the South and Midwest,
which saw jumps of 83.5% and 80.7%, respectively. As recently as February 2023, the
rates of business applications remain high, indicating a sustained recovery.
However, the pandemic affected existing Black businesses in distinctive ways,
dampening growth for some while buoying others. In addition to capturing the
pandemic’s overall impact on employer businesses, the most recent ABS data reports
subjective responses to the impact on business operations. Table 5 shows that a
slightly smaller proportion of Black-owned businesses (3.9%, versus 4.1% for all
businesses) were unable to maintain operations, while a higher proportion (36.2%,
versus 29.7% for all businesses) were able to maintain operations with minimal
impacts. However, a lower percentage of Black business owners reported they were
able to maintain operations at or above current levels—reflecting the fact that though
many Black employers were able to maintain operations, they may not have been able
to take full advantage of the subsequent boom in businesses or recovery funds.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 24/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 125
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table5_v2.png)
Black-owned firms were slightly more likely than all other businesses to report that
sales decreased due to the pandemic (37.7% compared to 35.56%). The number of
businesses impacted varied by firm size, with smaller and midsized employers less
likely to report a decrease in sales than larger employers. However, Black-owned
businesses were also more likely to report that sales increased due to the pandemic:
An average of 12% of Black-owned businesses reported an uptick in sales, compared
to only 5.5% of all businesses. These seemingly contradictory findings regarding sales
demonstrate a great deal of variability among Black-owned firms. Some were more
able to adapt and take advantage of the new economic environment while others were
hit hard.
Closure rates by industry provide a more nuanced insight into the types of Black-
owned businesses that were negatively impacted. Table 6 shows the percentage of
Black-owned businesses and overall businesses by industry in 2020. Unsurprisingly,
the hardest-hit industry for both groups was accommodations and food services,
which saw revenues plummet during lockdowns, especially for Black-owned
businesses (https://www.tandfonline.com/doi/epdf/10.1080/24694452.2022.2095971?
needAccess=true&role=button) . Black-owned businesses also saw slightly higher-
than-average closure rates in the transportation and warehousing sector as well as
the arts, entertainment, and recreation sector. These sectors are also large employers
of Black people
(http://dln.jaipuria.ac.in:8080/jspui/bitstream/123456789/1332/1/COVID-19s-effect-
on-minority-owned-small-businesses-in-the-United-States.pdf) .
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 25/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 126
on Black-owned of 801
businesses | Brookings
Black Americans had higher rates of job loss than other racial or ethnic groups
(https://edworkforce.house.gov/uploadedfiles/wilson_testimony.pdf) during the
pandemic. But in this regard, Black business owners’ decreasing “opportunity share”
(https://indicators.kauffman.org/wp-content/uploads/sites/2/2021/03/2020_Early-
Stage-Entrepreneurship-National-Report.pdf) in 2020 (an estimate of the relative
influence of opportunity over necessity in new business activity) shows that higher
unemployment seems to have been a push factor for the decision to start a business.
Sectors that have high proportions of Black-owned businesses—including health care
and professional, scientific, and technical services—were less impacted on average.
Non-store retailers grew substantially
(https://www.nber.org/system/files/working_papers/w28912/w28912.pdf) over the
first 40 weeks of the pandemic, and there is some evidence that COVID-19 relief
funding was associated with higher rates of startups in higher-income Black-majority
neighborhoods
(https://www.nber.org/system/files/working_papers/w28787/w28787.pdf) . Though a
shift toward virtual businesses may have lowered the barriers to entry for Black
owners, more work is needed to understand its effects on Black neighborhoods, where
physical assets can help drive investment and growth
(https://www.brookings.edu/research/the-devaluation-of-assets-in-black-
neighborhoods-the-case-of-commercial-property/) .
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 26/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 127
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table6.png)
An employer’s capacity to weather the pandemic partially depended on the ability of
employees to work from home. According to the Bureau of Labor Statistics
(https://www.bls.gov/ors/) , in May 2020, 29.3% of Black employees were teleworking,
compared to 35.4% overall. Teleworking rates were especially low for sectors in which
Black employees were over-represented, such as health care, where only 1.7% of
employees had the ability to telework.
Table 7 shows the proportion of Black-owned and non-Black-owned businesses that
reported having employees who worked from home and those that reported having
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 27/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 128
on Black-owned of 801
businesses | Brookings
employees who never worked from home. Results varied by the age of the business.
Newer Black-owned businesses seemed to have more flexibility for employees to work
from home, especially those between four to five years of operation, while businesses
over 10 years old were less flexible on average.
Table 8 shows some of the reasons that business owners gave for not having
employees work from home. These are similar for Black-owned and non-Black-owned
businesses, with slightly higher rates of Black employers citing security and IT
concerns as a limiting factor. This may be a remnant of the differences in average
revenue between Black-owned and other businesses, which may constrain
investments in expensive remote work technology.
(https://www.brookings.edu/wp-content/uploads/2023/05/Table7.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 28/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 129
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Table8.png)
BLACK WOMEN ARE STARTING MORE BUSINESSES
Black women are starting more businesses, but still lag
in revenue and payroll
Between 2017 and 2020, the number of employer businesses owned by Black women
rose to 52,374—encompassing 37.2% of all Black-owned businesses and representing
an increase of 1.76 percentage points. Yet despite this growth, there is still significant
progress to be made. Though Black women account for about 7% of the total U.S.
population and 13.9% of American women, their businesses accounted for only 0.9%
of total businesses and 4.2% of women-owned businesses in 2020.
Figure 9 shows how under-represented all women are in employer businesses,
accounting for only 23.6% of the total share of owners. White women own 82.9% of
these businesses (and 17.9% of all businesses). Latino or Hispanic women also have
lower rates of business ownership, representing 18.5% of American women but only
6.8% of all female business owners, and about 1.8% of all business owners. Asian
American women represent 6.3% of American women and 12.8% of female business
owners, and around 3% of all business owners.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 29/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 130
on Black-owned of 801
businesses | Brookings
Figure 10 shows that revenue is distributed even more unevenly. Black-women-owned
businesses represent 1% of all employer businesses, but bring in only 0.3% of total
revenue—behind businesses owned by white women (11%), Asian American women
(1.2%), and Latino or Hispanic women (0.6%). Figure 11 shows that most businesses
owned by Black women make mid-level revenue (between $100,000 to $250,000 per
year) and are over-represented in the lowest-earning categories (below $50,000 per
year).
Though far from equitable, business ownership among Black women is growing
rapidly. The number of Black-women-owned employer businesses increased by 18.1%
between 2017 and 2020, outpacing women-owned businesses in general (9.1%) and
Black-owned businesses (13.6%). These trends parallel the growth in small and sole
proprietorships owned by Black and Latino or Hispanic women. These
microbusinesses account for nearly 80% of all small businesses in the U.S.
(https://sbecouncil.org/about-us/facts-and-data/) , and could be a viable pathway to
greater representation in employer businesses (https://www.brookings.edu/blog/the-
avenue/2021/07/14/how-to-include-microbusinesses-in-economic-growth-
strategies/) if we can create a supportive business environment for owners of color,
and women in particular, to grow their businesses.
The extent to which these high growth rates are the result of structural changes in the
business and lending sectors that have made room for more inclusive growth is
questionable. For example, microbusinesses owned by women of color face unique
structural barriers in access to microfinance (https://smallbusinessmajority.org/our-
research/access-capital/small-businesses-struggling-access-capital-harming-their-
financial-recovery) , as well as opportunities for development and coaching. A closer
scrutiny of these trends could help develop more intentional strategies to uplift these
businesses and promote lasting growth by transitioning them into employer firms.
Ultimately, these disparities illustrate the still highly racialized and uneven business
environment that women of color—and Black women in particular—face. These
structural barriers are an impediment to the growing entrepreneurship among Black
women. For example, 17% of all Black women in the U.S. are in the process of starting
or running a new business (https://hbr.org/2021/05/black-women-are-more-likely-to-
start-a-business-than-white-men) , compared to 10% of white men and 15% of white
women. In 2017, Black women owned over 19% of all women-owned businesses
(https://ventureneer.com/wp-content/uploads/2017/11/2017-AMEX-SWOB-FINAL.pdf)
—above their proportional representation within the U.S. population. However, these
businesses were much more likely to be sole proprietorships than employer firms, with
eight out of 10 businesses owned by Black women being non-employer firms
(https://www.kauffman.org/currents/women-of-color-in-entrepreneurship-new-sbo-
data-and-what-it-means-for-the-economy/) . Due to decades of racial discrimination
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 30/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 131
on Black-owned of 801
businesses | Brookings
and sexist loaning practices, Black women face more structural barriers in accessing
capital, with about 61% of Black women self-funding
(https://www.gemconsortium.org/report/gem-usa-2015-report) their business and an
higher-than-average level of debt
(https://journals.sagepub.com/doi/abs/10.1111/etap.12137) . This is particularly the
case for growing businesses, which face greater barriers to capital access than other
businesses; only 3% of Black-women-owned companies survive longer than five years
(https://www.forbes.com/sites/jenniferpalumbo/2023/01/20/why-african-american-
women-entrepreneurs-have-found-success-through-buying-existing-companies/?
sh=574dd4071218) .
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure9.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 31/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 132
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure10.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 32/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 133
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure11.png)
FACTORS DRIVING GROWTH
Structural factors, rather than an equitable business
environment, seem to be driving growth in business
ownership among Black women
A deeper look at the 2020 data adds nuance to the economic and structural factors
that are driving higher rates of business creation for Black women, but also preventing
the establishment of larger and more profitable businesses.
Table 9 shows that businesses owned by Black women are more likely to be a first
business and less likely to be a primary source of income than the national average.
Among Black-women-owned businesses, 50.8% are the owner’s first business—about
2.2% higher than the rate for all businesses and about 3% lower than the rate for all
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 33/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 134
on Black-owned of 801
businesses | Brookings
women-owned businesses. Additionally, 67.6% of Black-women-owned businesses are
the owner’s primary source of income—about 3.3% lower than the average and 4.8%
lower than male-owned businesses. In other words, employer businesses owned by
Black women are more likely to be secondary to another income source, which limits
the potential for business growth. This may be an indicator of larger economic trends
that are leading people—especially low-income women—to supplement their income
with multiple jobs (https://www.census.gov/library/stories/2021/02/new-way-to-
measure-how-many-americans-work-more-than-one-job.html) to ensure financial
security.
While on the one hand, these statistics illustrate that Black women are starting
businesses at a higher rate than the national average, it also captures the ways that a
lack of financial support and structural barriers in access to capital
(https://www.weforum.org/agenda/2023/03/the-ebb-and-flow-of-vc-funding-how-to-
support-black-women-in-business/) have resulted in Black business owners having a
higher rate of failure for first businesses
(https://link.springer.com/article/10.1007/s41996-021-00081-6) than the national
average.
(https://www.brookings.edu/wp-content/uploads/2023/05/Table9.png)
Most Black-women-owned employer businesses are in the health care and social
services sector, which is also the largest sector for Black-owned businesses overall.
Working in certain health care occupations in 2020 put Black women on the frontlines
of the COVID-19 pandemic, meaning more than any other population, Black women
were being exposed to and witnessing the gaps in our health care, economic, and
political systems. Table 10 shows that in 2020, Black women started 5,364 health care
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 34/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 135
on Black-owned of 801
businesses | Brookings
businesses—the largest number by sector, followed by the professional, scientific, and
technical services sector.
Qualitative data collected during the pandemic suggested that Black women were
entering into health care to fill gaps left by federal and state policies
(https://time.com/5866854/black-women-health-care/) , and the ABS data seems to
confirm that. Despite often being the last to receive federal Paycheck Protection
Program loans (https://www.jstor.org/stable/j.ctv2vr9d90.11?seq=4) , there was a
boom in Black women opening health care businesses to support their communities,
which had higher COVID-19 death and morbidity rates.
(https://www.brookings.edu/wp-content/uploads/2023/05/Table10.png)
The reasons that Black women choose to start businesses are varied, but structural
pressures, such as poor job markets and a desire to help local communities, look to be
crucial factors. Figure 12 shows that in 2020, 24.6% of Black women cited an inability
to find work as a motivation to start a business—behind Black men (at over 27%), but
2.9% higher than the average rate and 6.8% higher than the rate for white women,
who cited this reason the least. At the onset of the COVID-19 pandemic, an extreme
drop in business revenues and stable employment struck Black communities at
proportionally higher rates than most other racial groups. The inequity already baked
into the economy—including in structural disparities in liquidity and credit
(http://dln.jaipuria.ac.in:8080/jspui/bitstream/123456789/1332/1/COVID-19s-effect-
on-minority-owned-small-businesses-in-the-United-States.pdf) —meant more Black-
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 35/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 136
on Black-owned of 801
businesses | Brookings
owned business closed quickly and permanently. Black workers suffered
disproportionately from job loss
(https://edworkforce.house.gov/uploadedfiles/wilson_testimony.pdf) , given their
overrepresentation in frontline sectors such as health care and service industries.
Black women were also more motivated by a desire to help their community. Over 75%
of Black women listed “wanting to help my community” as a key factor when deciding
to start a new business. This is 15.5% higher than the average percentage for all
businesses and over 18.3% higher than for white men, who were the least likely to cite
this as an influential factor. The ABS data captures some businesses created in 2020,
but includes a larger set of already established businesses from across the U.S. Thus,
while these figures may be higher due to the disproportionate impacts of COVID-19 on
Black communities, it also captures the realities of decades of structural racism
limiting the economic opportunities for Black Americans.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 36/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 137
on Black-owned of 801
businesses | Brookings
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 37/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 138
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure12.png)
Black women business owners tend to be younger and more educated than business
owners of other groups, possibly illustrating the higher barriers that exist for Black
women to access the capital and financial support needed to start a business, as well
as the entrepreneurial drive among younger Black women. Figure 13 shows that most
Black women business owners are 45 to 54 years old, compared to 55 to 64 for all
businesses. Additionally, 6.6% are 34 years old or younger, compared to 5.7% for all
businesses.
Table 11 shows that a comparatively substantial portion of Black women business
owners (37.9%) have an advanced degree, including 19.1% with a master’s degree. This
is more than 10 percentage points higher than the national average of 23.5%. This
discrepancy indicates that the barriers to business ownership are still higher for Black
women, who cannot receive equal economic support without earning more
credentials.
On the other hand, these figures also highlight the role that highly educated Black
women are playing in the growth of new employer businesses. Like in other arenas,
including voter participation and community engagement
(https://www.pewresearch.org/fact-tank/2022/10/12/key-facts-about-black-eligible-
voters-in-2022/) , Black women are a potent force, helping to shape a more inclusive
and innovative economy despite structural barriers. It is likely that decades of civil
rights activism have paved the way for generational change—finally creating viable
space for young, educated Black women to start businesses at such high rates.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 38/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 139
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure13.png)
(https://www.brookings.edu/wp-content/uploads/2023/05/Table11.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 39/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 140
on Black-owned of 801
businesses | Brookings
Figure 14 shows how the economic and business environment has changed for Black
women since the 1980s. The relatively small proportion of Black women who owned
businesses pre-2008 is testament to the lasting impacts of racist policies on business
longevity, which have lowered the security of Black businesses by blocking
opportunities for household and community wealth retention
(https://www.brookings.edu/essay/homeownership-racial-segregation-and-policies-
for-racial-wealth-equity/) , thus limiting the potential for sustained growth.
Most businesses owned by Black women today were started between 2013 and 2017,
with the number growing since at least 2008, and at proportionally higher rates than
other gender and racial or ethnicity groups. Conversely, businesses owned by white
men and women are older, with proportionally more businesses started between 1980
and 2007.
In some respects, this illustrates the low survival rate for Black-women-owned
businesses. However, the notable increases since 2008 could illustrate that greater
numbers of Black women are seeing business ownership as a pathway to a more
secure and stable livelihood, responding to increasingly precarious employment,
financial, and housing conditions. Like during the pandemic, the post-2008 global
financial economy involved heavy structural change, which undermined the stability of
many low-income and Black and brown households (https://www.epi.org/blog/the-
great-recession-education-race-and-homeownership/) . It may be that these
structural factors—rather than improvements in racial equity—have driven the growth
in Black-women-owned employer businesses.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 40/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 141
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure14.png)
CALIFORNIA BOOMING
California is booming for Black women business owners
Figure 15 shows that many of the fastest-growing areas for Black-women-owned
businesses are in California. Between 2017 and 2020, the number of employer
businesses owned by Black women grew by 93.14% in San Francisco (to 817
businesses); by 52% in Riverside (to 376 businesses); and by 44% in Los Angeles (to
2,310 businesses). Compared to Black men, the top metro areas for business growth
for Black women are spread more evenly across the geographic regions of the U.S.
Except for Los Angeles and Philadelphia, most of these large increases have occurred
in cities with smaller total numbers of Black-women-owned businesses (fewer than
500).
In contrast, some of the largest declines in the number of Black-women-owned
businesses have been in large eastern cities, including Hartford, Conn. and Baltimore,
as well as some southern cities, such as Houston.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 41/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 142
on Black-owned of 801
businesses | Brookings
Table 12 displays Black women business owners as a proportion of all women business
owners. It shows that seven of the top 10 metro areas are in the Southeast, including
Memphis, Tenn. (15%); Jackson, Miss. (15%); and Atlanta (11%). St. Louis has the
highest percentage of Black-women-owned businesses, at 21% of all women-owned
businesses and 2,566 businesses in total. Most of these metro areas (except Durham,
N.C.) also have higher percentages of Black-owned employer businesses.
(https://www.brookings.edu/wp-content/uploads/2023/05/Table12.png)
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 42/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 143
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure15-01.png)
Figure 16 displays the percentage point changes in proportional representation for
Black women employers. It shows that large southern cities—including Nashville,
Tenn.; Memphis, Tenn.; and Birmingham, Ala.—are becoming more equitable, while
Lansing, Mich.; Mobile, Ala.; and St. Louis have become more inequitable. Between
2017 and 2020, Killeen, Texas and Jackson, Miss. have had the largest shifts toward a
more equitable rate of business ownership.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 43/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 144
on Black-owned of 801
businesses | Brookings
(https://www.brookings.edu/wp-content/uploads/2023/05/Figure16-01.png)
MORE ACTION IS NEEDED
More action is needed to transform strong growth into
lasting change
Looking across the data from 2017 to 2020, it’s clear that there has been substantial
progress toward equal racial representation in owners of employer businesses. Black
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 44/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 145
on Black-owned of 801
businesses | Brookings
women in particular have made strides, with growth rates well beyond other racial,
ethnicity, and gender groups. And this progress ought to be celebrated.
However, more action is needed to deliver lasting and transformative change toward a
more inclusive economy. Realistically, these growth rates are well behind a pace that
could ameliorate racial disparities in employer-firm ownership anytime soon, and large
structural barriers across the economy—not least of which is the racialized nature of
investments in place (https://www.brookings.edu/research/what-the-lack-of-premium-
grocery-stores-says-about-disinvestment-in-black-neighborhoods/) —continue to
undermine transformative change.
Given that many of these disparities are structural in nature, the solutions must be as
well. Past reports in this series have highlighted policy targets to expediate the growth
in Black business ownership, including general targets
(https://www.brookings.edu/essay/to-expand-the-economy-invest-in-black-
businesses/) and more pointed structural changes in federal funding and philanthropy
(https://www.brookings.edu/research/black-owned-businesses-in-u-s-cities-the-
challenges-solutions-and-opportunities-for-prosperity/) , and we reiterate those here.
While there are strong signs of real progress, if we are to translate gains and policy
commitments into lasting change, we need to take seriously the notion that a more
inclusive economy is a powerful vehicle for delivering a more robust and prosperous
economy.
In this moment of historic federal funding across the innovation sector as well as in
climate change adaptation and mitigation, the opportunities are ripe to invest in and
grow Black business ownership, especially in historically underinvested areas. The
Biden administration has shown a real appetite to act on racial justice, including
guidelines (such as the Justice40 initiative and the executive order to strengthen
racial equity) to ensure investments flow to historically disadvantaged communities.
These systems must also be supported by local-level policy that can bend
investments toward more equitable ends, including proactively forecasting how
different racial groups will be impacted by how those investments are spent. This kind
of active approach to structural change is necessary if we are serious about fostering
and accelerating the development of Black-owned employer businesses.
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 45/46
8/31/23, 6:54 PM Case 1:23-cv-03424-TWT
Who is Document
driving Black business 59-4 from
growth? Insights Filed 08/31/23
the latest data Page 146
on Black-owned of 801
businesses | Brookings
AUTHORS
Andre M. Perry Senior Fellow - Brookings Metro @andreperryedu
Manann Donoghoe Senior Research Associate - Brookings Metro
@ManannanAD
Hannah Stephens Research Assistant - Brookings Metro
Footnotes
1. Throughout this report, figures for “Black residents” refers to those who identify as
Black alone or in combination with one or more other races, while “Black-owned
businesses” refers to business with owners who identify as Black alone. Latino or
Hispanic population and business owner figures refer to those who are Latino or
Hispanic alone, as do figures for Asian Americans and white Americans. Population
data is drawn from the 2020 census.
Copyright 2023 The Brookings Institution
https://www.brookings.edu/articles/who-is-driving-black-business-growth-insights-from-the-latest-data-on-black-owned-businesses/#:~:text=One key… 46/46
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 147 of 801
Exhibit E
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 148 of 801
Building supportive
ecosystems for Black-
owned US businesses
The right business ecosystems can mitigate or
negate the effects of structural obstacles to
business building for Black business owners—
and add $290 billion in business equity.
October 2020
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 149 of 801
Cover image:
© FG Trade/Getty Images
Copyright © 2020 McKinsey &
Company. All rights reserved.
This publication is not intended to
be used as the basis for trading in
the shares of any company or for
undertaking any other complex or
significant financial transaction
without consulting appropriate
professional advisers.
No part of this publication may be
copied or redistributed in any form
without the prior written consent of
McKinsey & Company.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 150 of 801
Public Sector Practice
Building supportive
ecosystems for Black-
owned US businesses
The right business ecosystems can mitigate or
negate the effects of structural obstacles to
business building for Black business owners—
and add $290 billion in business equity.
October 2020
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 151 of 801
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 152 of 801
Contents
2 Introduction
6 Barriers to business building for Black entrepreneurs
12 Interventions
15 Conclusion
16 Endnotes
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 153 of 801
Introduction
By David Baboolall, Entrepreneurship and business ownership—particularly of community-based businesses—are crucial
Kelemwork Cook, ways to develop community wealth, for both business owners and the people they employ. Healthy Black-
Nick Noel,
owned businesses could be a critical component for closing the United States’ Black–white wealth gap,
Shelley Stewart, and
Nina Yancy which we project will cost the economy $1 trillion to $1.5 trillion (in 2018 dollars) per year by 2028.1 The
COVID-19 crisis, however, has further stressed Black-owned businesses and may cause the racial wealth
gap to widen. This gap includes a $290 billion—and growing—opportunity to grow overall wealth by
achieving revenue parity between Black- and white-owned businesses in addition to providing aid to small
and medium-size businesses (SMBs)—those with up to 500 employees—with nonwhite owners.2
Black business owners have been disproportionately affected by the pandemic-linked economic
downturn,3 partly because they were more likely to already be in a precarious position, including more
likely to be located in communities with business environments that are more likely to produce poor
business outcomes. Indeed, about 58 percent of Black-owned businesses were at risk of financial
distress before the pandemic, compared with about 27 percent of white-owned businesses.4 The
pandemic contributed to tipping 41 percent of Black-owned US businesses into closure from February to
April 2020.5 More than 50 percent of the owners of surviving Black businesses surveyed in May reported
being very or extremely concerned about the viability of their businesses. This concern may be linked
to having a more difficult time accessing credit since the COVID-19 crisis began; 36 percent of Black
business owners responding to the survey said they had experienced this, compared with 29 percent of
all respondents.6
Black Americans have never had an equal ability to reap the benefits of business ownership. While about
15 percent of white Americans hold some business equity, only 5 percent of Black Americans do. Among
those with business equity, the average Black American’s business equity is worth about 50 percent of the
average American’s and a third of the average white American’s (Exhibit 1).7
Black-owned businesses also tend to earn lower revenues in most industries and are overrepresented in
low-growth, low-revenue industries such as food service and accommodations.8
This gap in business activity contributes to an overall lower level of prosperity for Black families: the median
white family’s wealth is more than ten times the wealth of the median Black family’s.9 This disparity is also a lost
opportunity for the US economy as a whole. If existing Black-owned businesses reached the same average
revenue as their white-owned industry counterparts (excluding publicly held companies), the result would be
an additional $200 billion in recurring direct revenues, which could equal about $190 billion in additional GDP,
or a roughly 1 percent increase in 2017 GDP. (For our analytical approach, see sidebar “Quantifying the impact
of revenue parity.”)
2 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 154 of 801
Exhibit 1
Black Americans hold
Black hold significantly
significantly less
lessbusiness
businessequity
equitythan
thantheir
theirwhite
white counterparts.
counterparts.
Status of business equity ownership in the United States, by race
Mean business equity, thousands of 2019 dollars
White Black Hispanic
1,500 1,475
1,373
1 in 20
Proportion of Black Americans
1,129 holding business equity
1,045
1,000
4×
Value of mean business
605 610 equity held by white Americans
relative to Black Americans
500 466 in 2019
385 346 360
338
105
0
2010 2013 2016 2019
Quantifying the impact of revenue parity
To quantify the economic impact of Our analysis shows that if Black-owned To estimate the macroeconomic effects of
achieving revenue parity between Black- companies were to attain the same revenue parity between Black- and white-
owned and white-owned US small and average revenue in their industries as owned businesses, we used a model that
medium-size businesses (SMBs), we used white-owned companies, their revenue draws on data from about 141 countries
the most recent results of the US Census gains would be about $200 billion. These and 60 sectors to translate disruptions,
Bureau’s Survey of Business Owners, from revenue gains would be concentrated shocks, and policy changes into cross-
2007 and 2012. We first estimated the rate in the healthcare, retail, and wholesale industry and macroeconomic effects. A
of new-business formation and revenue sectors, which would account for 51 $200 billion increase in revenues is worth
growth for businesses owned by Black percent of Black-owned SMBs’ total new about $190 billion in additional annual
and white US residents. Then we used revenue. This estimate does not account GDP, or a roughly 1 percent increase in
those growth rates to estimate the number for the higher revenue’s multiplier effects, 2017 GDP.¹
of companies owned by Black and white which would represent the impact of
Americans in each industry as well as to the change on the overall economy.
estimate the revenues of those businesses.
¹ For our analysis, we assumed that the economy can absorb most of the revenue increases, so any cannibalization of revenues from Black-owned firms’ revenue growth (to
attain parity with white-owned firms) is not offset by losses in white-owned firms.
Building supportive ecosystems for Black-owned US businesses 3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 155 of 801
A silver lining of COVID-19 and the
racially charged violence in 2020 may
end up being that some large companies
have now launched programs to support
Black-owned businesses.
Before the pandemic, the average American held 15.3 percent of his or her wealth in business equity, with
a mean value of $147,000. SMBs constituted 99 percent of US businesses and were responsible for
62 percent of the net increase in private-sector jobs from 1993 to 2017.10 Entrepreneurship and an ability
to start new businesses matter significantly here, as more than 50 percent of this net increase was the
result of the formation of new businesses, defined as those less than 3.5 years old.11
Our analysis mapped barriers throughout the entrepreneurial pathway: ideation and starting up, sustaining,
scaling, and exiting (see sidebar “Entrepreneurial pathway: Building a business, from ideation to exit”).
We focus here on the systemic barriers that hinder Black entrepreneurs’ efforts to ideate and start and
sustain local businesses—that is, community-based SMBs that are part of existing supply chains.12 We
then outline interventions that can counteract the effects of these obstacles, particularly opportunities
for coordinated efforts. Indeed, US institutions need to repair Black business owners’ trust in the business
ecosystem—particularly in companies in financial and business services.13 This historic lack of trust may be
slowing business creation; research shows that Black business owners may believe that they need to be
better qualified than their white counterparts: 30 percent of Black owners of employer firms (businesses
with at least one paid employee) hold an advanced degree, which is true of 22 percent of their white peers.14
A silver lining of COVID-19 and the racially charged violence in 2020 may end up being that some large
companies have now launched programs to support Black-owned businesses: a major social media
company dedicated $40 million in grants to support Black-owned US businesses with 50 or fewer
employees15; a financial services firm pledged $1.15 billion, including $350 million in procurement spending
on Black-owned businesses, to close the racial wealth gap.16 However, these individual actions will not
in themselves be enough to effect the needed change. For this work to have a systemic impact, entire
business ecosystems will need to be involved. Once successful, however, this effort would benefit not only
the US economy but US society as a whole.
4 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 156 of 801
Entrepreneurial pathway: Building a business, from ideation to exit
Entrepreneurship generally follows a on establishing the business and ensuring Once businesses attain annual revenues
five-step process that culminates in an exit its survival. They may hire employees, raise of at least $1 million, they can scale. The
for the founder (exhibit). additional capital, and develop a financial focus of this stage is on increasing working
cushion for the business. capital to support growth and expanding
During ideation, prospective entrepreneurs into new geographic or product markets.
research and assess business ideas before While sustaining the business, Businesses in this phase may also acquire
developing business plans in preparation entrepreneurs focus on increasing revenue or merge with another company.
for launch. Foundational activities at this and reaching profitability. In this stage,
stage include obtaining the relevant patents, entrepreneurs develop growth strategies, Finally, entrepreneurs can exit their
participating in business incubators, and further develop their companies’ sales and businesses through an acquisition or an
securing funding and support, often from marketing efforts, expand their teams, and initial public offering (IPO) that enables
the entrepreneur’s networks. scale business functions. Businesses may them to receive a one-time realization of
continue to raise venture capital. the business’s value.
In the start-up stage, which lasts about
three and a half years, entrepreneurs focus
Exhibit
Entrepreneurship follows five
Entrepreneurship stages.five stages.
follows
Focus areas
Ideate
Researching market attractiveness
Start up
Establishment and survival
Sustain
Growing profits
Scale
Market expansion
Exit
Capturing business value
Source: Dow Jones VentureSource, 2016; Global Entrepreneurship Monitor, 2017; 2016 annual survey of entrepreneurs, US Census Bureau
Building supportive ecosystems for Black-owned US businesses 5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 157 of 801
Barriers to business building
for Black entrepreneurs
Throughout the business-building process, Black business owners face economic, market, sociocultural,
and institutional barriers, which are all linked to racial discrimination in the United States. Economic barriers
relate to disempowerment and the costs of low starting levels of capital—for individuals, families, and
communities. Market barriers result from unaddressed needs, often related to challenges of access—to
capital, expertise, and services. Sociocultural barriers encompass the biased and exclusionary ways
in which Black entrepreneurs are more likely to be blocked from gaining social capital, such as helpful
relationships that make up business networks. Finally, institutional barriers are supported by the systems in
which Black-owned businesses operate and include factors as basic as their locations.
Ideating
Black entrepreneurs tend to make decisions in the business-ideation stage that are likely to keep their
businesses small. Black entrepreneurs generally pursue businesses in less lucrative sectors. Only five
industries represent employer businesses (that is, businesses that employ paid employees) owned by
74 percent of Black women business owners and 62 percent of Black male business owners: healthcare
and social assistance; professional, scientific, and technical services; administrative support and waste
management services; construction; and transportation and warehousing (Exhibit 2).17
These sectors represent only 20 percent of business revenues overall. Meanwhile, although wholesale
businesses represent 24 percent of business revenues, only 1 percent of Black women and 2 percent
of Black men who are entrepreneurs are in the sector (see sidebar “Uphill climb for Black women in
entrepreneurship”).18
Location can also limit Black entrepreneurs’ business potential. Sixty-five percent of Black Americans live
in 16 states that are below the US average on indicators of economic opportunity.19 Within their communities,
Black Americans are also disproportionately concentrated in economically disadvantaged neighborhoods.20
Black entrepreneurs from these communities are less likely than their white peers to have exposure and
access to lucrative business opportunities.
Black entrepreneurs might also lack access to the networks and relationships that could help them make
optimal business decisions. Research on New York–based start-ups shows that founders who are mentored
by top-performing entrepreneurs are three times more likely than their co-located peers without mentors
to become top performers themselves.21 Outside the start-up world, a knowledgeable contact may help a
prospective entrepreneur make decisions such as whether to buy a franchise for up-front capital or to build
an independent business.
6 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 158 of 801
Exhibit 2
Black-ownedemployer
Black-owned employerbusinesses
businesses are
are concentrated
concentrated in in only
only fivefive industries.
industries.
Black women Black men White women White men
Distribution of employer Revenues Distribution of employer Revenues
firms1 by sector, by sector, firms1 by sector, by sector,
% of businesses % of sales % of businesses % of sales
Healthcare and 48 6 Finance and 2 11
social assistance 19 insurance 5
15 4
9 5
Professional, 13 5 Real estate 2 2
scientific, and 14 3
17 7
technical services 5
16
Administrative, 7 2 Construction 2 4
support, and waste- 10 10
management 7 6
6 16
services
Retail trade 6 13 Transportation 2 3
8 and warehousing 9
12 2
10 4
Accommodation 5 2 Wholesale trade 1 24
and food services 6 2
8 4
7 6
Other services2 5 1
6
8
6
Note: Chart excludes the following sectors: agriculture, forestry, fishing, and hunting; arts, entertainment, and recreation; educational services; information;
management of companies and enterprises; manufacturing; mining, quarrying, and oil and gas extraction; utilities; and other or unclassified services and indus-
tries (except public administration).
1
A business with at least one paid employee.
2
Includes businesses such as advocacy, death-care services, dry cleaning and laundry, equipment and machinery repair, hair and nail salons, personal-care
services, pet-care services, photofinishing services, and religious activities.
Source: 2012 survey of business owners and 2016 annual survey of entrepreneurs, US Census Bureau
Building supportive ecosystems for Black-owned US businesses 7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 159 of 801
Uphill climb for Black women in entrepreneurship
About 20 percent of the 12.3 million sources of income or that they are unable year than Black women–owned employer
women-owned businesses in the United to focus on entrepreneurship. What’s more, businesses.⁶ Worse, Black women’s
States are owned by Black women.¹ Indeed, 68 percent of Black women entrepreneurs business revenues actually declined by
Black women were the first to exceed 50 with paid employees are 11 percent, or an average of $3,000 per
percent of business ownership within their in three industries that make up only year, from 2012 to 2018.⁷
racial or ethnic group.² However, Black 13 percent of overall business revenues³:
women encounter cross-cutting economic healthcare and social assistance, Black women are also disproportionately
and institutional barriers because of their professional services, and administrative, shut out of venture-capital funding,
race and gender. support, and waste-management services.⁴ which can rapidly fuel business growth.
Women of color receive less than 0.2
Black women’s high rates of entrepre- Black women also tend to have fewer percent of venture-capital funding, and at
neurship may reflect economic necessity employees than Black men do, even 4 percent of funded US start-ups,⁸ Black
rather than opportunity; 64 percent of though Black women own 50 percent more women founders are underrepresented
Black women spend less than 40 hours a businesses.⁵ As a result of these disparities, and underfunded. Indeed, the average
week on their businesses, which suggests employer businesses owned by Black men Black woman–led start-up that received
that the business may be one of several earn an average of $660,000 more per funding raised $42,000.
¹ The 2018 state of women-owned business report, American Express, americanexpress.com.
² Dell Gines, “Black women business startups,” Federal Reserve Bank of Kansas City, 2018, kansascityfed.org.
³ 2016 annual survey of entrepreneurs, US Census Bureau, 2016, census.gov.
⁴ 2016 annual survey of entrepreneurs, US Census Bureau, 2016, census.gov.
⁵ “Statistics for all U.S. firms by industry, gender, ethnicity, and race for the U.S., states, metro areas, counties, and places: 2012,” US Census Bureau, 2012, data.census.gov.
⁶ “Annual business survey: Statistics for employer firms by industry, sex, ethnicity, race, and veteran status for the U.S., states, metro areas, counties, and places: 2017,” US
Census Bureau, 2017, data.census.gov.
⁷ “The state of women-owned businesses,” American Express, 2018, womenentrepreneurscharleston.com.
⁸ https://www.projectdiane.com/.
Crucially, Black entrepreneurs struggle more to secure capital and access to credit. Even with strong
personal credit, Black business owners and other entrepreneurs from marginalized groups are about
half as likely as their white counterparts to receive full financing.22 Friends and family may not be able to
contribute capital either. Most Black families surveyed said that they didn’t know anyone who could lend
them $3,000.23
Starting up and sustaining
The focus of the start-up and sustain phases of entrepreneurship should be on ensuring the business’s
survival and increasing its profitability. However, Black entrepreneurs’ mistrust of institutions—based on
their experiences and knowledge of the United States’ history of discrimination—can make them wary of
using potential resources outside their communities. Combined with a lack of informal and geographic
connections to such resources—including venture-capital networks—this tendency can isolate Black
entrepreneurs from potential sources of support.24 (For an overview of the public sector’s attempts at
closing the gap between Black- and white-owned businesses—and their pitfalls—see sidebar “Legislating
for more equitable business outcomes.”)
8 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 160 of 801
Legislating for more equitable business outcomes
Since 1974, legislation has partially white neighborhoods are more likely to with its intention. This oversight may
closed the gap between Black- and white- be approved for credit cards than those in have impeded Black-owned businesses’
owned businesses. As a result of many of mostly Black neighborhoods.² participation, especially if they were
these advancements, the Small Business already skeptical of the program.⁴
Administration backed about $210 million In response to the savings and loan crisis,
in loans and helped award $2.3 billion Congress passed the Financial Institutions In 2010, the Dodd-Frank Wall Street
in federal contracts to disadvantaged Reform, Recovery, and Enforcement Act Reform and Consumer Protection Act
businesses in 2019.¹ However, the laws of 1989, which increased the ability of required some federal agencies, including
and programs are often imperfectly third-party watchdogs to analyze bank the Securities and Exchange Commission,
implemented, with gaps such as a lack of records and identify discriminatory lending to establish Offices of Minority and
information on how to apply to programs practices. However, loan originators Women Inclusion. However, the difficulty
that may benefit Black entrepreneurs. also shifted from denying loans to Black of enforcing the spirit of the act often
residents to targeting Black communities put Black-owned businesses behind.
In 1974, the Equal Credit Opportunity Act for higher-cost, higher-risk loans.³ Later, in During the most recent—and ongoing—
was passed. It prohibits credit issuers 1998, the Small Business Administration crisis, the Coronavirus Aid, Relief, and
from discriminating on the basis of directly targeted small and medium-size Economic Security Act aimed to provide
traits such as race, gender, and marital businesses (SMBs) from marginalized emergency relief to SMBs but failed to
status. The Community Reinvestment Act communities by creating the Mentor- reach many Black-owned businesses. Part
followed, in 1977. A tool to discourage Protégé Program. Designed to pair SMBs of the gap came from a failure to include
redlining, the act encourages financial from communities with relatively few many community-development funding
institutions to meet the credit needs of resources with experienced government institutions in its program. The act also
customers in low- and moderate-income contractors, the program was later found fails to provide information and training
neighborhoods. However, “credit redlining” to lack effective controls to ensure that the on how to apply for the programs and
persists: consumers who live in mostly application review process was consistent access funds.⁵
¹ A shley D. Bell, “Opportunities for black-owned businesses are thriving,” press release, February 21, 2020, US Small Business Administration, sba.gov.
² Connie Prater, “Who gets credit cards may be a matter of black and white,” CreditCards.com, June 4, 2008, creditcards.com. The rules that govern the Community
Reinvestment Act’s implementation are under revision by the US Office of the Comptroller of the Currency, the Federal Deposit Insurance Corporation, and the Federal
Reserve Board.
³ Justin P. Steil et al., “The social structure of mortgage discrimination,” Housing Studies, Volume 33, Issue 5, 2018, pp. 759–76, ncbi.nlm.nih.gov.
⁴ Evaluation of SBA’s All Small Mentor-Protégé Program, US Small Business Administration, September 17, 2019, buildsmartbradley.com.
⁵ Ben Popken, “Why are so many black-owned small businesses shut out of PPP loans?” NBC News, April 29, 2020, nbcnews.com.
Only 4 percent of Black American businesses survive the start-up stage, even though 20 percent of
Black Americans start businesses.25 Even if they survive the start-up stage, Black-owned businesses still
disproportionately struggle with debt and raising capital in addition to challenges such as a lack of helpful
relationships in the business community. Indeed, participants in focus groups said that they feel that their
race—and sometimes age and gender—makes loan officers hesitant to lend to them.26 Such obstacles
make Black-owned businesses less likely to survive and grow.
Building supportive ecosystems for Black-owned US businesses 9
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 161 of 801
Start-up capital is associated with better business performance,27 but Black entrepreneurs have less
of it. Black entrepreneurs start their businesses with about $35,000 of capital, white entrepreneurs
$107,000.28 As a partial consequence, Black-owned businesses report higher levels of debt relative to
revenues. Almost 30 percent of Black-owned businesses spent more than 50 percent of their revenues to
service their debt in 2019.29 Indeed, Black entrepreneurs are three times as likely as white entrepreneurs
to say that a lack of access to capital negatively affects their businesses’ profitability and almost twice
as likely to cite the cost of capital.30 This lack of capital occurs even though the requirements for Black
applicants are more stringent. In one study, 73 percent of Black loan applicants were asked to provide
financial statements for their businesses, compared with 50 percent of white applicants with comparable
profiles. In addition, 31 percent of Black applicants were asked to provide their personal W-2 forms. No
white applicants received such a request.31
These interactions with financial institutions are consistent with findings that nonwhite ownership of a
business is a strong predictor of loan denial outside of majority-nonwhite communities.32 However, simply
being located in a co-ethnic market—where community members, customers, and business owners are of
the same ethnicity—results in lower revenue, even for businesses of comparable quality.33 Indeed, research
shows that even when they have better Yelp reviews, businesses in majority-Black communities have lower
revenues than those outside majority-Black communities. Black-owned businesses are in a double bind
that results in missing revenues.
Black entrepreneurs can also have difficulty accessing expertise and business services. Black owners
of employer firms, which are more likely to benefit from services such as legal and financial advisory, are
less likely to seek them out. Only 58 percent of Black owners sought professional services, for reasons
including expense, inaccessibility, and mistrust, compared with 70 percent of white owners.34
Black entrepreneurs are three times as
likely as white entrepreneurs to say that
a lack of access to capital negatively
affects their businesses’ profitability and
almost twice as likely to cite the cost
of capital.
10 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 162 of 801
Business networks can support Black entrepreneurs, but Black entrepreneurs are less likely to know
and hear about relevant networks that can help support and promote their businesses. Indeed, Black
entrepreneurs are likely to be excluded from receiving information about high-potential opportunities,
even though focus-group participants said that they would like to connect with a variety of business
professionals and mentors. This exclusion translates into fewer connections to formal hubs, such as
banks and venture-capital funds,35 and from informal networks.36 One sign of this disconnection is
that corporate and government procurement programs that target Black-owned businesses tend to
be underused. The Minority Business Development Agency (MBDA) found in 2016 that Black-owned
businesses saw less utilization relative to their availability than white-owned businesses did in the same
industries. The median share of contract dollars awarded to Black-owned businesses across five key
industries—architecture and engineering, construction, goods and supplies, professional services,
and other services—was 4 to 44 percent of contractors’ availability, compared with 49 to 61 percent
for white-owned businesses.37
The aggregate barriers to starting and sustaining a Black-owned business translate structural bias into
less access to capital, lower revenue, and dimmer prospects for business growth. The mainstream financial
system’s role in those barriers has helped to maintain Black Americans’ distrust of the financial sector as
well as fear of debt.38
Building supportive ecosystems for Black-owned US businesses 11
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 163 of 801
Interventions
Insufficient access to capital, knowledge, and support ultimately leaves many Black entrepreneurs
less economically mobile and limits the potential of entrepreneurship to grow wealth for Black families
and communities. Interventions to tackle these barriers will require public-, private-, and social-sector
stakeholders to evaluate current business ecosystems and rebuild them to be equitable and more supportive
for Black business owners. In particular, local anchor institutions such as hospitals and universities can take
the lead on collaborating with the public and social sectors to coordinate the work of disparate organizations.
Local economic-development institutions, community development financial institutions, chambers of
commerce, and other government agencies such as the Small Business Administration or MBDA are already
engaged in these activities. But because anchor institutions are tethered to their communities, they stand to
reap long-term benefits from investing in more equitable—and healthier—business communities.
Indeed, this work requires instituting policies that create more equitable outcomes in addition to assuring
equitable access to capital. Black leaders’ input can help create positive experiences and build Black
entrepreneurs’ trust in institutions and ecosystems that have felt exclusionary. In addition, the private and
social sectors should help Black-owned businesses build capabilities and facilitate knowledge sharing.
Expanded opportunities for mentorship (advisory) and sponsorship (advocacy) can further solidify bonds
between Black entrepreneurs and other stakeholders in business ecosystems.
Implement policies that produce equitable outcomes
The public, private, and social sectors can all participate in removing institutional barriers for Black-
owned businesses. Stakeholders could ensure that laws, policies, and practices are designed to produce
equitable opportunities and outcomes.
Policy makers, industry groups, affinity groups, and coalitions of stakeholders could enforce laws and
policies aimed at preventing inequitable processes and outcomes. After a review of the current laws and
policies and their effects, stakeholders could consider updating antidiscrimination policies. Crucially,
stakeholders could set up—or support—internal and external watchdog groups dedicated to creating
research on equity in processes and policy outcomes.
In addition, procurement practices, especially at anchor institutions and large organizations, can evolve
to be more inclusive of Black-owned businesses. Besides dedicating funds to procurement from Black-
owned businesses, large organizations could simplify their minority-supplier certification processes so
that they can take on new suppliers more quickly. Organizations should also dedicate funding to supplier-
development programs that can help Black-owned suppliers better participate in supply chains.
Once their supplier programs are established, organizations should track their procurement spending
with important categories of suppliers and make the data more widely available to track progress and
promote accountability. For instance, one North American bank committed to tracking its spending with
diverse suppliers, the number of minority-certified suppliers it engages, and the number of requests for
information and proposals to these suppliers.39 This work is especially urgent during the pandemic-linked
economic crisis, and entities that have an interest in keeping business ecosystems vibrant may be able to
lead the effort.
12 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 164 of 801
Enable equitable access to capital
To overcome economic barriers, Black-owned SMBs need direct investment or in-kind equity
contributions, including grants, subsidies, loans, and revenue-participation agreements. Direct
investment is especially important during the COVID-19 crisis. Our analysis suggests that an additional
$7.6 billion to $15.4 billion in liquidity for Black-owned SMBs in the 2020–21 time frame—less than 3
percent of the $659 billion authorized under the Paycheck Protection Program—could preserve 460,000
to 815,000 jobs,40 for an average of $9,325 to $33,478 per job.
The public and private sectors’ immediate priority during the COVID-19 economic crisis has been to
provide SMBs with grants and loan deferrals. However, increasing the availability of financial resources,
particularly start-up and expansion capital, for Black entrepreneurs, will be crucial to improving
business owners’ experience and replenishing the pipeline of growing businesses. For example,
banks, conventional and social-impact investors, foundations, and public programs could make more
capital available to Black-owned businesses. Possible programs include ones that make extending
capital to Black-owned businesses less risky through measures such as guaranteeing funds and
programs that help entrepreneurs from marginalized backgrounds acquire more capabilities. A short-
term counterweight to the effects of bias in the loan process could be conducting fewer in-person
transactions. Indeed, Black entrepreneurs are already more likely to borrow online, where their race may
be less salient.41 A program for small-business investment companies that provides high-growth small
businesses with capital and R&D funding could also accelerate the growth of Black-owned businesses
that hold intellectual property and may foster Black entrepreneurship in high-growth industries.
A network of volunteer professional services providers and coaches could help Black-owned businesses
navigate the process of attaining loans, grants, and other affordable capital, including from large
corporations and nonprofit organizations.42
An additional $7.6 billion to $15.4 billion
in liquidity for Black-owned SMBs
in the 2020–21 time frame—less than
3 percent of the $659 billion authorized
under the Paycheck Protection
Program—could preserve 460,000 to
815,000 jobs, for an average of $9,325
to $33,478 per job.
Building supportive ecosystems for Black-owned US businesses 13
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 165 of 801
In addition to continuing the long-term project of cultivating Black leaders in positions of influence on
capital funding, cross-sector entities could convene and support a consortium of Black fund managers
to participate in issuing and managing debt and equity investments in small businesses. The SBA already
licenses small-business investment companies that guarantee funds that invest in SMBs. Increasing the
share of Black fund managers in this program and simultaneously investing in Black managers—and the
pipeline of Black leaders—in the private sector could have compounding effects.
Build business capabilities and facilitate knowledge sharing
To overcome market barriers, Black-owned SMBs need support for building capabilities and sharing a
greater amount of knowledge. Black service providers, compensated by organizations working to support
equity in entrepreneurship, can lead these capability-building efforts. This work would protect and
strengthen Black-owned businesses and build business networks with Black-owned SMBs as hubs.
The private and social sectors—particularly anchor institutions—could provide resources, including help
with reskilling and upskilling Black-owned businesses’ workers, to make Black-owned SMBs nimbler.
For instance, a major social-media platform built a resource hub to provide businesses with industry-
specific information they could use during the pandemic. Similarly, on-the-job training and free web-
based courses are both resources that can be easily shared among multiple—indeed, many—businesses.
Business-services providers could also facilitate digital transformations to help businesses identify new
market opportunities.
Finally, digital capabilities will increase Black entrepreneurs’ share of opportunities, but the corresponding
business services are long-unmet needs. Many Black-owned businesses lack the resources to hire service
providers that can help them digitize their businesses, but private-sector and social-sector organizations
can provide free technology services and managerial assistance. Free or subsidized installation, tech
support, and staff training can help Black-owned businesses acquire more digital capabilities and become
more able to share this knowledge with other Black-owned businesses in their communities.
Expand opportunities for mentorship and sponsorship
Representation and participation in networking, mentorship, and sponsorship programs can help Black
entrepreneurs overcome some sociocultural barriers. The private sector should take the lead by building
more inclusive teams. Managers at every level should consciously develop—and sponsor the careers
of—more Black leaders to counteract the effects of structural bias. Indeed, only about 3 percent of
current Fortune 100 company CEOs are Black, and fewer than 4 percent of leaders with responsibility for
departments’ profit and loss—roles that tend to accelerate career progression—are Black.43 Moreover,
unconscious bias can affect mentorship and sponsorship, because humans tend to gravitate toward
mentoring people they view as similar to themselves.44
Community programs can help Black entrepreneurs connect with role models and commercial networks
that can help Black prospective entrepreneurs pursue business ownership with more confidence and
support. For instance, the private and social sectors could facilitate networking and partnerships
between established businesses and compatible start-ups.
14 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 166 of 801
Conclusion
An investment in more business ecosystems that provide Black business owners equitable access to
resources and opportunities can unlock part of the $1 trillion to $1.5 trillion in annual GDP that would come
from closing the racial wealth gap. More important, by making social and economic institutions supportive
of a wider swath of people, stakeholders can rectify the mistrust that has developed between Black
entrepreneurs and institutions. That restorative effect, along with economic gains, could be a step forward
for US society.
David Baboolall and Nina Yancy are consultants in McKinsey’s New York office; Kelemwork Cook is a consultant in the Cleveland
office; Nick Noel is a consultant in the Washington, DC, office; and Shelley Stewart is a partner in the New Jersey office.
The authors wish to thank Katie Chen, Michael Chui, JP Julien, Mike Kerlin, Michael Lazar, Ricardo Pena, and Duwain Pinder for
their contributions to this article.
Copyright © 2020 McKinsey & Company. All rights reserved.
Building supportive ecosystems for Black-owned US businesses 15
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 167 of 801
Endnotes
1 Nick Noel, Duwain Pinder, Shelley Stewart, and Jason Wright, “The economic impact of closing the racial wealth gap,”
August 13, 2019, McKinsey.com.
2 Our estimate of the wealth-building opportunity is based on an additional $200 billion per year in revenue for Black-
owned businesses that attain revenue parity. We assumed that US private companies’ enterprise value is 2.54 times
annual revenue and applied a debt-to-equity ratio of 0.43. These assumptions translate the $200 billion revenue
increase to $290 billion in additional business equity. André Dua, Deepa Mahajan, Ingrid Millán, and Shelley Stewart,
“COVID-19’s effect on minority-owned small businesses in the United States,” May 27, 2020, McKinsey.com; Electronic
Code of Federal Regulations, Part 121—Small Business Size Regulations, October 5, 2020, sba.gov.
3 For more on the effects of the COVID-19 pandemic on Black Americans, see Aria Florant, Nick Noel, Shelley Stewart, and
Jason Wright, “COVID-19: Investing in Black lives and livelihoods,” April 14, 2020, McKinsey.com.
4 Claire Kramer Mills and Jessica Battisto, “Double jeopardy: COVID-19’s concentrated health and wealth effects in Black
communities,” Federal Reserve Bank of New York, August 2020, newyorkfed.org.
5 Robert Fairlie, “The impact of COVID-19 on small business owners: Evidence of early-stage losses from the April 2020
current population survey,” Stanford Institute for Economic Policy Research, working paper 20-022, May 2020, siepr.
stanford.edu.
6 McKinsey & Company COVID-19 US SMB Financial Pulse Survey, n = 1,004; responses collected May 8–13, 2020, from
small and medium-size businesses with less than $500 million in annual revenue.
7 McKinsey Survey of Consumer Finances, 2019.
8 For more on COVID-19’s effect on minority-owned small businesses, see Andre Dua, Deepa Mahajan, Ingrid Millán,
and Shelley Stewart, “COVID-19’s effect on minority-owned small businesses in the United States,” May 27, 2020,
McKinsey.com.
9 For more on the racial wealth gap, see Nick Noel, Duwain Pinder, Shelley Stewart, and Jason Wright, “The economic
impact of closing the racial wealth gap,” August 13, 2019, McKinsey.com.
10 “Frequently asked questions,” US Small Business Administration Office of Advocacy, October 2017, sba.gov.
11 For more on what constitutes an SMB and for a data set on US SMBs, see “Size standards tools,” US Small Business
Administration, sba.gov; and “Frequently asked questions,” US Small Business Administration Office of Advocacy,
August 2017, sba.gov.
12 The same systemic obstacles affect Black entrepreneurs who lead potentially high-growth start-ups, which merit their
own future discussion.
13 A business ecosystem is a network of entities that collaborate and compete to produce a product or service.
14 2016 annual survey of entrepreneurs, US Census Bureau, census.gov.
15 Sheryl Sandberg, “How we’re supporting Black-owned businesses,” Facebook, August 19, 2020, fb.com.
16 Chauncey Alcorn, “Citi is spending $1.15 billion to help close America’s racial wealth gap,” CNN Business, September 23,
2020, cnn.com.
17 2012 survey of business owners, US Census Bureau, census.gov; 2016 annual survey of entrepreneurs, US Census
Bureau, census.gov.
16 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 168 of 801
18 2012 survey of business owners, US Census Bureau, census.gov; 2016 annual survey of entrepreneurs, US Census
Bureau, census.gov.
19 2017 American community survey, US Census Bureau, census.gov; Leading States Index, 2018, McKinsey.com.
20 For more, see Nick Noel, Duwain Pinder, Shelley Stewart, and Jason Wright, “The economic impact of closing the racial
wealth gap,” August 13, 2019, McKinsey.com.
21 Rhett Morris, “Mentors are the secret weapons of successful startups,” TechCrunch, March 22, 2015, techcrunch.com.
22 2016 small business credit survey: Report on minority-owned firms, Federal Reserve, November 30, 2017,
fedsmallbusiness.org. Minority population includes Black, Hispanic, and Asian-American business owners. “Based on
self-reported business credit score or personal credit score, depending on which is used to obtain financing for their
business. If the firm uses both, the highest risk rating is used. ‘Low credit risk’ is an 80–100 business credit score or
720+ personal credit score.”
23 What resources do families have for financial emergencies? The role of emergency savings in family financial security,
Pew Charitable Trusts, November 2015, pewtrusts.org; and Lisa J. Dettling et al., “Recent trends in wealth-holding
by race and ethnicity: Evidence from the survey of consumer finances,” Federal Reserve Board, September 27, 2017,
federalreserve.gov.
24 Venture networks can be especially geographically concentrated. See also Michael S. Barr, Minority and women
entrepreneurs: Building capital, networks, and skills, Policy Proposal 2015-031, The Hamilton Project, March 9, 2015,
ww.hamiltonproject.org.
25 Global Entrepreneurship Monitor, 2017, gemconsortium.org; Dow Jones VentureSource, 2018.
26 Sponsored by a major US financial institution, the focus groups were conducted in May 2019. Researchers recruited both
Black and non-Black entrepreneurs in Atlanta, Detroit, New York, and Oakland. The entrepreneurs’ reported business
revenues were $100,000 to $1 million.
27 Brian Headd, “Redefining business success: distinguishing between closure and failure,” Small Business Economics,
Volume 21, Number 1, August 2003, pp. 51–61, springer.com.
28 Robert Fairlie, Alicia Robb, and David T. Robinson, “Black and white: Access to capital among minority-owned startups,”
Stanford Institute for Economic Policy Research, working paper 17-003, February 2017, siepr.stanford.edu.
29 McKinsey & Company COVID-19 US SMB Financial Pulse Survey, n = 1,004; responses collected: May 8–13, 2020, from
small and medium-size businesses with less than $500 million in annual revenue.
30 Alicia Robb and Arnobio Morelix, Startup financing trends by race: How access to capital impacts profitability, Ewing
Marion Kauffman Foundation, October 2016, kauffman.org.
31 Sterling A. Bone et al., “Shaping small business lending policy through matched-pair mystery shopping,” Journal of
Public Policy & Marketing, February 26, 2019, journals.sagepub.com.
32 Timothy Bates and Alicia Robb, Loan availability among small businesses operating in urban minority communities,
Reserve System Community Development Research Conference, April 11, 2013. frbatlanta.org.
33 Rachel S. Shinnar, Michael B. Aguilera, and Thomas S. Lyons, “Co-ethnic markets: Financial penalty or opportunity?”
International Business Review, Volume 11, Issue 6, December 2002, pp. 646–58.
34 2016 Annual Survey of Entrepreneurs, “Providers of Business Advice or Mentoring,” US Census Bureau, census.gov.
For more on Black entrepreneurs’ higher levels of mistrust, see “The tapestry of Black business ownership in America:
Untapped for success,” Association for Enterprise Opportunity, 2016, aeoworks.org.
35 Robb and Morelix, Startup financing trends by race: How access to capital impacts profitability; Transparent Collective,
transparentcollective.com.
Building supportive ecosystems for Black-owned US businesses 17
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 169 of 801
36 For more on Black professionals’ exclusion from information networks, see Nick Noel, Duwain Pinder, Shelley Stewart,
and Jason Wright, “The economic impact of closing the racial wealth gap,” August 13, 2019, McKinsey.com.
37 “Contracting barriers and factors affecting minority business enterprises: A review of existing disparity studies,” Minority
Business Development Agency, mbda.gov.
38 2018 Annual Business Survey, US Census Bureau, census.gov. For more on inclusion in the financial services sector, see
Aria Florant, JP Julien, Shelley Stewart, Nina Yancy, and Jason Wright, “The case for accelerating financial inclusion in
black communities,” February 25, 2020, McKinsey.com.
39 “Supplier diversity at RBC,” Royal Bank of Canada, rbc.com.
40 “The CARES Act provides assistance to small businesses,” US Department of the Treasury, treasury.gov.
41 Cameron Costa, “Minority entrepreneurs at a tipping point as Black-owned banks dwindle in the U.S.” CNBC, August 25,
2020, cnbc.com.
42 “Verizon small business recovery fund,” Local Initiatives Support Corporation, 2020, lisc.org.
43 David F. Larcker and Brian Tayan, Fortune 100 C-suite organizational charts, Corporate Governance Research Initiative,
Stanford Graduate School of Business, February 2020, gsb.stanford.edu.
44 Alexandra E. Petri, “When potential mentors are mostly white and male,” Atlantic, July 7, 2017, theatlantic.com.
18 Building supportive ecosystems for Black-owned US businesses
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 170 of 801
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 171 of 801
Copyright © McKinsey & Company. All rights reserved.
McKinsey.com
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 172 of 801
Exhibit F
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 173 of 801
NBER WORKING PAPER SERIES
BLACK AND WHITE:
ACCESS TO CAPITAL AMONG MINORITY-OWNED STARTUPS
Robert W. Fairlie
Alicia Robb
David T. Robinson
Working Paper 28154
http://www.nber.org/papers/w28154
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
November 2020
We are grateful for comments and suggestions from Pat Bayer, Elijah Brewer, Scott Frame, John
Graham, Melinda Petre, Amit Seru, Per Stromberg, participants at the AEA meetings, the Society
for Government Economists meetings, the CESifo Conference on Entrepreneurship and
Economics, the Federal Deposit Insurance Corporation, Federal Reserve Bank of Cleveland and
Kauffman Foundation Conference on Entrepreneurial Finance, the International Conference on
Panel Data, the APPAM meetings, as well as seminar participants at Illinois, IFN Stockholm,
McGill University, Stockholm University, Stockholm School of Economics, University of British
Columbia, University of Melbourne, University of Southern California, and Vanderbilt
University. The views expressed herein are those of the authors and do not necessarily reflect the
views of the National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been
peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies
official NBER publications.
© 2020 by Robert W. Fairlie, Alicia Robb, and David T. Robinson. All rights reserved. Short
sections of text, not to exceed two paragraphs, may be quoted without explicit permission
provided that full credit, including © notice, is given to the source.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 174 of 801
Black and White: Access to Capital among Minority-Owned Startups
Robert W. Fairlie, Alicia Robb, and David T. Robinson
NBER Working Paper No. 28154
November 2020
JEL No. J15,J71,L26
ABSTRACT
We use confidential and restricted-access data from the Kauffman Firm Survey and matched
administrative data on credit scores to explore racial disparities in access to capital for new
business ventures. The novel results on racial inequality in startup financing indicate that black-
owned startups start smaller and stay smaller over the entire first eight years of their existence.
Black startups face more difficulty in raising external capital, especially external debt. We find
that disparities in credit-worthiness constrain black entreprenuers, but perceptions of treatment by
banks also hold them back. Black entrepreneurs apply for loans less often than white
entrepreneurs largely because they expect to be denied credit, even when they have a good credit
history and in settings where strong local banks favor new business development.
Robert W. Fairlie David T. Robinson
Department of Economics Fuqua School of Business
Engineering 2 Building Duke University
University of California at Santa Cruz 100 Fuqua Drive
Santa Cruz, CA 95064 Durham, NC 27708
and NBER and NBER
rfairlie@ucsc.edu davidr@duke.edu
Alicia Robb
333 18th Street
Boulder, CO 80302
alicia@nextwaveimpact.com
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 175 of 801
1 Introduction
More than half a century after the passage of the Civil Rights Act, economic differences
between whites and African-Americans continue to be a source of social and political
tension in the United States. Median black and white households live under substantially
different economic circumstances. For example, the median household income for black
families is $37,000; for white families the number is $63,000. One out of four black families
live in poverty; the poverty rate for white families is 9 percent (U.S. Census Bureau 2016).
Inequality is even higher for wealth and financial assets. For example, the ratio of median
household net worth for black families to that of white familes is 11 to 1, and only 7
percent of black families own stocks or mutual funds compared with 23 percent of white
families (U.S. Census Bureau 2019).
Entrepreneurship is often viewed as a mechanism for promoting economic mobility,
wealth accumulation and job creation in minority communities, representing a poten-
tial tool for alleviating these racial disparities (Bradford and Osborne 1976; Borjas 1999;
Boston, 1999, 2006; Bradford 2003). Yet, access to financial capital is a critical element of
new business formation (Kerr and Nanda 2011; Simoes et al. 2016). This paper explores
racial differences in capital market outcomes associated with launching a new businesses.
Although previous research provides evidence that established minority-owned firms ex-
perience higher loan denial probabilities, we know little about the racial differences in
financing that occur when firms are initially started.1 To our knowledge, our analysis is
the first to provide a detailed analysis of race, financing, and creditworthiness at the time
a business is first launched.
To explore these racial differences, we use the confidential, restricted-access version of
the Kauffman Firm Survey (KFS) with matched administrative data on credit scores. The
KFS is the only dataset that provides panel data on startups with detailed information
on financing outcomes, credit worthiness and credit expectations. The panel structure of
1
See Bostic and Lampani 1999, Cole 1999, Cavalluzzo, Cavaluzzo and Wolken 2002, Blanchflower,
Levine and Zimmerman 2003, and Blanchard et al. 2008 for evidence of racial disparities in loan outcomes
among established businesses in the SSBF. See Bates (1989, 1991) for evidence of racial disparities in total
capital and loan outcomes from a sample of businesses started in the past 6 years in the 1987 CBO.
1
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 176 of 801
the KFS allows us to focus on both the initial capital that firms receive in their founding
year, as well as later capital injections secured over the firm’s next seven years of oper-
ations. Ultimately this allows us not only to measure initial differences, but also study
whether any differences diminish or persist over time as a startup builds an observable
track record of performance.
The panel structure of our data offers two key advantages relative to previous work
on racial differences in funding, which has focused on cross-sectional differences in firms
that are already operating. First, it allows us to avoid problems with restropective ques-
tion recall bias and survival bias found with cross-sectional data. Second, if we restrict
our analysis to the sample of firms that survive eight years, the initial racial differences
in startup capital are considerably smaller than if we look at the full sample, which in-
cludes firms that do not survive. This suggests that conditioning on survival understates
the degree of racial differences in access to capital. Previous research has highlighted the
differences in rates of job creation, responsiveness, and growth between young firms and
small firms (Hurst and Pugsley, 2011; Haltiwanger, Jarmin and Miranda, 2013; Adelino,
Ma and Robinson, 2016). Previous research showing racial differences in capital access in
a cross-section of established businesses could be attributed to racial differences in human
capital that have played out over time, inducing sorting of minority-owned businesses
into low-growth industries where small firms are the dominant mode of organization.
Our work instead demonstrates that there are within-industry, within-geography differ-
ences in access to capital at firm inception, which may have important implications for
understanding racial differences in regional economic growth and employment.
Our analysis proceeds in three steps. In the first step we demonstrate large racial dif-
ferences in the sources and amounts of financial capital that are used to launch businesses.
Black-owned startups start smaller in terms of overall financial capital and invest less on
average as they mature. Racial differences in outside debt account for more than half of
the difference in total financial capital. Indeed, the ratio of debt to total capital (i.e., the
leverage ratio) for black-owned startups is persistently below that observed for white-
2
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 177 of 801
owned startups. But, the disparities do not end here: alternative sources of capital such
as loans from friends and family, personal equity and credit cards do little to attenuate
these differences. Black-owned startups also have lower levels of all other major sources
of funding than do white-owned startups.
The second step is to explore the underlying causes of these financing patterns. Through-
out the paper, we use the term access to capital to capture the amount of capital obtained by
a particular business, understanding that this quantity is an equilibrium capital market
outcome affected by both supply-side and demand-side factors. Large racial disparities
in access to capital could reflect racial differences in either demand for capital, in the
underlying quality of the business opportunity, or in attitudes towards credit markets.
Under these demand-side explanations, black borrowers obtain less capital because they
need or want less, because they are more risk averse (perhaps the stigma of bankruptcy
affects them more greatly) or because they anticipate rejection when they apply for credit.
There are also supply-side channels, through which race matters to lenders. A long lit-
erature in economics going back to Becker (1971) and Phelps (1972) debates whether this
ultimately traces back to taste-based discrimination rooted in racial animus or instead sta-
tistical discrimination based on differences in endowments and incomplete information.
Under both sets of explanations the race of the borrower affects the level of capital they
receive.
Although we cannot definitively rule out any particular explanation, our data allow
us to paint a rich descriptive picture of racial differences in access to capital by exploring
these potential explanations in considerable detail. First, because we have new confiden-
tial administrative data on credit ratings from Dun & Bradstreet that have been matched
to all businesses in the restricted-access version of the KFS, as well as information on
founder net worth, we can condition on an extensive set of founder and business char-
acteristics that are correlated with race, and likely affect lending decisions. Thus, we can
identify key traits contributing to inequality and can examine whether correlated traits
are the primary source of racial disparities. After we control for industry, business credit
3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 178 of 801
scores, founder net worth, education and experience, as well as many business charac-
teristics that may ultimately be endogenous to the amount of funding received, we can
explain about one-third of the initial funding gap between black-owned and white-owned
startups. Lower credit scores among black startups contribute the most among correlated
traits.
Nevertheless, as is common with much work that attempts to explain firm-level differ-
ences in capital structure, our analysis cannot fully account for the unobserved differences
in opportunity sets that might drive firm-level differences in borrowing. Including fixed
effects for business location dramatically increases the explanatory power of our regres-
sions, but does little to alter the estimated differences between black- and white-owned
businesses in initial size. That is, including a fixed effect for the core-based statistical
area (CBSA) of the business raises the regression R2 from around 15% to around 30%,
and specifications including zipcode fixed effects produce R2 values over 60%.2 In short,
racial differences across neighborhoods within the same city are as large as racial differ-
ences across cities.
Moreover, the initial differences in funding are not erased by later injections of cap-
ital. In order for black- and white-owned businesses to converge in size, black-owned
businesses would need substantially larger capital injections in the years after inception
to make up for differences at founding. Racial differences in the size of later injections of
new funding are smaller than the initial differences, but they remain significantly smaller
in later years. Thus, on average, businesses started by black founders do not converge to
the size of white-owned businesses as they age.
This persistent difference in funding is driven primarily by differences in the amount
of bank loans and other bank credit products, which in turn are not substituted by other
sources of capital. Lower amounts of banking services could reflect worse treatment by
banks, less demand for banking services, or could reflect differences in borrower attitudes
2
Core-based statistical areas include metropolitan statistical areas (MSAs) but also include “micropoli-
tan” statistical areas, defined by the US Census as “areas that have at least one urban cluster of at least
10,000 but less than 50,000 population, plus adjacent territory that has a high degree of social and economic
integration with the core as measured by commuting ties.”
4
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 179 of 801
and expectations, and ultimately less willingness to approach banks. One important ad-
vantage of our data is that they include measures of loan application expectations, even
among those who did not seek funding. Typically, differential average participation rates
confound the measurement of discrimination; here, detailed questions in the KFS mea-
suring the demand for loans, the rate of loan rejections, and the expected fear of denial
among borrowers who chose not to attempt to borrow allow us to explore how expecta-
tions of discrimination may impact participation in financial markets.
Black entrepreneurs apply for bank loans less frequently than white entrepreneurs.
This stems largely from differences in the fear of rejection. Overall, black entrepreneurs
are about three times more likely to state that they did not apply for credit when needed
for fear of having their loan application denied. Similarly, black-owned startups are about
three times less likely than white-owned startups to report that their loan requests are
always approved. These differences persist even after controlling for credit scores and
net worth: indeed, even black founders in the top quartile of the credit score distribution
are more than twice as likely to report a fear of denial than white founders with below
median credit scores. These effects are stronger in areas where historical and current
racial tension is higher, and weaker in areas where racial tensions are less severe, which
suggests that actual or perceived statistical or taste-based discrimination could be a factor
in these results.
Banks use both hard information (objective, easily codified and transmitted informa-
tion like credit scores) and soft information (potentially more precise, but subjective and
difficult to verify information) in their lending decisions, and to varying degrees based
on bank characteristics. Because black-owned startups tend to be at a hard-information
disadvantage relative to white-owned startups, we next explore whether they face fewer
constraints in settings where soft information is potentially more actionable. Given that
large national banks tend to rely more on hard information when making lending deci-
sions, while local banks tend to rely more on soft information (Berger, Miller, Petersen,
Rajan, and Stein, 2005; Petersen and Rajan, 2002), we exploit regional variation in the
5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 180 of 801
strength of local banks to ask whether these attitudes and outcomes are different in re-
gions where soft information could play a bigger role in the lending decision. Areas with
stronger local banks are indeed areas where the average founder is less afraid of loan de-
nial, and where average business loan amounts to startups in our sample is higher. But
these effects are exclusive to white-owned startups. Black founders are not less afraid of
loan denial in these markets; if anything, they are somewhat more likely to report that
they did not apply for fear of denial in regions with stronger local banks. In these areas,
white-owned startups receive larger amounts of bank debt on average but black-owned
startups do not.
The third step and final piece of our analysis attempts to assess the importance of
these differences to cumulative capital disparities. For this, we use decomposition tech-
niques developed by Blinder (1973) and Oaxaca (1973) to assess how much of the size
difference between black-owned and white-owned businesses is attributable to the char-
acteristics we observe. We can explain around one-half of the total difference in firm size
with observables. Of these, business credit scores and founder net worth (which presum-
ably measures collateral) account for about two-thirds of the difference. Differences in
education and experience account for only a modest portion of the difference. If the aver-
age black-owned business had the observable characteristics of the average white-owned
business, it would be about 75% larger.
This paper adds to the literature on racial differences in financial market outcomes.
Chatterji and Seamans (2012) find that the expansion of credit card availability stimulated
entry into entrepreneurship especially for black entrepreneurs, and find that the strongest
results in areas with high rates of historical racial discrimination. Dougal, Gao, Mayew
and Parsons (2017) find that historically black colleges pay higher issuing costs for bonds
than other higher-education bond issuers, and attribute these higher spreads to racial
animus among wealthy white bond purchasers. Earlier studies provide cross-sectional
evidence from the SSSBF of racial differences in lending markets for established busi-
nesses (Bostic and Lampani 1999; Cole 1999; Cavalluzzo, Cavaluzzo and Wolken 2002;
6
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 181 of 801
Blanchflower, Levine and Zimmerman 2003; Blanchard et al. 2008; Mitchell and Pearce
2011). Apart from our study being the first to focus on new business ventures, rather than
more-established, existing businesses, our work departs from earlier work in the breadth
and depth of our empirical measures of overall capital sources, creditworthiness and loan
expectations, and use of longitudinal data on a cohort of firms.
The balance of the paper is organized as follows. In Section 2, we describe the restricted-
access KFS panel that follows startups from their founding through seven years of oper-
ations after their startup year and the matched Dun and Bradstreet (D&B) administra-
tive data on credit scores. In Section 3, we examine whether there are differences in the
use of financial capital (levels and detailed sources) between black and white firms at
startup and in the years following startup. Section 4 explores potential causes of racial
differences in financial capital. In Section 5, we explore racial differences in credit mar-
ket explanations. Section 6 explores the potential role of racial bias in capital markets,
and Section 7 explores the question of how much of the racial gap in funding disappears
after controlling for startup characteristics. Section 8 concludes. An online appendix pro-
vides additional details regarding racial differences in survival, profitability, and funding
sources.
2 The Kauffman Firm Survey
We use the confidential, restricted access version of the Kauffman Firm Survey (KFS)
to study how startups access capital markets. The KFS is a longitudinal survey of new
businesses in the United States, collecting annual information for a sample of 4,928 firms
that began operations in 2004. The underlying sample frame for the KFS is Dun and
Bradstreet (D&B) data.
The KFS data contain unprecedented detail on the financing patterns of startups, as
well as detailed information on both the firm itself and up to ten business owners of the
firm. In addition to the 2004 baseline year data, we also use the seven years of follow
up data covering calendar years 2005 through 2011. Detailed information on the owners
7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 182 of 801
includes race, gender, age, education, previous startup experience, and previous work ex-
perience. Detailed information on the firm includes industry, physical location, employ-
ment, sales, intellectual property, and financial capital used at start-up and over time. The
detailed financing information in the KFS allows us to examine the relative importance
of each source of financing at start up and over time. The confidential, restricted-access
version of the KFS includes credit scores, continuous measures of key variables, such
as financing, and more detail on industries and geographic locations than the publicly-
available KFS. The KFS was also designed using sample weights to be representative of
all new businesses in the U.S. economy and not restricted to a narrow set of industries or
business types.
Our administrative data on credit scores from D&B for all firms in the KFS allows for
a novel analysis of racial differences among startups. Credit scores are not available on
most surveys, perhaps because most entrepreneurs do not know readily know what their
scores are. To be sure, the Survey of Small Business Finances (SSBF) includes information
on credit scores, but only for larger, more established, and older businesses (Cavalluzo
and Wolken 2005). While the KFS contains unprecedented detail on the business for-
mation process, the availability of business credit scores allows us to control for many
differences in firm characteristics that would be observable by bank lending personnel
but typically unobservable to the econometrician.
The KFS is the only large, nationally representative, longitudinal dataset providing
detailed information on new firms and their financing activities. Most previous research
on the use of financial capital among small businesses has relied on cross-sectional data
on existing businesses. For example, the Survey of Business Owner (SBO) data provide
information on the amount of startup capital, but provide only retrospective information
for surviving businesses and do not provide information on the relative importance of
the different sources of financing. Another commonly-used dataset, the Federal Reserve
Board’s Survey of Small Business Finances (SSBF), provides information on recent financ-
ing, but does not provide information on financing at startup or the early stages of firm
8
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 183 of 801
growth (and was discontinued after 2003). Furthermore, both the SBO and the SSBF are
cross sectional surveys that do not provide information on firm financing over time for
the same sets of firms. Finally, fundraising levels in the KFS are measured annually, and
are thus less prone to recall bias as is the case with both the SBO and the SSBF.
We restrict our attention to the set of firms that either survived over the sample pe-
riod or that have been verified as going out of business over the sample period. In most
analyses, we condition on survival in that year, but we also conduct robustness checks
taking alternative approaches to addressing survival. Our main results are not sensitive
to the approach, and we discuss the robustness check results below. We also specifically
focus on firms that have a white or black primary owner. These restrictions result in a
sample of 3,551 startups out of the total sample of 4,124 startups with owners of any race
that began operations in 2004 and either continued through the final year in the sample
period (2011) or can be verified to have exited sometime over the period.
We assign owner demographics at the firm level based on the primary owner. For
firms with multiple owners (35 percent of the sample), the primary owner is designated
by having the largest equity share in the business. In cases where two or more owners
owned equal shares, hours worked and a series of other variables are used to create a rank
ordering of owners in order to define a primary owner following the algorithm proposed
in Ballou et al (2008). We include businesses with owners of all races in the regression
analysis, but focus our comparisons on black- and white-owned businesses. Following
standard conventions in the literature, the white category includes only non-Hispanic
whites. Using these definitions, we find that 9.1 percent of the KFS sample of startups is
black-owned. The percentage of black-owned startups does not notably change over time
indicating similar survival rates. In the seventh year after startup we find that 8.4 percent
of the KFS sample is black-owned.
Because so much of our analysis centers around founder net worth and creditworthi-
ness, we also compare the distribution of net worth among startup owners in our data
with that of the broader U.S. population as a whole. The most recent government source
9
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 184 of 801
for data on U.S. net worth is from the 2013 Survey of Income and Program Participation
(SIPP). Figure 1 compares the KFS and SIPP net worth distributions. Solid bars represent
the U.S. population and dotted bars represent startup owners. The bottom two quartile
categories are collapsed because of reporting restrictions. Additionally, the quartiles are
inexact due to data availability in the published net worth statistics from SIPP.
Insert Figure 1 here
There are two key findings here. First, both black and white owners have net worth
distributions to the right of their respective population net worth distributions. Thus,
both black and white startup owners are less likely to be from the lower tail of the wealth
distribution than the population as a whole. Second, the wealth disparity between whites
and blacks found in the overall U.S. population also holds among startup owners. Black
startup owners have a wealth distribution to the left of the white startup owners distri-
bution, and the same holds for the U.S. population.
3 Are There Racial Differences in Access To Startup Capital?
Table I reports average amounts of capital by type of capital for startups (and Figures
1 and 2). The KFS contains finely detailed sources of funding for startups, which are
reported along with summary statistics in Appendix Table I. To facilitate an analysis of
broad patterns in the data, in most of our analysis we follow Robb and Robinson (2014)
and group the detailed categories into six broad buckets based on the source of capi-
tal and the structure of the capital (reported in Table I). The three alternative sources of
capital are owners, insiders, and outsiders; the two alternative types of capital are debt
and equity. The distinction between sources captures whether the funding source is the
founder, informal channels such as friends or close associates of the founder who are not
direct owners of the business, or formal channels such as banks, venture capital firms,
and angel investors. Robb and Robinson (2014) make distinctions along these lines be-
cause the personal balance sheets of business owners and the balance sheets of the firms
10
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 185 of 801
themselves are often deeply intertwined at the time the business is founded, and there-
fore there is little practical distinction between, for instance, a business credit card and a
personal credit card, or a personal bank loan and a business bank loan.
Insert Table I Here
In the initial year of the KFS, black-owned startups are started with substantially less
capital than white entrepreneurs. The average level of startup capital among black en-
trepreneurs is $35,205 compared with $106,720 for white entrepreneurs. Racial differences
in the sources of capital are also pronounced. In the year the business is founded, black
owners contribute around $19,500 of personal equity, compared with around $34,500 for
white business owners. Inside equity–equity stakes taken by family members or other
business insiders–are relatively modest for both groups, but are about five times larger
for white-owned than black-owned startups.
Differences in outside equity—venture capital, angel financing, and the like are even
more stark. The average black-owned startup has around $500 of outside equity, whereas
the average white-owned business has more than $18,500 from outside equity at found-
ing. These numbers are a reflection of the fact that while outside equity is relatively
uncommon for white-owned businesses, it is exceedingly rare for black-owned startups.
Owner debt includes personal loans extended to the business by the founder. These
are small on average for both black-owned and white-owned firms, but white-owned
businesses have higher average amounts here as well, by a factor of five in the initial
year. Inside debt–money lent to the firm by family members or business insiders–is about
the same order of magnitude as owner debt, although there is no statistically significant
difference across racial groups.
The largest quantitative difference between white- and black-owned startups is in the
amount of outside debt associated with their businesses. Outside debt includes personal
loans, business loans, personal and business credit cards, as well as other types of loans
made by banks either directly to business owners for the purpose starting their business
or else to the business itself. Robb and Robinson (2014) show that on average, this is
11
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 186 of 801
the largest source of financing for firms in the KFS. Here, we see that this is only true
of white-owned firms. At startup, black-owned firms borrow about one-half as much
as they put in of their own capital, whereas white-owned firms borrow about 1.7 times
what they put in of their own capital. In the year of founding, white-owned firms on
average borrow nearly six times as much black-owned firms. Although the amount of
outside debt accessed by black-owned startups grows steadily over time, average outside
debt for black-owned startups is substantially lower than that seen among white-owned
firms.
Insert Figure 2 here
In the later years of the survey, there is significant convergence in the average amounts
of personal equity injected into the business, but this largely reflects the fact that personal
equity injections from white startup owners dramatically decline in the years after found-
ing: the average amount drops to around $11,000 in years 1-3 after startup and to around
$4,000 by years 4-7 after startup on average for white-owned businesses. On average,
insider equity (that is, equity injections from friends, family or other non-business owner
acquaintances) is a negligible source of financing for most firms after founding, and the
differences between white- and black-owned startups is not statistically significant. In-
deed, across most of the individual categories, differences in new capital cease to be sta-
tistically different after the initial founding year. Because these numbers track new dollars
coming into the firm, however, this means that the accumulated difference in size grows
over time.
Insert Figure 3 here
In the appendix, we dig deeper into the differences in access to debt for minority
and white-owned startups by looking at the specific sources of debt financing. This is
presented in Table A.4 digs. In the founding year, there are differences between black and
white owned businesses across a wide array of debt sources. Only one percent of black
owners obtain business loans, compared with 7% for white-owned firms. While 30% of
12
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 187 of 801
white-owned businesses use business credit cards in their founding year, only 15% of
black owned businesses do. Similarly, 18% of white business owners rely on personal
loans for their business in the founding year, while only 14% of black-owned startups do.
All these differences are statistically significant.
What sources offset these differences? As we show in the Appendix, it is not the
case that black-owned startups rely more on personal credit cards. In fact, the opposite
is true. Instead, black-owned startups appear to rely more on informal borrowing from
family members: 14% of black-owned startups relied on family loans in their founding
year, while only 9% of white-owned businesses do. Interestingly, the average amounts
borrowed from family and other sources are not statistically different between minority
and non-minority businesses. This could be a reflection of liquidity constraints in the
network of family members that are stronger for black-owned startups than for white-
owned firms (Fairlie and Robb 2008). Average amounts of capital from personal bank
loans and business bank loans are statistically smaller for black-owned startups. Black-
owned startups continue to rely on family loans to a greater degree than white-owned
firms in the three years following the firm’s founding. This suggests that access to formal
debt channels remains limited for minorities.
All told, the descriptive evidence thus far indicates that black-owned startups access
less formal credit. It suggests that they partially substitute for this with a heavier re-
liance on informal channels and personal equity, but this substitution is an imperfect one
(perhaps due to lower levels of personal and family wealth). This results in businesses
that start with smaller amounts of financial capital and that do not converge over time.
To illustrate this, Figure 4 reports average firm size, for all firms as well as white- and
black-owned firms, over time from startup to seven years after startup.
4 What Explains Racial Differences in Access to Capital?
In this section, we investigate the causes of racial inequality in financial capital reported in
the previous section. We focus on the question of whether credit scores, and other founder
13
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 188 of 801
and business characteristics limit the ability of black startups to obtain comparable levels
of financial capital as white startups. We first examine differences in access to capital in
the firm’s initial year, then examine differences as the startup ages.
4.1 Differences in Initial Capital
We begin by examining the difference in total capital raised across all sources. Given its
importance, we then turn to examining differences in the amount of outside debt. The
final step is to examine differences in business bank loans.
4.1.1 Total Financial Capital
Table II models variation in the natural log of the total amount of capital (from all sources)
in the startup year based on race, owner characteristics and business characteristics. In-
dustry fixed effects at the two-digit NAICS level are included in all specifications to cap-
ture general differences in capital levels based on types of businesses started. The inclu-
sion of industry fixed effects partly addresses the concern that black and white businesses
differ in their need for capital because they cluster in industries with different capital re-
quirements.
In column (1) we report the baseline specification, which includes only a dummy for
the race of the founder and industry fixed effects with no additional controls. The loading
on the black-owned startup dummy variable illustrates that black-owned startups have
total capital investments that are roughly 0.73 log points lower in terms of initial total
capital than white-owned businesses.
Insert Table II Here
The remaining columns of the table in some sense seek to explain away this differ-
ence with a variety of control variables. Including the credit score lowers the loading on
the black-owned startup dummy from -0.73 to -0.60. Credit scores are much lower among
black startups than white startups, and the loading on the credit score indicates that credit
14
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 189 of 801
scores have a large positive effect on the amount of capital raised.3 We find that moving
up 10 percentile points in the credit score distribution is associated with an increase in fi-
nancial capital by roughly 20 percent. These results are consistent with previous research
focusing on larger, established businesses, which finds that credit scores have a negative
effect on loan denial rates (Cavalluzzo and Wolken 2005). But, even after controlling for
credit scores, the black indicator estimate remains large and statistically significant.
In Column (3) we introduce founder net worth. Although founder net worth is not
available in the survey until the fourth followup year, we rely on the high persistence
in net worth, especially as measured categorically in the KFS. We treat this as a proxy of
owner’s net worth in the startup year and note some caution in interpreting the estimates.
The net worth categories included in Column (3) indicate that high net worth individuals
launch businesses at a much larger scale than others. Controlling for net worth attenu-
ates the loading on the black-owned startup indicator variable but does not diminish its
statistical significance.
Next we include measures of formal education (in the form of dummy variables for
levels), prior work experience to starting the business (both industry specific and non-
industry specific), and previous entrepreneurial experience. These are included in Col-
umn (4), and capture the human capital of the entrepreneur. Education and prior work
experience in the same industry have been found to be important determinants of busi-
ness success in previous research (Van Praag et al. 2005; Parker 2009). We find some
evidence that education is important, but no evidence of important effects for prior work
experience. Previous entrepreneurial experience is positively associated with capital in-
vestments, perhaps due to prior knowledge of finding capital. Rather than further erase
the difference between white-owned and black-owned startups, controlling for human
capital widens the racial difference slightly. The loading on the black-owned startup
dummy remains statistically significant in these specifications.
Columns (5) and (6) introduce a range of detailed additional controls for business
3
In unreported regressions, we tested whether credit scores had a different effect for white and black
owned startups and found no statistically significant difference.
15
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 190 of 801
type, growth goals and performance. These variables may be endogenous to the amount
of capital the firm was able to raise, but including them does not diminish the racial dif-
ference in total capital. In column (5) we add controls for firm characteristics to condition
on the fact that black and white founders may open different types of businesses with dif-
ferent capital needs. We include dummies for whether the firm sells a product or service,
whether it is based out of the founder’s home, and whether it has patents or other intel-
lectual property. In column (6) we include a dummy for whether the business is full-time
or part-time, its incorporation status, and employment level (i.e. employees). Interest-
ingly, when we control for the type of business started (i.e., whether it sells a product or
service, whether it has intellectual property, and whether it is incorporated) the effect of
prior startup experience drops in half and becomes statistically insignificant: serial en-
trepreneurs, on average, start observationally different types of businesses than first-time
entrepreneurs.
The inclusion of controls for business characteristics in Columns (5) and (6) has little
affect on the measured racial difference in startup capital, but the controls themselves in-
dicate that home-based businesses invest less capital, and product-centered businesses
and businesses with intellectual property invest more capital, as would be expected.
When we further add additional controls for firm performance and growth goals, such
as whether the business is full-time or part-time, its incorporation status, and employ-
ment level, the black-founder loading does not change. Although many of these controls
may well be endogenous, the stability of the black-owner loading across different speci-
fications suggests that remaining black/white differences in capital use are not primarily
driven by easily observable differences in firm characteristics. Moreover, the addition
of these variables does not substantially change the coefficient estimates on credit scores
and human capital measures, which suggests that credit scores are not simply proxying
for the type of business.
In the remaining two columns we attempt to control for the effect that business loca-
tion may have on demand conditions, unobservable business quality, and hence demand
16
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 191 of 801
for capital. In Column (7) we introduce Core-Based Statistical Area (CBSA) fixed effects.
CBSAs include the standard MSAs but add to them ‘micropolitan’ statistical areas, which
the Census describes as 1a new set of statistical areas that have at least one urban cluster
of at least 10,000 but less than 50,000 population, plus adjacent territory that has a high
degree of social and economic integration with the core as measured by commuting ties.’
As column (7) shows, including a CBSA fixed effect does little to change the point esti-
mates on the main initial lending outcomes. This suggests that unobserved differences
in business quality captured by coarse location measures—the difference between being
located in Duluth, Minnesota instead of Mobile, Alabama, for example—does little to
explain away the observed racial difference in startup capital.
In Column (8) we include zipcode-level fixed effects. Because this results in an ex-
tremely large number of model parameters, we cannot use the sampling weights included
in Columns (1)-(7), thus we urge caution in comparing the point estimate on the Black-
owned Startup indicator with the preceding estimates. In addition, this parameter is
only identified using survey zipcodes which contain both black and white survey re-
spondents, limiting the sample size. Nevertheless, there remains a statistically significant
racial difference in total capital. Thus, black-owned startups access less capital than their
white-owned neighbors in the same zip code.
4.1.2 Outside Debt
Given the importance of outside debt as illustrated in Section 3, we now turn to exploring
the potential causes of racial differences in access to outside debt. Exploring potential
explanations for differences in outside debt may also be useful for shedding further light
on the importance of credit scores and provide a useful consistency check on this variable.
Credit ratings are undoubtedly one of the most important pieces of information used by
banks and other financial institutions in loan determination. Table III reports regression
results, which follow the same format as Table II, except that the dependent variable is
the log of total outside debt instead of the log of total financial capital.
Insert Table III Here
17
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 192 of 801
The results for the determinants and patterns across the regression specifications for
outside debt are similar to those for total financial capital. Credit scores exert a strong
influence on the ability of businesses to find outside debt. Even controlling for an exten-
sive list of business characteristics proxying for need and ability to raise capital (i.e. make
products, intellectual property, home-based, part-time, incorporated, and employment)
the coefficient on credit scores is large, positive and statistically significant. The results
for human capital measures are also similar, with previous startup experience demon-
strating the strongest association with outside debt capital, but also some evidence of the
influence of education and work experience. Wealth is a stronger predictor of outside
debt, which may be due to the importance of personal wealth as collateral in obtaining
loans. Racial differences persist even after controlling for business location using either
coarser CBSA fixed effects or narrower zipcode fixed effects.
4.1.3 Business Bank Loans
To zero in on borrower/lender effects, we refine our analysis one step further by exam-
ining only business bank loans. Whereas total capital includes all sources of debt and
equity financing, and total outside debt includes many forms of debt (e.g. credit cards)
that do not require any interaction between a borrower and a loan officer, by studying
business bank loans separately we are honing in on the empirical setting in which there is
the greatest scope for personal interactions between the borrower and lender to influence
outcomes.
Insert Table IV Here
Table IV reports regressions of the log of business bank loans on the same set of observ-
ables that were used to explain total capital and total outside debt. The results are largely
consistent with the previous analysis, in that about 1/3 of the initial industry-adjusted
racial difference is attenuated with controls for credit score, net worth, and business char-
acteristics. The raw magnitude of the racial difference is smaller for business bank loans
than for total debt, which reflects the fact that differences in access to business bank loans
18
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 193 of 801
are not attenuated by access to other forms of outside debt. These results remain statisti-
cally significant in the presence of location fixed effects.
4.2 Differences in Capital at Later Stages
The previous tables examine racial differences in the year of founding and demonstrate
that controlling for a rich set of observable characteristics only partially removes the large
difference in funding between white-owned and black-owned startups. In Table V we
ask whether these racial differences abate over time, as startups build track records that
might help them overcome information asymmetries with lenders. We repeat the same
basic specification from Column (6) of the previous three tables, but form two groups, one
for years 1-3 and one for years 4-7 after startup. We include followup year fixed effects in
each model to absorb variation over time in access to capital.4
In Columns (1) and (2), the dependent variable is the log of business bank debt that
the business received; this is the narrowest of the three sources of capital investigated
in the preceding tables, the source of capital with the greatest scope to be influenced
by direct, personal borrower/lender interactions. In both the years 1-3 period and the
years 4-7 period, there continues to be a statistically significant difference between black-
owned and white-owned businesses in the amount of business bank debt they receive.
In terms of magnitudes, the years 4-7 point estimate is about 1/3 the size of the point
estimate in the initial survey year, meaning that the black-white funding gap persists but
is considerably smaller.
Columns (3) and (4) focus on total outside debt from all sources. In the years 1-3
sample, the point estimate is about half as large as the comparable point estimate in the
initial year, meaning that about half the black-white difference is erased over the next
4
These followup year fixed effects also capture differences in survival rates between black and white
startups. The results are not sensitive to their inclusion. We also examine the sensitivity of the results
to survival bias by conditioning the sample on including only firms surviving through the last year in
the survey (year 7 after startup). Taking this approach, we also find similar results. To push the analysis
further, we also take an approach that is in the spirit of a bounds analysis (e.g. Fairlie, Karlan and Zinman
2015). We estimate the regressions assuming as a lower bound that all non-surviving businesses would
have used zero financial capital in that year. And, as a potential upper bound we alternatively impute all
non-surviving firm observations as equal to the median level of financial capital among surviving firms.
The regression results are not sensitive to this imputation.
19
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 194 of 801
three years of the firm’s life. In the years 4-7 period, the difference between black-owned
and white-owned businesses is no longer statistically significant. The final two columns
broaden the scope further to include all forms of financial capital. Here the differences
between black-owned and white-owned businesses ceases to be statistically significant,
even in the years 1-3 sample.
Taken together, these point estimates illustrate that differences in bank lending to
black-owned and white-owned businesses persist over time, but that over time black
borrowers are able to substitute into other forms of capital. The fact that we are able to
condition on a rich set of observables means that the remaining differences are unlikely
to be explained by creditworthiness, collateral, aspects of the business operating strategy,
or the industry in which operates. It is important to recognize that the dependent vari-
able here is measured in terms of new dollars flowing in during a given survey year: it
is a measure of the flow of new capital, not the outstanding stock of capital. This in turn
means that the initial differences in funding do not dissipate; they do not converge in the
level of cumulative total capital over time. In the Appendix we provide estimations that
include zipcode-level fixed effects. These specifications produce results that are quality
similar to those presented here.
5 Do Black Borrowers Expect To Be Treated Differently
The previous section asks whether observable differences in borrower characteristics that
might be important for lenders can explain the large unconditional differences in the lev-
els of capital that white-owned and black-owned businesses receive. In this section, we
ask whether differences in attitudes and expectations about the bank borrowing expe-
rience are important for understanding differences in access to capital. To explore this
question, we use survey information in the KFS that gauges demand and unmet need for
credit among entrepreneurs.
Access to measures of attitudes towards borrowing among entrepreneurs is rare in
survey data sets, but beginning in the third followup year, the KFS included a series
20
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 195 of 801
of questions gauging borrowing intentions. The new questions ask whether the startup
business applied for a loan that year, and whether it did not apply for a loan that year
because of a fear of rejection. Among those startups that did apply, a follow-up question
asks whether they were always approved, always denied, or sometimes approved and
sometimes denied.
Racial differences in responses to these questions are analyzed in Table VI. We re-
port survey-weighted averages by minority ownership status, both for the sample as a
whole, as well as splits based on notable points in the distribution of credit scores. White
entrepreneurs are more likely to apply for loans than black entrepreneurs, which poten-
tially reflects different capital needs, but could also reflect different attitudes and expec-
tations of the loan application process. When we focus on borrowers with below-median
credit scores, there is no statistical difference in the rates of loan application, but among
above-median borrowers, loan application rates are lower for blacks than for whites.
Turning to those who did not apply for loans that year, we also study racial differ-
ences in whether they did not apply for fear of rejection in Table VI.5 There are massive
differences in fear of rejection between white and black business owners. Overall, black
business owners are about three times more likely to not apply for loans because of fear of
rejection than white business owners. This difference is highly statistically significant. Al-
though it is even more pronounced among below-median credit borrowers, even among
credit worthy borrowers we find that blacks are more than twice as likely than whites to
fear rejection. Black business owners whose credit scores are above the 75th percentile for
the entire sample are still more than twice as likely as white business owners of similar
creditworthiness to not apply for a loan for fear of having their loan application denied.
Another measure of unmet financing needs is whether loans are always approved,
always denied, or sometimes approved and sometimes denied. Here, the results mirror
those from the discussion above. Black business owners are significantly less likely to
5
Although the question, did not apply for fear of rejection, is asked of all respondents, some owners
who applied for loans might have wanted to apply for additional loans. We do not include these owners
and focus on only those firms who did not apply for a new loan for clarity. The results are unchanged if we
examine all responses to this question.
21
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 196 of 801
report that they are always approved for loans. This holds throughout the distribution of
credit scores.
A useful summary measure of whether a startup experiences unmet capital need com-
bines responses to being denied a loan application and not applying for a loan because of
fear of rejection. Affirmative answers to these two questions imply that the startup did
not obtain all of the capital it needed. Using this measure, black startups are much more
likely to face unmet need for capital than are white startups.
Taken together, these results provide further evidence that the lower levels of borrow-
ing among black-owned businesses are a reflection of unmet need, stemming at least in
part from different attitudes and perceptions of the banking process, and not simply be-
cause black-owned startups need or want less capital. But they are still unconditional in
nature; to address this, Table VII examines these findings in a multivariate setting.
Even controlling for a detailed set of firm and founder characteristics, we still observe
pronounced differences in the fear of denial and loan denial rates based on the race of
the firm founder. These findings are consistent with previous findings for larger, more
established and older businesses (i.e. SSBF data) that minority-owned firms experience
higher loan denial probabilities than white-owned businesses even after controlling for
differences in credit-worthiness and other factors (Bostic and Lampani 1999; Cole 1999;
Cavalluzzo, Cavaluzzo and Wolken 2002; Blanchflower, Levine and Zimmerman 2003;
Blanchard et al. 2008; Bates and Robb, 2014). Finally, these findings also provide evidence
that racial differences in financing patterns are not simply due to lower levels of financing
needs among black startups.
Of course, one reason why a borrower might fear denial is because they had already
received a lot of debt in prior years, so that they were near their maximum debt capacity
for the business. Thus, one reason why black founders might be fearful of borrowing
is that they had already borrowed. To explore this possibility we split the sample into
black-owned businesses and all other businesses and regressed a dummy variable for
fear of denial or denied credit on the amount of prior accumulated debt as well as the
22
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 197 of 801
same set of controls we have used throughout the preceding analysis.
The results are presented in Table VIII. Columns (1) and (2) focus on the fear of denial.
Among white-owned businesses, high levels of past borrowing are a strong predictor of
failing to apply for a loan for fear of denial. The opposite is true for black borrowers;
those with more past borrowing are less likely to indicate that they are afraid to apply.
Similarly, in Columns (3) and (4) we find that among white-owned businesses, high past
borrowing is associated with a greater likelihood of being denied credit. Among black
borrowers, there is no statistically significant relationship, and the sign of the relation is
the opposite of what we find in the white-owned sample.
These results suggest that not only are there pronounced racial differences in the fear
of loan denial, but the determinants of having this fear are a function of race. Among
white borrowers, fear of denial is correlated with remaining debt capacity: white bor-
rowers are more likely to fear denial when they have borrowed heavily in the past and
perhaps worry about perceived debt levels being too high. Among black borrowers, fear
of denial is correlated with past borrowing experience: those who have borrowed in the
past are less afraid of denial than those who have not perhaps because they perceive dis-
crimination to have declined.
6 Do Banks Treat Black Borrowers Differently
The previous sections demonstrate pronounced differences in capital access based on the
race of the business founder. This section attempts to discern whether discrimination is
the root cause of the differences we find. We do this in three steps. First, we make use of
the fact that small, local banks rely more on soft information, while larger, national banks
rely more heavily on credit scores and other types of quantifiable borrower characteristics
to examine whether differences in local banking conditions exacerbate or alleviate racial
differences. Next, we develop two measures of regional variation in the degree of racial
bias and ask how perceptions about access to capital by black borrowers vary according
to these measures. One is based on a historical measure of racial inequality, the other a
23
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 198 of 801
contemporaneous measure. Both measures provide evidence that areas where racial bias
is stronger are areas where black business founders are more likely to anticipate being
denied credit.
6.1 Do Stronger Local Banks Help?
A large literature in banking draws a distinction between soft information and hard infor-
mation. Hard information—like that contained in credit scores—is quantitative and im-
personal, and can be easily transmitted, while soft information is qualitative, and while
it may be very precise, it is difficult to communicate credibly (see Petersen and Rajan,
2002). While large, national banks have been shown to have an advantage in obtaining
hard information, small banks tend to have a comparative advantage in lending to small
businesses, which are traditionally more informationally opaque, since small banks tend
to rely more on soft information than do large banks (Berger, Cerquiero and Penas, 2014;
Berger, Miller, Petersen, Rajan, and Stein, 2005; Brickley, Link, and Smith, 2003).
In this section we ask whether racial differences in startup funding vary with the
strength of local banks. On the one hand, minority-owned businesses should face fewer
financing hurdles in areas with stronger local banks if the main source of their disadvan-
tage is that they have good ideas but little ability to signal their quality objectively. In
this case, the funding gap between black- and white-owned startups would be smaller
in areas with stronger local banks, because local banks, with their increased reliance on
soft information, would award capital to minority borrowers with good ideas but po-
tentially weaker verifiable credit history. On the other hand, a greater reliance on soft
information might create greater scope for lenders to cater to racial preferences or biases,
which could mean that black-owned businesses face greater funding challenges in en-
vironments where more objective creditworthiness criteria might receive less weight in
lending decisions.
Table IX explores these issues. In Panel A, we estimate models for did not apply for
fear of denial. Column (1) verifies the previous finding that black startups have higher
rates of fear of denial than white startups. In column (2) we add the share of county bank
24
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 199 of 801
deposits held by local banks and find that areas with higher local bank concentration are
areas in which new businesses are much less likely to report that they do not apply for
fear of denial. 6 This comports with a wide body of evidence suggesting that small, in-
formationally opaque businesses have an easier time securing bank loans in areas where
local bank concentrations are higher. Column (2) suggests that startups recognize that
they will face an easier time in markets where local banks are stronger.
Column (3) introduces an interaction term to explore whether black and white-owned
businesses experience different outcomes in high local bank concentration areas. If black-
owned startups found it easier to borrow in these markets, presumably because they
expected lenders acting on soft information to be easier to work with, then we would
expect the interaction term to be negative–their reluctance to apply for loans for fear of
denial would be attenuated in these markets.
Instead, we do not find evidence that black-owned startups receive more financing in
these markets. The interaction term is statistically significant, but has the wrong sign. Of
course, we cannot rule out the possibility that borrower perceptions of their own cred-
itworthiness differ according to bank market structure, nevertheless, the results do not
provide evidence that minority business owners expect it to be easier to obtain bank loans
in markets where local banks are stronger.
To guard against the possibility that black borrowers fear rejection due to concerns
about underlying credit quality, in Columns (4)-(6) repeats the analysis of columns (1)-(3)
but includes the borrowers credit score and net worth as controls. This effectively holds
constant the hard information available to lenders. This has no qualitative impact on the
findings. Black founders continue to be more afraid of denial, not less afraid of denial, in
higher soft information environments when we condition on available hard information.
To examine how these perceptions are correlated with financing, in Panel B of Table
IX, we report regressions of log business bank debt on race and interactions with the
local banking variables. In keeping with prior research, areas with higher local bank
6
We follow Cortes (2015) and Adelino, Ma and Robinson (2017) and define a local bank as one with
at least 75% of its deposits coming from that MSA. Deposit data are taken from the FDIC Summary of
Deposits. See https://www5.fdic.gov/sod/.
25
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 200 of 801
concentration are areas with higher bank lending to startups. But while entrepreneurs in
areas with stronger local banks receive larger amounts of bank loans, this is an effect that
is confined almost entirely to white borrowers. Comparing the main effect of local bank
share with the interaction term between race and local bank share suggests that the effect
of stronger local banks is almost zero for black-owned businesses.
As a further check, we also examine whether the competitiveness of the local banking
market affects our results. A more competitive local banking market could make it more
likely that black borrowers obtained loans in those markets by increasing a borrower’s
ability to shop for a loan. These results are presented in the Appendix. Here, we also
find no impact on the black dummy variable after including a Herfindahl index of local
banking competition.
In sum, areas with stronger local banks are areas where banks are perceived, and
indeed act, more favorably towards startups. But there is no evidence that areas with
stronger local banks are areas where black-owned businesses have an easier time raising
capital. Black founders are not less afraid of loan denial in these markets, nor do they
receive larger amounts of capital in these markets. The pro-startup effects of a strong
local banking community do not appear to accrue to minority business founders.
6.2 Historical Inequality and Racial Bias
Because contemporaneous measures of inequality are likely to be correlated with con-
temporaneous business conditions, we use a measure of historical inequality obtained
from Braggion, Dwarkasing, and Ongena (2015). They instrument current measures of
income inequality at the MSA level with data on the historical distribution of farm plot-
sizes in 1890. Braggion, Dwarkasing and Ongena (2015) show that this historical distri-
bution of plot sizes in 1890 is highly correlated with current measures of inequality and
use this measure to show that more historically unequal regions have lower rates of self-
employment. Based on the fact that areas with high degree of skewness in the historical
size distribution of landholdings are areas in which slavery was common, we build on
their insight and ask whether racial differences in borrowing attitudes and outcomes are
26
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 201 of 801
more pronounced in these areas by exploring interactions of the Gini coefficient with the
business owner’s race.
The main idea is to ask whether perceptions of lending outcomes are different in areas
with high historical inequality. The first three columns of Table X indicate that they are.
In Panel A, we report regression results for the fear of denial on race, the historical Gini
coefficient, and the race/gini interaction, along with all the variables listed in Table IV.
Local areas with high levels of historical inequality have much higher levels of the fear
of denial among black entrepreneurs relative to white entrepreneurs than areas with low
levels of inequality. In columns (4) through (6), we repeat the analysis in the first three
columns but include the business credit score as an independent variable. The results are
qualitatively identical.
In Panel B of Table X, we report a probit analysis for unmet capital need on race,
the historical Gini coefficient, and the race/gini interaction. Regions with high levels of
historical inequality have higher average levels of respondents reporting that they have
unmet capital need, and these effects are more pronounced among black borrowers in
areas with high inequality. As in Panel (A), this conclusion holds even when we include
the business credit score as a control variable in Columns (4) through (6).
6.3 Contemporary Inequality and Racial Bias
Next we turn to a contemporaneous measure of potential discrimination that varies re-
gionally and is likely to be correlated with racial bias, but not necessarily with contem-
poraneous business conditions. Views about interracial marriage and resulting actual
rates of interracial marriage are likely to be associated with racial prejudice. Racial preju-
dice measured along other dimensions and wage disparities are higher and when views
against interracial marriage are more negative (Charles and Guryan 2008). Thus, a find-
ing of lower levels of fear of denial and unmet capital needs in geographical areas with
high interracial marriage rates provides evidence that is at least consistent with black
entrepreneurs perceiving and facing racial bias in lending markets.
To create regional interracial marriage rates we use Census 2000 5 Percent PUMS mi-
27
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 202 of 801
crodata. We condition the sample on married couples that involve at least one black or
white partner.7 At the state level, we calculate the percentage of blacks who have white
marital partners. We then normalize the interracial marriage rate by the probability of
marriage to a white partner that would occur if this were random. For example, if 10 per-
cent of blacks are married to white partners and the population is 90 percent white then
0.10
the normalized interracial marriage rate is 0.90 =0.11, whereas the normalized interracial
marriage rate for an area with 10 percent of blacks married to whites and a population
that is 70 percent white would have a higher normalized rate (0.14) because the underly-
ing probability of an interracial marriage for a black is lower.8
Table XI reports the same set of specifications as Table X. In Panel A, we report regres-
sion results for the fear of denial on race, the interracial marriage rate coefficient, and the
race/marriage interaction, along with all the variables listed in Table IV.
Local areas with high levels of interracial marriage have much lower levels of fear
of denial among black entrepreneurs relative to white entrepreneurs than areas with low
levels of interracial marriage. In Panel B of Table XI, we report a probit analysis for unmet
capital need on race, the interracial marriage coefficient, and the race/marriage interac-
tion. Regions with high levels of interracial marriage have lower average levels of unmet
capital need among black borrowers relative to white borrowers. Overall, fear of denial
and unmet capital are lower among black entrepreneurs relative to white entrepreneurs
in areas where interracial marriage is higher, and thus potentially racial bias is lower.
7 What if Black-owned Startups Looked Like White-Owned
Ones?
The final part of our analysis asks how much of the racial gap in funding documented
above would disappear if black-owned startups had similar observable characteristics to
white-owned startups. To explore this, we use a technique pioneered by Blinder (1973)
7
We use the Census microdata to match heads of households to their spouses using household identifier
codes.
8
The normalization also results in similar rates when the focus is shifted to the percentage of whites
married to blacks. In these two examples, we would have 1.11 and 1.43 percent of whites married to blacks,
with the same normalized interracial marriage rates of 0.11 and 0.14.
28
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 203 of 801
and Oaxaca (1973) that decomposes the inter-group differences in a dependent variable
into those due to different observable characteristics across groups (sometime referred to
as the endowment effect) and those due to different “prices” of characteristics of groups.
Consider a regression Y = Xβ + with group means of the independent variables for
the black and white subpopulations given by X̄ B and X̄ W . To implement the standard
Blinder-Oaxaca decomposition, we begin by writing the inter-group difference in the av-
erage value of a dependent variable, Y, as:
h i
Ȳ W − Ȳ B = X̄ W − X̄ B β̂ W + X¯B β̂ W − β̂ B
(1)
The first term, X̄ W − X̄ B β̂ W , reflects the part of the inter-group difference that can be at-
tributed to differences in the group averages of the independent variables X—differences
in observables. The second term reflects the different “prices” or factor loadings of the
characteristics across the two groups.
There are two issues associated with implementing Equation 1. The first concerns
h i
how to deal with the second term of the equation, X¯B β̂ W − β̂ B . This “unexplained”
component of the decomposition partly captures contributions from group differences in
unobserved characteristics. This part is sensitive to the choice of omitted characteristics
making the results difficult to interpret. Another issue that arises is the “index” problem
is that the decomposition itself can either be written using coefficient weights β W or β B .9
To deal with this issue, we use an alternative method developed by Oaxaca and Ran-
som (2004), which is to weight the first term of the decomposition expression using coef-
ficient estimates from a pooled sample of the two groups. Following this approach, we
calculate the decompositions by using coefficient estimates from regressions that includes
a sample of all racial groups. We thus calculate the first term in the decompositions as:
X̄ W − X̄ B βˆ∗
(2)
where X j are means of firm characteristics of race j, β̂ ∗ is a vector of pooled coefficient
h i
9
Note that an alternative formulation of Equation 1 is Ȳ W − Ȳ B = X̄ W − X̄ B β̂ B + X¯W β̂ W − β̂ B .
29
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 204 of 801
estimates, and j = W or B for white or black, respectively.
We report estimates using pooled estimates from a regression that includes both white
and black observations (Oaxaca and Ransom 1994). It is becoming increasingly popular
when studying racial differences to use the full sample of all races to estimate the coeffi-
cients (Fairlie and Robb 2007). This version of the pooled sample is advantageous in that
it incorporates the full market response and does not exclude other racial groups. The
full set of racial and ethnic dummies in the regression specification are included to allow
us to remove any influence on the coefficients from racial differences that are correlated
with any of the explanatory variables.
Table XII reports Blinder/Oaxaca decompositions of the difference in business size in
the seventh year after startup, which is the final year of the KFS.10 The top panel of Ta-
ble XII shows that the accumulated difference in business size by the end of the survey is
about $336,000. Around one-half of this difference can be explained with observable char-
acteristics. The lower panel shows how much can be attributable to each set of observable
characteristics.
Roughly speaking, the explanatory components of this difference can be grouped into
three equally sized categories. About one-third of the difference is attributable to differ-
ences in business credit scores. Another one-third is attributable to differences in founder
net worth. And the final one-third is attributable to all other observable characteristics:
gender, founder education and work experience (collectively labeled human capital), as
well as business characteristics such as incorporation status, whether it generates a prod-
uct or service, whether it operates in or outside the home, and whether it owns intellectual
property.
Given that the average black-owned business is around $200,000 in size in year 7,
assigning average white characteristics to an average black-owned business would result
in it being about seventy-five percent larger. Merely assigning white credit scores to a
black-owned business would result in a business about 25% larger. On the one hand, to
10
Similar decompositions for individual sources of capital and for individual years mirrors the results
presented here and are available from the authors upon request.
30
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 205 of 801
the extent that this score can be improved by better financial management, rather than
simply being a manifestation of circumstances that are difficult to control, these results
suggest that improving credit scores would have a non-trivial impact on the racial gap
in funding. These results also illustrate that about half the difference in size cannot be
explained by observables, which illustrates the importance of attitudes and perceptions
by and about black borrowers in credit markets.
8 Conclusion
This paper uses confidential, restricted-access microdata from the KFS and matched ad-
ministrative data on credit scores to explore racial inequality in access to capital among
startups. Our analysis of detailed financial data available in the KFS and panel data fol-
lowing startups through the first seven years of existence provides several novel find-
ings. Black entrepreneurs start businesses at a substantially smaller scale than white en-
trepreneurs, and while the disparity in later-stage capital injections narrows over time,
they continue to take on less capital in the early years of the firm’s operation than white
entrepreneurs. Thus, initial funding differences persist. We also find that black en-
trepreneurs access less outside debt in the founding year and in the years that follow,
which is by the far the largest cause of disparities in total financial capital. Alternative
sources of capital such as loans from friends and family, personal equity and credit cards
also do little to attenuate these disparities. Black-owned startups have lower levels of all
sources of funding than do white-owned startups.
These differences in financial capital use do not appear to be due to differences in the
need for capital between black and white entrepreneurs. Black startups report substan-
tially higher levels of loan denials and overall unmet need for capital than white startups,
even after controlling for differences in credit scores and founder wealth. Moreover, in-
dustry differences, which should represent first-order differences in need for capital do
not explain racial disparities. The inclusion of detailed, potentially endogenous business
characteristics such as goals for growth and type of business also has little effect on the
31
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 206 of 801
racial differences we find providing further evidence against need differences.
Focusing on supply-side channels, we find that racial differences in financial capital
cannot be attributed entirely to white lenders looking unfavorably upon black borrowers.
There are large differences in credit worthiness between black and white entrepreneurs.
Detailed administrative data on credit ratings linked to all KFS businesses provides the
first evidence in the literature of extensive differences in creditworthiness between black
and white startups and their effects on financing outcomes. Our analysis also reveals that
the relatively low credit scores for black business owners explain a substantial amount of
the gaps in both financing at startup and in the years after startup. These results imply
that a great deal of the capital investment differences between black- and white-owned
businesses is the result of persistent differences in the founder’s financial health that are
present at the very inception of the firm. This connects our findings to an increasing
concern over inequality in household finance and financial literacy and suggests inter-
esting connections between household financial planning, behavioral finance, race and
entrepreneurship.
At the same time, on the demand side our evidence clearly indicates an enduring be-
lief among even the most credit-worthy black borrowers that they will be turned away
by banks. The fact that many well-qualified black entrepreneurs do not apply for credit,
even when they feel they need it, because they anticipate being denied credit suggests
that overcoming differences between black and white borrowers is not simply a matter
of expanding the supply of credit available to lower income borrowers. Interestingly,
we also find that simply increasing the strength of local banks is unlikely to help – al-
though white-owned startups receive large amounts of bank debt on average in areas
with stronger local banks black-owned startups do not. 11 Getting to the root cause of
racial differences in the way that new businesses are financed likely requires changes in
perceptions and financial planning behaviors as much as it requires augmenting the sup-
ply of credit to traditionally underserved borrowers.
11
Further increases in credit card access might help reduce disparities (Chatterji and Seamans 2012), but
this source provides only high interest borrowing which might be prohibitive for larger borrowing needs.
32
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 207 of 801
References
[1] Adelino, M. , Ma, S. and Robinson, D. (2017), Firm Age, Investment Opportunities,
and Job Creation. The Journal of Finance, 72: 999-1038. doi:10.1111/jofi.12495
[2] Altonji, J. G., and Pierret, C. R. (2001). Employer learning and statistical discrimina-
tion. The Quarterly Journal of Economics, 116(1), 313-350.
[3] Ballou, J., Barton, T., DesRoches, D., Potter, F., Reedy, E.J., Robb, A., Shane, S. and
Zhao, Z. (2008). Kauffman Firm Survey: Results from the Baseline and First Follow-
Up Surveys. Kauffman Foundation.
[4] Bates, Timothy. “The changing nature of minority business: A comparative analysis
of Asian, Non-Minority and Black owned businesses.” The Review of Black Political
Economy (1989): 25-42.
[5] Bates, Timothy. “Commercial bank financing of white and black owned small busi-
ness start-ups.” Quarterly Review of Economics and Business (1991): 64-80.
[6] Bates, Timothy and Alicia Robb (2014). “Has the Community Reinvestment Act In-
creased Loan Availability Among Small Businesses Operating in Minority Neigh-
borhoods?” Urban Studies, May, pp. 1-20.
[7] Berger, Allen, Geraldo Cerquiero, and Maria Fabiana Penas (2014). “Market size
structure and small business lending: Are crisis times different from normal times?”
Review of Finance 19(5): 1965-1995.
[8] Berger, Allen, Nathan Miller, Mitchell Petersen, Raghuram Rajan, and Jeremy Stein.
2005. “Does function follow organizational form? Evidence from the lending prac-
tices of large and small banks.” Journal of Financial Economics, 76(2): 237-269.
[9] Blanchflower, David G., P. Levine and D. Zimmerman. 2003. “Discrimination in the
small business credit market”, Review of Economics and Statistics, November, 85(4),
pp. 930-943.
[10] Blinder, Alan S. 1973. “Wage Discrimination: Reduced Form and Structural Vari-
ables.” Journal of Human Resources 8: 436-455.
[11] Board of Governors of the Federal Reserve System. 2007. “Report to the Congress on
Credit Scoring and Its Effects on the Availability and Affordability of Credit.”
[12] Bone, Sterling A., Glenn Christensen and Jerome Williams. 2014. “Rejected, Shack-
led, and Alone: The Impact of Systemic Restricted Choice on Minority Consumers’
Construction of Self” Journal of Consumer Research 41. 451–474.
[13] Borjas, George J. 1999. “The Wage Structure and Self-Selection into Self-
Employment.” Working paper. Cambridge, MA: Harvard University.
[14] Bostic, R. W. and K. P. Lampani. 1999. “Racial Differences in Patterns of Small Busi-
ness Finance: The Importance of Local Geography,” Proceedings of the Conference
on Business Access to Capital and Credit, August, 149-179.
33
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 208 of 801
[15] Boston, Thomas D. 1999. “Generating Jobs Through African American Business De-
velopment”, in J. Whitehead and C. Harris, eds. Readings in Black Political Economy
(Dubuque: Kendall-Hunt)
[16] Boston, Thomas D. 2006. “Black Patronage of Black-owned startups and Black Em-
ployment” in J. Whitehead, J. Stewart and C. Conrad (eds). African Americans in the
United States (Rowman & Littlefield Publishers, Inc).
[17] Bradford, William D., and Alfred E. Osborne. 1976. “The entrepreneurship decision
and black economic development.” The American Economic Review 66.2: 316-319.
[18] Bradford, William D. 2003. “The Wealth Dynamics of Entrepreneurship for Black and
White Families in the U.S.,” Review of Income and Wealth, 49(1): 89-116.
[19] Brickley, James, James Link, and Clifford Smith. 2003. “Boundaries of the Firm: Evi-
dence from the Banking Industry.” Journal of Financial Economics, 70(3): 351-383.
[20] Cavalluzzo, Ken, Linda Cavalluzzo, and John Wolken. 2002. “Competition, Small
Business Financing, and Discrimination: Evidence from a New Survey,” Journal of
Business, Vol. 75(4): 641-679.
[21] Cavalluzzo, Ken and John Wolken. 2005. “Small Business Loan Turndowns, Personal
Wealth and Discrimination.” Journal of Business, 78(6): 2153-2177.
[22] Charles, Kerwin Kofi, and Jonathan Guryan. (2008). ”Prejudice and wages: an em-
pirical assessment of Becker?s The Economics of Discrimination.” Journal of Political
Economy, 116(5): 773-809.
[23] Charles, Kerwin Kofi, and Jonathan Guryan. (2011). “Studying discrimination: Fun-
damental challenges and recent progress.” Annual Review of Economics, 3(1), 479-
511.
[24] Chatterji, Aaron K., and Robert C. Seamans. 2012. “Entrepreneurial finance, credit
cards, and race.” Journal of Financial Economics 106(1): 182-195.
[25] Cole, R. A. (1999). “Availability of credit to small and minority-owned businesses:
Evidence from the 1993 National Survey of Small Business Finances.”
[26] Cole, Rebel A., and Tatyana Sokolyk. 2013. “How do start-up firms finance their
assets? Evidence from the Kauffman Firm Surveys.”
[27] Cortes, Kristle R., 2015, “Rebuilding after disaster strikes: How local lenders aid in
the recovery,” Federal Reserve Bank of Cleveland Working Paper.
[28] Dougal, Casey, Pengjie Gao, Bill Mayhew and Chris Parsons, 2017. “What’s in a
(School) Name? Racial Discrimination in Higher Education Bond Markets.” Working
paper.
[29] Fairlie, Robert W.,Dean Karlan, and Jonathan Zinman. 2015. “Behind the GATE Ex-
periment: Evidence on Effects of and Rationales for Subsidized Entrepreneurship
Training,” American Economic Journal: Economic Policy, 7(2): 125-61.
34
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 209 of 801
[30] Fairlie, Robert W., and Alicia M. Robb. 2007. “Why are Black-owned startups Less
Successful than White-Owned Businesses: The Role of Families, Inheritances, and
Business Human Capital,” Journal of Labor Economics, 25(2): 289-323.
[31] Fairlie, Robert W., and Alicia M. Robb. 2008. “Race and Entrepreneurial Success:
Black-, Asian-, and White-Owned Businesses in the United States,” Cambridge: MIT
Press.
[32] Farhat, Joseph B., and Alicia Robb. “Applied survey data analysis using Stata: The
Kauffman firm survey data.” (2014).
[33] Haltiwanger, John, Ron Jarmin and Javier Miranda. 2013. “Who Creates Jobs? Small
vs. Large vs. Young.” Review of Economics and Statistics, 347-361.
[34] Hurst, Erik and Benjamin Pugsley. 2011. “What do Small Businesses Do?” NBER
Working Paper #17041.
[35] Kerr, William R., and Ramana Nanda. 2011. “Financing constraints and en-
trepreneurship.” Handbook of Research on Innovation and Entrepreneurship eds.
David B. Audretsch, Oliver Falck, Stephan Heblich and Adam Leder: 88.
[36] Koellinger, Phillipp, and Maria Minniti. 2006. “Not for lack of trying: American en-
trepreneurship in black and white,” Small Business Economics 27(1): 59-79.
[37] Mitchell, Karlyn, and Douglas Pearce. 2011 “Lending technologies, lending special-
ization, and minority access to small-business loans.” Small Business Economics 37,
no. 3: 277-304.
[38] Oaxaca, Ronald. 1973. “Male-Female Wage Differentials in Urban Labor Markets,”
International Economic Review, 14 (October), 693-709.
[39] Oaxaca, Ronald, and Michael Ransom. 1994. “On Discrimination and the Decompo-
sition of Wage Differentials,” Journal of Econometrics, 61, 5-21.
[40] Parker, Simon C. (2009). The Economics of Entrepreneurship. Cambridge: Cam-
bridge University Press.
[41] Petersen, Mitchell and Raghuram Rajan, 2002. “Does Distance Still Matter? The In-
formation Revolution in Small Business Lending,” Journal of Finance 57(6): 2533-
2570.
[42] Robb, Alicia, et al. “An overview of the Kauffman Firm Survey: Results from the
2004-2008.” Ewing Marion Kauffman Foundation Research Paper (2010).
[43] Robb, Alicia and David Robinson, “The Capital Structure Decisions of New Firms.”
Review of Financial Studies, Vol 27, No 1, 2014.
[44] Stangler, Dane, Inara Tareque, and Arnobio Morelix. “Trends in Venture Capital,
Angel Investments, and Crowdfunding across the Fifty Largest US Metropolitan Ar-
eas.” Angel Investments, and Crowdfunding across the Fifty Largest US Metropoli-
tan Areas (December 1, 2016) (2016).
35
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 210 of 801
[45] Simoes, Nadia, Nuno Crespo, and Sandrina B. Moreira. “Individual determinants
of self-employment entry: What do we really know?.” Journal of economic surveys
30.4 (2016): 783-806.
[46] U.S. Bureau of Labor Statistics. 2014. Table A-2. Employment status of the civilian
population by race, sex, and age, http://www.bls.gov/news.release/empsit.t02.htm
[47] U.S. Census Bureau. 2012. Wealth and Asset Ownership: Table 1. Median Value of
Assets for Households, by Type of Asset Owned and Selected Characteristics: 2011,
http://www.census.gov/people/wealth/
[48] U.S. Census Bureau. 2015. 2012 Survey of Business Owners: Company Summary
[49] U.S. Census Bureau. 2016. Income and Poverty in the United States: 2015, In-
come and Earnings Summary Measures by Selected Characteristics: 2014 and 2015,
http://www.census.gov/library/publications/2016/demo/p60-256.html
[50] U.S. Census Bureau. 2017. Wealth, Asset Ownership, & Debt of Households Detailed
Tables: 2013, https://www.census.gov/data/tables/2013/demo/wealth/wealth-
asset-ownership.html.
[51] U.S. Census Bureau. 1997. 1992 Economic Census: Characteristics of Business Own-
ers. Washington, D.C.: U.S. Government Printing Office.
[52] van der Sluis, J., van Praag, M., and Vijverberg, W. (2005), “Education and En-
trepreneurship in Industrialized Countries: A Meta-Analysis,” World Bank Eco-
nomic Review, 19(2): 225-261.
[53] Walstad, William B. and Marilyn L. Kourilsky. 1998. “Entrepreneurial Attitudes and
Knowledge of Black Youth,” Entrepreneurship Theory & Practice, 23(2): 5-18.
36
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 211 of 801
Figure 1: Racial Differences in Wealth in the Kauffman Firm Survey
Net Worth Distribution by Race
U.S. Population (SIPP) and Startup Owners (KFS)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1st+2nd Quartile (1-44%) 3rd Quartile (45-70%) 4th Quartile (71-99%)
<$50,000 $50,000-249,999 $250,000+
Total Population Total Startup Owners White Population White Startup Owners Black Population Black Startup Owners
37
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 212 of 801
Figure 2: Racial Differences in Sources of Initial Capital for Startups
120000
100000
80000
60000
40000
20000
0
Owner's Informal Formal Owner Informal Formal Total
Equity Equity Equity Debt Debt Debt Financial
Capital
Overall White Black
38
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 213 of 801
Figure 3: Racial Differences in Sources of New Capital for Startups, Years 1-3
80000
70000
60000
50000
40000
30000
20000
10000
0
Owner's Formal Informal Owner Informal Formal Total
Equity Equity Equity Debt Debt Debt Financial
Capital
Overall White Black
39
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 214 of 801
Figure 4: Racial Differences in the Evolution of Total Assets over Time
400000
350000
300000
250000
200000
150000
100000
50000
0
0 1 2 3 4 5 6 7
ALL White Black
40
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 215 of 801
Table I: Racial Differences in Sources of Funding
This table reports survey-weighted mean values by race for broad funding categories. The
components of the classifications (Owner, Insider, Outsider/Debt, Equity) are described
in detail in Appendix Table A.1. The final column reports p-values from the t-test of the
difference between black- and white-owned businesses.
Overall Mean White Mean Black Mean p-value(diff)
KFS Initial Survey Year
Owner’s Equity 33,078 34,426 19,562 0.00
Inside Equity 2,117 2,139 440 0.14
Outside Equity 16,768 18,543 536 0.10
Owner Debt 4,890 5,228 1,010 0.05
Inside Debt 6,663 7,195 2,849 0.17
Outside Debt 51,680 56,663 10,809 0.01
Total Financial Capital 99,344 106,720 35,205 0.00
Leverage Ratio 0.19 0.19 0.12 0.00
KFS Survey Years 1-3
Owner’s Equity 13,047 13,308 8,555 0.13
Outside Equity 14,864 16,499 551 0.07
Inside Equity 1,206 1,284 664 0.48
Owner Debt 4,200 4,336 2,297 0.15
Inside Debt 5,385 5,713 2,491 0.49
Outside Debt 51,147 54,813 14,883 0.19
Total Financial Capital 69,256 72,958 29,107 0.00
Leverage Ratio 0.29 0.30 0.21 0.00
KFS Survey Years 4-7
Owner’s Equity 8,327 7,944 4,678 0.42
Outside Equity 7,663 8,339 1,227 0.20
Inside Equity 1,037 1,047 254 0.63
Owner Debt 3,618 3,671 3,482 0.42
Inside Debt 4,898 5,176 979 0.21
Outside Debt 48,616 49,809 20,265 0.64
Total Financial Capital 58,684 59,825 27,348 0.54
Leverage Ratio 0.29 0.29 0.20 0.00
41
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 216 of 801
Table II: Initial Differences in Log Total Capital
This table models variation in the amount of total capital from all sources, include founder, insider and outside debt and
equity. All columns include 2-digit NAICS industry fixed effects and controls for gender and other racial categories (Asian, Hispanic,
and Other). Missing or negative net worth is the omitted category. Column (7) includes CBSA fixed effects, while Column (8) includes
zipcode-level fixed effects. Survey weights are used in Columns (1)-(7), but not in Column (8). One, two and three asterisks denote
statistical significance at the 10, 5 and 1% level, respectively.
(1) (2) (3) (4) (5) (6) (7) (8)
Black-owned Startup -0.731*** -0.599*** -0.485*** -0.512*** -0.496*** -0.501*** -0.502*** -0.857***
(0.113) (0.111) (0.110) (0.109) (0.111) (0.102) (0.114) (0.330)
Credit Score 0.021*** 0.020*** 0.020*** 0.020*** 0.012*** 0.012*** 0.010***
(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.004)
Net Worth: Up to 50K -0.517*** -0.498*** -0.504*** -0.313*** -0.288** -0.465*
(0.119) (0.121) (0.120) (0.109) (0.127) (0.278)
Net Worth: 50-100K -0.581*** -0.566*** -0.566*** -0.392*** -0.332** -0.050
(0.130) (0.130) (0.129) (0.116) (0.131) (0.283)
Net Worth: 100-250K -0.167 -0.151 -0.153 -0.061 -0.035 -0.167
(0.108) (0.108) (0.108) (0.103) (0.114) (0.241)
Net Worth: Over 250K 0.360*** 0.332*** 0.330*** 0.277*** 0.232** 0.099
(0.100) (0.102) (0.101) (0.095) (0.104) (0.186)
Previous Industry Experience -0.007 -0.006 -0.006 -0.008* -0.005
(0.004) (0.004) (0.004) (0.004) (0.008)
Experience Outside Industry 0.001 0.002 0.011*** 0.011*** 0.012
(0.004) (0.004) (0.003) (0.004) (0.008)
Some College 0.051 0.048 0.001 0.041 0.217
(0.109) (0.108) (0.101) (0.115) (0.258)
College Deg. 0.113 0.096 0.002 0.023 0.009
(0.121) (0.120) (0.115) (0.128) (0.276)
Grad. Deg. 0.306** 0.310** 0.139 0.182 0.028
(0.140) (0.139) (0.131) (0.143) (0.288)
Prev. Startup Exp. 0.289*** 0.264*** 0.105 0.109 0.100
(0.073) (0.073) (0.068) (0.075) (0.148)
Makes Product 0.344*** 0.219*** 0.211** 0.321**
(0.083) (0.076) (0.085) (0.162)
Intel. Property 0.216** 0.083 0.017 0.372**
(0.091) (0.081) (0.090) (0.173)
Home-based -0.728*** -0.699*** -0.825***
(0.075) (0.082) (0.167)
Part time Bus. -0.821*** -0.861*** -0.854***
(0.085) (0.091) (0.179)
Incorporated 0.658*** 0.705*** 0.506***
(0.071) (0.082) (0.170)
Employment 0.060*** 0.061*** 0.045***
(0.012) (0.012) (0.009)
Observations 4,124 4,038 4,038 3,975 3,975 3,840 3,590 1,214
R-squared 0.055 0.097 0.117 0.131 0.139 0.281 0.394 0.624
42
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 217 of 801
Table III: Initial Differences in Total Outside Debt
This table models variation in the log of the amount of total outside debt. All columns include 2-digit NAICS industry fixed
effects and controls for gender and other racial categories (Asian, Hispanic, and Other). Missing or negative net worth is the omitted
category. Column (7) includes CBSA fixed effects, while Column (8) includes zipcode-level fixed effects. Survey weights are used
in Columns (1)-(7), but not in Column (8). One, two and three asterisks denote statistical significance at the 10, 5 and 1% level,
respectively.
(1) (2) (3) (4) (5) (6) (7) (8)
Black-owned Startup -0.654*** -0.544*** -0.442*** -0.438*** -0.419*** -0.431*** -0.417*** -0.922***
(0.111) (0.109) (0.111) (0.112) (0.112) (0.110) (0.121) (0.355)
Credit Score 0.019*** 0.018*** 0.018*** 0.018*** 0.013*** 0.013*** 0.011***
(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.004)
Net Worth: Up to 50K -0.406*** -0.377*** -0.380*** -0.247** -0.216* -0.171
(0.121) (0.122) (0.122) (0.119) (0.129) (0.299)
Net Worth: 50-100K -0.296** -0.273** -0.268** -0.133 -0.100 0.120
(0.130) (0.130) (0.129) (0.126) (0.137) (0.305)
Net Worth: 100-250K -0.044 -0.029 -0.032 0.032 0.099 -0.245
(0.125) (0.125) (0.126) (0.124) (0.137) (0.260)
Net Worth: Over 250K 0.328*** 0.305** 0.306** 0.258** 0.207* 0.150
(0.117) (0.121) (0.120) (0.119) (0.125) (0.201)
Previous Industry Experience -0.003 -0.003 -0.004 -0.003 -0.006
(0.005) (0.005) (0.005) (0.005) (0.009)
Experience Outside Industry 0.005 0.005 0.010** 0.011** 0.013
(0.004) (0.004) (0.004) (0.004) (0.008)
Some College 0.069 0.072 0.032 0.101 -0.012
(0.124) (0.124) (0.122) (0.133) (0.278)
College Deg. 0.048 0.045 -0.046 0.060 -0.124
(0.136) (0.136) (0.136) (0.147) (0.297)
Grad. Deg. 0.318** 0.337** 0.208 0.263 -0.076
(0.158) (0.159) (0.158) (0.167) (0.310)
Prev. Startup Exp. 0.229*** 0.214*** 0.093 0.133 -0.238
(0.082) (0.082) (0.082) (0.087) (0.159)
Makes Product 0.337*** 0.258*** 0.206** 0.198
(0.092) (0.090) (0.098) (0.175)
Intel. Property 0.003 -0.125 -0.141 -0.130
(0.101) (0.097) (0.103) (0.187)
Home-based -0.385*** -0.322*** -0.464**
(0.083) (0.090) (0.180)
Part time Bus. -0.326*** -0.368*** -0.659***
(0.089) (0.092) (0.193)
Incorporated 0.460*** 0.479*** 0.249
(0.083) (0.093) (0.183)
Employment 0.075*** 0.072*** 0.068***
(0.013) (0.012) (0.010)
Observations 4,124 4,038 4,038 3,975 3,975 3,840 3,590 1,214
R-squared 0.032 0.064 0.074 0.082 0.086 0.155 0.310 0.585
43
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 218 of 801
Table IV: Initial Differences in Total Business Debt
This table models variation in the amount of total debt for the business. All columns include 2-digit NAICS industry fixed
effects and controls for gender and other racial categories (Asian, Hispanic, and Other). Missing or negative net worth is the omitted
category. Column (7) includes CBSA fixed effects, while Column (8) includes zipcode-level fixed effects. Survey weights are used
in Columns (1)-(7), but not in Column (8). One, two and three asterisks denote statistical significance at the 10, 5 and 1% level,
respectively.
(1) (2) (3) (4) (5) (6) (7) (8)
Black-owned Startup -0.339*** -0.302*** -0.251*** -0.239*** -0.222*** -0.212*** -0.200*** -0.423*
(0.044) (0.044) (0.045) (0.047) (0.047) (0.048) (0.055) (0.218)
Credit Score 0.006*** 0.006*** 0.006*** 0.006*** 0.003** 0.004*** 0.006**
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.003)
Net Worth: Up to 50K -0.005 0.009 0.006 0.072 -0.018 0.146
(0.079) (0.079) (0.079) (0.079) (0.073) (0.184)
Net Worth: 50-100K -0.074 -0.060 -0.055 0.005 -0.106* -0.056
(0.070) (0.071) (0.071) (0.072) (0.060) (0.187)
Net Worth: 100-250K -0.057 -0.045 -0.048 -0.005 -0.050 -0.179
(0.074) (0.073) (0.074) (0.075) (0.077) (0.159)
Net Worth: Over 250K 0.239*** 0.235*** 0.236*** 0.210*** 0.182** 0.141
(0.079) (0.082) (0.081) (0.081) (0.083) (0.123)
Previous Industry Experience -0.002 -0.002 -0.003 -0.003 -0.003
(0.003) (0.003) (0.003) (0.003) (0.006)
Experience Outside Industry -0.001 -0.001 -0.000 -0.001 -0.001
(0.002) (0.002) (0.002) (0.002) (0.005)
Some College -0.119 -0.115 -0.144* 0.009 -0.219
(0.076) (0.075) (0.075) (0.075) (0.170)
College Deg. -0.033 -0.032 -0.082 0.100 -0.071
(0.087) (0.087) (0.086) (0.086) (0.182)
Grad. Deg. 0.092 0.112 0.038 0.149 -0.059
(0.105) (0.105) (0.105) (0.096) (0.190)
Prev. Startup Exp. 0.084* 0.074 0.012 0.024 -0.288***
(0.051) (0.050) (0.050) (0.051) (0.097)
Makes Product 0.279*** 0.243*** 0.230*** 0.217**
(0.059) (0.060) (0.060) (0.107)
Intel. Property -0.050 -0.102* -0.052 -0.190*
(0.066) (0.062) (0.058) (0.114)
Home-based -0.190*** -0.127** -0.117
(0.048) (0.051) (0.110)
Part time Bus. 0.039 -0.013 -0.127
(0.056) (0.054) (0.118)
Incorporated 0.144*** 0.095* -0.003
(0.050) (0.054) (0.112)
Employment 0.056*** 0.053*** 0.069***
(0.011) (0.010) (0.006)
Observations 4,124 4,038 4,038 3,975 3,975 3,840 3,590 1,214
R-squared 0.022 0.032 0.039 0.044 0.052 0.110 0.346 0.590
44
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 219 of 801
Table V: Later-Stage Differences in Debt
This table models variation in the amount of total debt, outside debt, and business debt for the later survey years. The re-
gression specifications mirror those in Column (6) of the previous three tables. All columns include industry fixed effects controls for
gender and other racial categories and dummy variables for the survey years. Human capital controls include education, previous
work experience and previous startup experience. Product characteristics control for whether the business sells a product or a service
(or both), and whether it has intellectual property. Firm characteristics control for whether the business is fulltime or parttime,
whether it is home-based, incorporated and has employees. Standard errors appear in parentheses below point estimates. One, two
and three asterisks denote significance at the ten, five and one per cent level, respectively.
Business Total Total
Bank Debt Outside Debt Financial Capital
Years 1-3 Years 4-7 Years 1-3 Years 4-7 Years 1-3 Years 4-7
(1) (2) (3) (4) (5) (6)
Black-owned Startup -0.163*** -0.084*** -0.260*** -0.122 -0.017 0.067
(0.031) (0.032) (0.084) (0.079) (0.092) (0.096)
Controls:
Credit Score Yes Yes Yes Yes Yes Yes
Net Worth Yes Yes Yes Yes Yes Yes
Human Capital Yes Yes Yes Yes Yes Yes
Product Characteristics Yes Yes Yes Yes Yes Yes
Firm Characteristics Yes Yes Yes Yes Yes Yes
Industry Dummies Yes Yes Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes Yes Yes
Observations 9,482 8,979 9,608 8,981 9,608 8,981
R-squared 0.157 0.124 0.135 0.139 0.051 0.046
45
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 220 of 801
Table VI: Racial Differences in Attitudes Towards Formal Debt
This table reports survey-weighted averages by racial group to questions in the KFS that capture attitudes and intentions with respect
to borrowing. “Applied for a loan” is a dummy equaling one if the respondent applied for a loan, regardless of whether the loan was
approved. “Did not apply for fear of rejection” is one for those borrowers who did not apply for a loan, but who did not only because
they anticipated the loan being denied. “Loan Always Approved” is only available for those who applied for a loan: it is a dummy for
whether the respondent received the full amount they were asking for, or whether sometimes their loans are denied or reduced in size.
“Unmet Need” is 1 if the respondent either did not apply for fear of rejection, or else applied but did not always get the full amount.
The column labeled Overall is for all respondents. The remaining columns split the sample on whether the respondent had below or
above median credit score, or whether credit scores were above the 75th percentile of observed scores across the whole sample.
Credit Score:
Overall Below Median Above Median Above 75th
Applied for a Loan
White 0.1200 0.0838 0.1414 0.1617
Black 0.0785 0.0752 0.0834 0.1125
Loan Always Approved
White 0.6826 0.6201 0.7038 0.7225
Black 0.2240 0.1153 0.3862 0.2530
Did Not Apply For Fear of Rejection
White 0.1617 0.1666 0.1590 0.1497
Black 0.4181 0.4746 0.3244 0.3228
Unmet Need
White 0.1633 0.1671 0.1611 0.1525
Black 0.4295 0.4929 0.3246 0.3174
46
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 221 of 801
Table VII: Race and the Demand for Capital
This table provides a multivariate analysis of the relation between founder race and the demand for capital. In Panel A, the dependent
variable is a dummy for whether the borrower did not apply for a loan for fear of denial. In Panel B, the dependent variable is a dummy
for whether they applied but were denied credit or else received less than they asked for. Human capital controls include education,
previous work experience and previous startup experience. Product characteristics control for whether the business sells a product
or a service (or both), and whether it has intellectual property. Firm characteristics control for whether the business is fulltime or
parttime, whether it is home-based, incorporated and has employees. Standard errors appear in parentheses below point estimates.
One, two and three asterisks denote significance at the ten, five and one per cent level, respectively.
Panel A: Did Not Apply for Fear of Denial
(1) (2) (3) (4) (5) (6)
Black-owned Startup 0.856*** 0.798*** 0.638*** 0.621*** 0.628*** 0.617***
(0.058) (0.059) (0.070) (0.070) (0.070) (0.074)
Controls:
Credit Score No Yes Yes Yes Yes Yes
Net Worth No No Yes Yes Yes Yes
Human Capital No No No Yes Yes Yes
Product Characteristics No No No No Yes Yes
Firm Characteristics No No No No No Yes
Industry Dummies Yes Yes Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes Yes Yes
Observations 11,380 11,337 8,982 8,878 8,878 8,620
Panel B: Denied Credit or Received Less than Requested
(1) (2) (3) (4) (5) (6)
Black-owned Startup 0.450*** 0.432*** 0.275** 0.288** 0.307** 0.292**
(0.101) (0.102) (0.124) (0.122) (0.122) (0.128)
Controls:
Credit Score No Yes Yes Yes Yes Yes
Net Worth No No Yes Yes Yes Yes
Human Capital No No No Yes Yes Yes
Product Characteristics No No No No Yes Yes
Firm Characteristics No No No No No Yes
Industry Dummies Yes Yes Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes Yes Yes
Observations 9,954 9,915 7,829 7,736 7,736 7,515
47
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 222 of 801
Table VIII: Past Borrowing and Credit Beliefs
This table explores the relation between prior loan balances and lack of access to capital by race. The
dependent variable in the first two columns is a dummy variable for whether the respondent was afraid
to apply for a loan for fear of denial. Column (1) includes all businesses that are not black-owned, while
Column (2) focuses only on black-owned startups. In columns (3) and (4), the dependent variable is a
dummy for whether the respondent reported that they were either denied credit or they received less
than they asked for: again, the columns split the samples according to the race of the founder. Prior
Accumulated Debt is the sum of all outside debt up through the previous survey round, in hundreds
of thousands of dollars. Human capital controls include education, previous work experience and
previous startup experience. Product characteristics control for whether the business sells a product or
a service (or both), and whether it has intellectual property. Firm characteristics control for whether the
business is fulltime or parttime, whether it is home-based, incorporated and has employees. Standard
errors appear in parentheses below point estimates. One, two and three asterisks denote significance at
the ten, five and one per cent level, respectively.
Fear of Denial Denied Credit
(1) (2) (3) (4)
Prior Accumulated Debt 0.007*** -0.011* 0.007*** -0.005
(0.002) (0.007) (0.002) (0.004)
Controls:
Credit Score Yes Yes Yes Yes
Net Worth Yes Yes Yes Yes
Human Capital Yes Yes Yes Yes
Product Characteristics Yes Yes Yes Yes
Firm Characteristics Yes Yes Yes Yes
Industry Dummies Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes
Black-Owned Startup Sample No Yes No Yes
Observations 8,209 639 7,265 405
48
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 223 of 801
Table IX: Local Banking Conditions and Racial Differences in Access to Credit
Panel A reports Probit regressions in which the dependent variable is a dummy if the respondent answered yes to “Did Not Apply for
Fear of Rejection” or if they reported that they did not always get the full amount they asked for. Panel B reports regressions (Pooled
OLS with year dummies) in which the dependent variable is the natural log of total business debt. Local bank share is the share of
total county-level deposits held by local banks. Standard errors in parentheses, with one, two and three asterisks denoting significance
at the 10, 5 and 1percent level. Controls from Table IV included but not shown.
Panel A: Dependent variable is Did Not Apply for Fear of Denial
(1) (2) (3) (4) (5) (6)
Black-Owned Startup 0.820*** 0.815*** 0.794*** 0.621*** 0.618*** 0.591***
(0.066) (0.066) (0.086) (0.070) (0.070) (0.091)
Local Bank Share -0.306*** -0.321*** -0.241** -0.262**
(0.116) (0.121) (0.120) (0.127)
Local Bank Share × Black 0.146 0.192
(0.407) (0.414)
Credit Score No No No Yes Yes Yes
Net Worth No No No Yes Yes Yes
Human Capital Yes Yes Yes Yes Yes Yes
Product Characteristics Yes Yes Yes Yes Yes Yes
Firm Characteristics Yes Yes Yes Yes Yes Yes
Industry Dummies Yes Yes Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes Yes Yes
Observations 8,433 8,424 8,424 8,412 8,403 8,403
Panel B: Dependent variable is log Business Bank Debt
(1) (2) (3) (4) (5) (6)
Black-Owned Startup -0.206*** -0.199*** -0.148*** -0.133*** -0.125*** -0.082***
(0.019) (0.019) (0.024) (0.020) (0.020) (0.025)
Local Bank Share 0.327*** 0.352*** 0.328*** 0.350***
(0.059) (0.063) (0.060) (0.063)
Local Bank Share × Black -0.360*** -0.309**
(0.128) (0.130)
Credit Score No No No Yes Yes Yes
Net Worth No No No Yes Yes Yes
Human Capital Yes Yes Yes Yes Yes Yes
Product Characteristics Yes Yes Yes Yes Yes Yes
Firm Characteristics Yes Yes Yes Yes Yes Yes
Industry Dummies Yes Yes Yes Yes Yes Yes
Survey-Year Dummies Yes Yes Yes Yes Yes Yes
Observations 21,677 21,648 21,648 21,441 21,412 21,412
R-squared 0.015 0.017 0.017 0.023 0.025 0.025
49
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 224 of 801
Table X: Historical Inequality and Racial Bias
Panel A reports regressions (Pooled OLS with year dummies) in which the dependent variable is a dummy equaling one if the
respondent answered yes to “Did Not Apply for Fear of Rejection.” The dependent variable in Panel B is a dummy equaling one if
they reported that they did not always get the full amount they asked for. Regional Historical Gini is the gini coefficient of the MSA in
1890; data from Braggion, Dwarkasing, and Ongena (2015). In each panel a constant is estimated but suppressed for brevity. Standard
errors in parentheses, with one, two and three asterisks denoting significance at the 10, 5 and 1percent level. Controls from Table IV
included but not shown.
Panel A: Dependent Variable is Did Not Apply For Fear of Denial
Black 0.8141*** 0.7514*** 0.2008 0.7605*** 0.7022***
0.1819
(0.058) (0.061) (0.276) (0.059) (0.061)
(0.274)
Historical Inequality 0.5565*** 0.4507*** 0.6057***
0.5056***
(0.159) (0.166) (0.159)
(0.166)
Gini × Minority 1.1465** 1.0848*
(0.568) (0.563)
Credit Score -0.0035*** -0.0032*** -0.0032***
(0.001) (0.001) (0.001)
Constant -1.0867*** -1.3712*** -1.3198*** -0.9541*** -1.2731*** -1.2252***
(0.097) (0.133) (0.135) (0.099) (0.135) (0.137)
Observations 11,247 9,436 9,436 11,204 9,396 9,396
Panel B: Dependent variable is Unmet Capital Need
(1) (2) (3) (4) (5) (6)
Black 0.8323*** 0.7622*** 0.1579 0.7808*** 0.7165***
0.1405
(0.058) (0.060) (0.275) (0.059) (0.061)
(0.273)
Historical Inequality 0.5762*** 0.4609*** 0.6239***
0.5139***
(0.157) (0.164) (0.157)
(0.165)
Gini × Minority 1.2586** 1.2012**
(0.566) (0.560)
Credit Score -0.0034*** -0.0030*** -0.0030***
(0.001) (0.001) (0.001)
Constant -1.0384*** -1.3343*** -1.2784*** -0.9091*** -1.2429*** -1.1903***
(0.096) (0.132) (0.134) (0.098) (0.134) (0.136)
Observations 11,249 9,437 9,437 11,206 9,397 9,397
50
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 225 of 801
Table XI: Attitudes towards Interracial Marriage and Access to Capital
Panel A reports regressions (Pooled OLS with year dummies) in which the dependent variable is a dummy equaling one if the
respondent answered yes to “Did Not Apply for Fear of Rejection.” The dependent variable in Panel B is a dummy equaling one
if they reported that they did not always get the full amount they asked for. “Interracial Marriage” is the state level percentage of
black married persons who have white marital partners, scaled by the proportion of white married persons in the state. In each
panel a constant is estimated but suppressed for brevity. Standard errors in parentheses, with one, two and three asterisks denoting
significance at the 10, 5 and 1percent level. Controls from Table IV included but not shown.
Panel A: Dependent Variable is Did Not Apply For Fear of Denial
(1) (2) (3) (4) (5) (6)
Black 0.8141*** 0.8105*** 1.0072*** 0.7605*** 0.7559***
0.9670***
(0.058) (0.059) (0.101) (0.059) (0.060)
(0.102)
Interracial Marriage 0.2741 0.3860* 0.2895
0.4099*
(0.207) (0.210) (0.210)
(0.213)
Black× Interracial -2.3124** -2.4790**
(0.982) (0.983)
Credit Score -0.0035*** -0.0036*** -0.0036***
(0.001) (0.001) (0.001)
Constant -1.0867*** -1.0932*** -1.1125*** -0.9541*** -0.9596*** -0.9802***
(0.097) (0.101) (0.102) (0.099) (0.103) (0.104)
Observations 11,247 10,830 10,830 11,204 10,787 10,787
Panel B: Dependent variable is Unmet Capital Need
(1) (2) (3) (4) (5) (6)
Black 0.8323*** 0.8370*** 1.0509*** 0.7808*** 0.7848***
1.0139***
(0.058) (0.059) (0.101) (0.059) (0.060)
(0.102)
Interracial Marriage 0.4813** 0.6007*** 0.5034**
0.6315***
(0.207) (0.210) (0.210)
(0.213)
Black× Interracial -2.5134** -2.6890***
(0.981) (0.982)
Credit Score -0.0034*** -0.0035*** -0.0035***
(0.001) (0.001) (0.001)
Constant -1.0384*** -1.0696*** -1.0906*** -0.9091*** -0.9403*** -0.9627***
(0.096) (0.100) (0.100) (0.098) (0.102) (0.103)
Observations 11,249 10,832 10,832 11,206 10,789 10,789
51
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 226 of 801
Table XII: Oaxaca-Blinder Decompositions of Business Size
This table presents Oaxaca-Blinder decompositions of size differences in businesses in the
final survey year based on whether or not the founder is black. The upper panel reports
differences in the mean values of black-owned and white-owned size, expressed both in
dollars and in log size. The bottom panel decomposes the mean difference into amounts
explained by each set of independent variables. Human capital controls include education,
previous work experience and previous startup experience. Business characteristics control
for whether the business sells a product or a service (or both), whether it has intellectual
property, and whether the business is fulltime or parttime, whether it is home-based, incor-
porated and has employees.
(1) (2)
Average Business Size, 2011: Dollar Value Log(Size)
White-Owned $533,726.05 11.693
Black-Owned $197,634.84 10.845
Difference $336,091.21 0.848
(1) (2)
Explanatory Components: Dollar Value Log(Size)
Race and Gender $5,459.03 0.047
Human Capital $40,695.11 0.059
Business Characteristics $4,966.81 0.119
Business Credit Score $51,105.07 0.175
Founder Net Worth $59,158.65 0.097
Total Explained $161,384.67 0.5510
52
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 227 of 801
A Appendix
This appendix presents information on financing, and racial differences in owner and
business characteristics in the KFS.
Table A.1 describes the detailed financing choices in the founding year for the firms
in our sample (2004). The detailed sources are grouped into six broad categories, based
on the source of the capital and the type of capital following Robb and Robinson (2014).
These are (owner, informal, formal)×(debt, equity). The first column, labelled “Full KFS”,
includes all 4,928 firms in the Kauffman Firm Survey. For some of these firms, it cannot
be verified that they either went out of business or remain in operations, therefore in the
remaining columns we include 3,972 firms that either survived over the followup years
1-3 period or were verified as going out of business over the same period. This Column is
labelled “Analysis Sample.” These two columns report means that include firms with $0
amounts of a particular source of capital. The third column, labelled “Mean if > 0” reports
the mean, in dollars, for only firms with positive amounts of that source of funding. The
number of respondents reporting a positive amount of each source of funding is reported
in the final column.
Table A.2 provides summary statistics for key variables based on the race of the firm
owner.
53
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 228 of 801
Table A.1: Detailed Sources of Financing for All Startups in the KFS
Full Analysis
Category Funding Source KFS Sample Mean if > 0 Count
Owner Equity 33,640 31,734 40,536 3,093
Total Owner Debt: 4,952 5,037 15,765 1,241
Personal Credit Card balance, owner 2,812 2,811 9,375 1,158
Personal Credit Card balance, others 1,906 238 7,415 132
Personal loan, other owners 235 1,989 124,124 67
Total Insider Equity: 2,221 2,102 44,956 177
Spouse equity 524 646 40,436 62
Parent equity 1,697 1,456 42,509 126
Total Insider Debt: 7,257 6,362 47,873 480
Family loan 2,760 2,749 29,232 327
Family loan to other owners 1,719 284 34,509 29
Personal loan to other owners 272 550 28,988 73
Other personal loans 649 924 81,452 45
Business loan by family 1,156 1,760 57,207 115
Business loan by owner 635 15 9,411 5
Business loan by other employees 52 79 22,198 9
Total Outsider Equity: 19,257 15,935 354,540 205
Angels and other investors 5,148 6,350 244,707 110
Business equity 6,621 3,645 321,351 56
Govt. equity 5,242 798 146,624 27
VC equity 701 4,804 1,162,898 26
Other equity 1,546 337 187,046 8
Total Outsider Debt: 50,130 47,847 128,706 1,439
Personal bank loan 18,031 15,859 92,433 641
Owner business credit card 16,213 1,009 7,107 543
Personal bank loan by other owners 5,017 1,859 80,650 92
Business credit card 4,227 812 6,976 452
Other Business credit cards 2,275 135 7,852 62
Business bank loans 1,591 17,075 261,358 243
Credit line balance 1,030 5,057 95,058 210
Nonbank business loan 133 3,627 214,920 72
Business loan from Govt. 857 1,331 154,743 34
Other business loan 241 231 78,281 19
Other individual loan 206 226 43,202 22
Other debt 308 626 119,493 22
Total Financial Capital 117,458 109,016 121,981 3,536
54
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 229 of 801
Table A.2: Summary Statistics by Race
Overall Mean White Mean Black Mean p-value(diff)
KFS Initial Survey Year
Female 0.31 0.30 0.36 0.00
Yrs. Work Experience 11.70 11.88 9.91 0.00
Yrs. Non-Work Experience 13.54 13.57 13.23 0.21
Previous Startup Experience 0.43 0.43 0.38 0.45
Attended Some College 0.36 0.36 0.48 0.00
Graduated College 0.30 0.31 0.24 0.05
Graduate Degree 0.17 0.18 0.16 0.54
Credit Score 35.99 36.50 30.47 0.00
KFS Survey Years 1-3
Female 0.30 0.29 0.35 0.00
Yrs. Work Experience 12.07 12.25 10.11 0.00
Yrs. Non-Work Experience 13.30 13.33 13.06 0.09
Previous Startup Experience 0.43 0.44 0.38 0.43
Attended Some College 0.36 0.36 0.48 0.00
Graduated College 0.31 0.32 0.26 0.00
Graduate Degree 0.18 0.19 0.16 1.00
Credit Score 41.39 42.27 32.28 0.00
KFS Survey Years 4-7
Female 0.30 0.29 0.36 0.00
Yrs. Work Experience 12.70 12.84 11.12 0.00
Yrs. Non-Work Experience 12.73 12.76 12.35 0.21
Previous Startup Experience 0.44 0.44 0.40 0.43
Attended Some College 0.35 0.35 0.47 0.00
Graduated College 0.33 0.34 0.26 0.00
Graduate Degree 0.19 0.20 0.16 0.49
Credit Score 52.88 54.51 35.80 0.00
Net Worth: Neg. or Zero 0.07 0.06 0.18 0.00
Net Worth: $1-50,000 0.15 0.13 0.30 0.00
Net Worth: $51,000-100,000 0.14 0.14 0.16 0.01
Net Worth: $100,001-250,000 0.18 0.19 0.13 0.00
Net Worth: $250,000+ 0.39 0.42 0.16 0.00
Net Worth: Missing 0.07 0.07 0.08 0.23
55
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 230 of 801
Table A.3: Summary Statistics on Key Firm and Founder Characteristics
Survey Wave:
Variable Group 0 1 2 3 4 5 6 7
% Black 9.1 8.8 9.0 8.4 8.1 7.9 8.4 8.4
% Survived ALL 100 91 78 65 60 54 48 45
(0.0) (0.5) (0.8) (0.9) (0.9) (0.9) (0.9) (0.9)
White 100 91 79 66 61 55 49 45
(0.0) (0.6) (0.8) (1.0) (1.0) (1.0) (1.0) (0.9)
Black 100 88 73 56 50 45 42 40
(0.0) (2.2) (3.0) (3.4) (3.3) (3.2) (3.1) (3.0)
Total Assets ALL 152917 223911 274108 296380 306185 340459 340769 343096
( 12797) ( 17935) ( 22636) ( 26416) ( 27481) ( 30976) ( 31799) ( 31094)
White 166211 239464 277122 311040 318626 356439 348634 360265
( 14269) ( 19894) ( 23123) ( 28931) ( 30178) ( 33909) ( 33568) ( 34125)
Black 40942 74559 194978 93689 139103 141007 148748 177364
( 10372) ( 16548) ( 82296) ( 22010) ( 40682) ( 50197) ( 43440) ( 63926)
Outside Debt ALL 51847 48261 55766 49380 46340 53783 48539 40157
( 4779) ( 6652) ( 14600) ( 7768) ( 6008) ( 7523) ( 9049) ( 5703)
White 56663 52368 59274 52669 45360 56523 48735 42657
( 5336) ( 7430) ( 16317) ( 8614) ( 5392) ( 8303) ( 9877) ( 6331)
Black 10809 14425 17053 12637 17301 17447 31149 15557
( 2611) ( 3648) ( 4406) ( 3300) ( 4109) ( 5366) ( 16485) ( 5992)
Outside Equity ALL 16619 19978 12666 10523 4847 11263 6512 8196
( 3369) ( 5369) ( 5174) ( 5360) ( 1399) ( 6378) ( 2638) ( 3581)
White 18543 22116 14159 11660 5330 12419 6880 8858
( 3772) ( 6007) ( 5796) ( 5981) ( 1557) ( 7084) ( 2942) ( 3994)
Black 536 604 206 912 164 678 1628 2672
( 300) ( 275) ( 106) ( 752) ( 164) ( 666) ( 1601) ( 2589)
Total Financing ALL 115233 96546 90549 79693 70293 84940 69914 64108
( 7102) ( 10098) ( 16471) ( 10893) ( 7002) ( 11458) ( 10752) ( 8745)
White 124195 103653 96096 85164 69169 89334 70518 67081
( 7898) ( 11273) ( 18394) ( 12104) ( 6602) ( 12644) ( 11724) ( 9707)
Black 35205 34462 29033 22647 32194 26015 39460 25566
( 6002) ( 5245) ( 5187) ( 3950) ( 7344) ( 6026) ( 16755) ( 6751)
% Have Out. Debt ALL 37.5 42.4 43.8 42.6 42.2 40.3 35.5 34.0
( 0.9) ( 1.0) ( 1.1) ( 1.2) ( 1.2) ( 1.2) ( 1.3) ( 1.3)
White 38.5 42.9 44.6 43.4 43.0 40.6 36.2 34.7
( 0.9) ( 1.0) ( 1.1) ( 1.2) ( 1.3) ( 1.3) ( 1.3) ( 1.3)
Black 29.4 37.7 37.7 36.9 35.1 35.4 28.3 28.2
( 2.7) ( 3.4) ( 3.7) ( 4.3) ( 4.3) ( 4.7) ( 4.2) ( 4.4)
% Out. Debt Ratio ALL 49.4 63.1 69.5 71.2 72.7 73.6 77.5 75.8
(Out. Debt≥0) ( 0.9) ( 1.0) ( 1.0) ( 1.2) ( 1.2) ( 1.3) ( 1.3) ( 1.5)
White 50.1 64.2 70.1 72.5 73.6 73.7 78.3 77.3
( 1.0) ( 1.1) ( 1.1) ( 1.2) ( 1.2) ( 1.3) ( 1.4) ( 1.5)
Black 41.8 52.4 61.5 55.2 62.6 66.6 66.0 58.7
( 3.1) ( 4.0) ( 3.4) ( 5.1) ( 4.8) ( 5.5) ( 6.5) ( 6.3)
56
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 231 of 801
(1) (2) (3) (4
VARIABLES survive survive survive netpro
black -0.199*** -0.014 0.038** -53,72
(0.035) (0.037) (0.018) (48,306
score 0.011*** 0.002***
(0.000) (0.000)
Racial Differences in Profitability and Survival
Constant -0.109*** -0.643*** 2,399
(0.023) (0.030) (5,367
Observations 28,527 28,309 30,608 23,1
R-squared 0.609 0.0
Standard errors in paren
*** p<0.01, ** p<0.05, *
57
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 232 of 801
Table A.4: A Closer Look at Sources of Debt
This table reports survey-weighted mean values by race for dummy variables that track the use of
particular types of credit, as well as for mean values of these sources of credit. The final column
reports p-values from the t-test of the difference between black- and white-owned businesses.
The first block of numbers for each year range (“Uses”) reports the proportion of the sample that
indicates they use that form of debt. The lower block of numbers for each year range (indicated
with $) reports the average dollar amounts for that funding category.
Overall White Mean Black Mean p-value(diff)
KFS Initial Survey Year
Uses Personal Credit Cards 0.48 0.49 0.34 0.00
Uses Personal Bank Loan 0.18 0.18 0.14 0.01
Uses Business Credit Cards 0.28 0.30 0.15 0.00
Uses Loans from Family Members 0.10 0.09 0.14 0.00
Uses Business Bank Loans 0.06 0.07 0.01 0.00
Personal Bank Loan ($) 13,660 14,497 6,971 0.04
Personal Loans from Fam. ($) 2,465 2,571 1,801 0.36
Personal Loans, Other Sources ($) 4,360 4,659 2,161 0.24
Business Bank Loan ($) 9,540 10,551 1,106 0.03
Business Non-bank Loans ($) 5,510 6,035 866 0.06
KFS Survey Years 1-3
Uses Personal Credit Cards 0.38 0.38 0.35 0.84
Uses Personal Bank Loan 0.13 0.13 0.08 0.01
Uses Business Credit Cards 0.42 0.43 0.32 0.00
Uses Loans from Family Members 0.07 0.07 0.12 0.00
Uses Business Bank Loans 0.06 0.06 0.03 0.02
Personal Bank Loan ($) 7,992 8,228 4,171 0.05
Personal Loans from Fam. ($) 1,454 1,491 1,323 0.17
Personal Loans, Other Sources ($) 1,999 2,070 1,451 0.60
Business Bank Loan ($) 5,039 5,589 625 0.01
Business Non-bank Loans ($) 2,933 3,085 742 0.08
KFS Survey Years 4-7
Uses Personal Credit Cards 0.33 0.33 0.30 0.23
Uses Personal Bank Loan 0.08 0.09 0.05 0.08
Uses Business Credit Cards 0.41 0.43 0.27 0.00
Uses Loans from Family Members 0.06 0.06 0.08 0.00
Uses Business Bank Loans 0.05 0.05 0.02 0.08
Personal Bank Loan ($) 2,523 2,719 635 0.04
Personal Loans from Fam. ($) 677 702 298 0.34
Personal Loans, Other Sources ($) 944 973 343 0.58
Business Bank Loan ($) 2,589 2,624 1,392 0.25
Business Non-bank Loans ($) 1,484 1,507 521 0.43
58
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 233 of 801
Table A.5: Later Stage Differences in Access to Capital with Zipcode Fixed Effects
(1) (2) (3) (4) (5) (6)
VARIABLES 2005-2007 2008-2011 2005-2007 2008-2011 2005-2007 2008-2011
Black -0.317** -0.265* -0.462** -0.511** 0.120 -0.473*
(0.133) (0.151) (0.211) (0.225) (0.226) (0.244)
Asian -0.121 0.075 -0.302 -0.810*** 0.154 -0.068
(0.160) (0.174) (0.255) (0.260) (0.269) (0.279)
Hispanic -0.296** 0.116 -0.439** 0.291 0.097 0.417
(0.140) (0.167) (0.223) (0.249) (0.238) (0.267)
Other 0.064 -0.376 -1.276*** -0.773** -0.707** -0.084
(0.198) (0.263) (0.316) (0.392) (0.337) (0.431)
Female -0.207*** -0.107 -0.139 -0.127 -0.178 -0.265**
(0.073) (0.084) (0.117) (0.125) (0.124) (0.135)
Previous Industry Experience -0.001 0.004 0.002 -0.009 0.011* -0.003
(0.003) (0.004) (0.006) (0.006) (0.006) (0.006)
Experience Outside Industry -0.000 -0.003 0.015*** 0.004 0.019*** 0.013**
(0.003) (0.004) (0.005) (0.006) (0.005) (0.006)
Some College 0.113 0.182 0.156 0.109 0.229 -0.046
(0.111) (0.137) (0.176) (0.205) (0.187) (0.220)
College Deg. 0.025 0.020 0.061 -0.303 0.209 -0.325
(0.118) (0.146) (0.188) (0.217) (0.200) (0.233)
Grad. Deg. -0.042 -0.029 -0.027 -0.239 0.146 -0.305
(0.123) (0.148) (0.195) (0.222) (0.208) (0.238)
Prev. Startup Exp. -0.088 0.035 0.178* 0.262** 0.188* 0.174
(0.060) (0.068) (0.096) (0.102) (0.102) (0.110)
Makes Product 0.080* -0.009 0.204*** -0.002 0.364*** 0.109
(0.044) (0.046) (0.070) (0.068) (0.074) (0.073)
Intel. Property 0.028 -0.028 0.156** 0.145** 0.481*** 0.203***
(0.045) (0.049) (0.072) (0.073) (0.076) (0.078)
Home-based -0.086 -0.167** -0.421*** -0.575*** -0.832*** -0.885***
(0.059) (0.066) (0.094) (0.099) (0.100) (0.106)
Part time Bus. -0.121 0.084 -0.743*** 0.088 -0.614*** -0.208
(0.074) (0.083) (0.118) (0.124) (0.125) (0.133)
Incorporated -0.029 0.037 0.340*** 0.367*** 0.390*** 0.247**
(0.063) (0.073) (0.100) (0.109) (0.106) (0.117)
Employment 0.008*** 0.013*** 0.009*** 0.008** 0.008*** 0.009**
(0.001) (0.002) (0.002) (0.003) (0.002) (0.004)
Credit Score 0.002** 0.002* 0.003** 0.006*** 0.001 0.003*
(0.001) (0.001) (0.001) (0.001) (0.001) (0.002)
Observations 7,903 7,819 7,903 7,819 7,736 7,790
R-squared 0.531 0.501 0.640 0.659 0.650 0.661
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
59
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 234 of 801
Table A.6: Local Banking Conditions and Racial Bias
Panel A reports regressions (Pooled OLS with year dummies) in which the dependent variable is the natural log of total business debt.
Panel B reports Probit regressions in which the dependent variable is a dummy if the respondent had unmet capital needs, which
is true if the respondent answered yes to “Did Not Apply for Fear of Rejection” or if they reported that they did not always get the
full amount they asked for. Local bank share is the share of total county-level deposits held by local banks. Local bank herfindahl is
the herfindahl of local banks at the county level computed using deposits. Standard errors in parentheses, with one, two and three
asterisks denoting significance at the 10, 5 and 1percent level. Controls from Table IV included but not shown.
Panel A: Dependent variable is Did Not Apply for Fear of Denial
Black-owned Startup 0.7605*** 0.7571*** 0.7238*** 0.7117*** 0.5778***
(0.059) (0.059) (0.077) (0.078) (0.128)
Local Bank Share -0.2623** -0.2881*** -0.2594** -0.2557**
(0.105) (0.111) (0.115) (0.116)
Local Bank Share × Minority 0.2358 0.3086 0.1578
(0.366) (0.381) (0.387)
Local Bank Herfindahl -0.5995*** -0.6727***
(0.191) (0.203)
Bank Conc. × Minority 0.9212
(0.706)
Credit Score -0.0035*** -0.0035*** -0.0035*** -0.0035*** -0.0035***
(0.001) (0.001) (0.001) (0.001) (0.001)
Observations 11,204 11,193 11,193 11,193 11,193
Panel B: Dependent variable is Log(Business Debt)
(1) (2) (3) (4) (5)
Black-owned Startup -0.1604*** -0.1527*** -0.1071*** -0.1044*** -0.0241
(0.020) (0.020) (0.025) (0.025) (0.037)
Local Bank Share 0.3285*** 0.3508*** 0.3354*** 0.3333***
(0.060) (0.063) (0.064) (0.064)
Local Bank Share × Minority -0.3260** -0.3357** -0.2801**
(0.131) (0.132) (0.132)
Local Bank Herfindahl 0.2481** 0.2798***
(0.101) (0.107)
Bank Conc. × Minority -0.5215***
(0.159)
Credit Score 0.0038*** 0.0038*** 0.0038*** 0.0037*** 0.0038***
(0.000) (0.000) (0.000) (0.000) (0.000)
Observations 21,441 21,412 21,412 21,412 21,412
R-squared 0.020 0.022 0.022 0.023 0.023
60
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 235 of 801
Exhibit G
620484
research-article2015
EDQXXX10.1177/0891242415620484Economic Development QuarterlyBates and Robb
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 236 of 801
Research and Practice
Economic Development Quarterly
Impacts of Owner Race and
2016, Vol. 30(2) 159–170
© The Author(s) 2015
Reprints and permissions:
Geographic Context on Access sagepub.com/journalsPermissions.nav
DOI: 10.1177/0891242415620484
to Small-Business Financing edq.sagepub.com
Timothy Bates1 and Alicia Robb2
Abstract
Multiple studies document both the scarcity of small-business financing in inner-city minority communities and the higher
loan application rejection rates among minority business enterprises (MBEs), compared with equally creditworthy White-
owned firms. When MBEs do receive bank financing, they get smaller loans than Whites. Two aspects of these findings are
troublesome. First, urban MBEs are heavily concentrated geographically in minority communities, raising the issue of whether
difficulties accessing financing reflect firm location, minority ownership, or both. Second, applicable studies rest on weak
theoretical foundations. The authors’ findings suggest that banks engage in discriminatory practices limiting credit availability
to MBEs. Controlling for risk factors, however, firm location in a minority or inner-city neighborhood has no apparent impact
on loan availability or size. Owner race/ethnicity, in contrast, is important. Subtle processes discourage MBEs from seeking
bank loans. Owner race and wealth both powerfully shape loan access: high wealth opens doors, minority ownership closes
them.
Keywords
bank loans, racial differentials, redlining, minority neighborhood firms
Small businesses located in inner-city minority neighborhoods Background: Discriminatory Practices
undisputedly face multiple difficulties when they seek com- Limiting Access to Bank Loans
mercial bank financing. Neighborhoods, where poverty rates
are high and minority residents are numerous, have tradition- Access to bank loans is important because financial institu-
ally been viewed as unattractive markets by risk-averse finan- tions provide more debt capital to small businesses than all
cial institutions. This much we agree on. Exactly why local other sources combined. According to the Federal Reserve
firms struggle to obtain financing in these environs is unclear. Board’s Survey of Small Business Finances (SSBF) data,
Varying degrees of bank redlining in minority neighborhoods banks account for over 90% of all small-business debt
have been documented in past studies, but those findings have financing. Very young firms and minority business enter-
not ruled out alternative explanations for loan scarcity. Bank prises (MBEs) of all ages, although somewhat less reliant on
lending patterns may be because of firm location in minority bank loans than others, nonetheless obtain the majority of
communities or, alternatively, lender discriminatory treatment their debt capital from banks (Bates, 2011).
of subgroups (minority owners) overrepresented in minority Passage of the 1977 Community Reinvestment Act (CRA)
neighborhoods, greater applicant credit risk, low credit provides our analytical starting point. The CRA sought to
demand, or a combination of these factors. Our study probes make banks responsive to the credit needs of the communi-
these explanations to develop a better understanding of why ties in which they operate. A key CRA rationale was to pres-
small-business credit availability is limited. sure financial institutions to expand their provision of loans
Our primary objective is to investigate causes of small- to support the housing and economic development needs of
business racial and spatial differences in unmet credit needs urban minority neighborhoods. Although serious CRA
and differentials in loan terms while proposing theoretical
justifications for these differentials: We seek to disentangle 1
Wayne State University, Detroit, MI, USA
the effects of race and space. Using Kauffman Firm Survey 2
Ewing Marion Kauffman Foundation, Kansas City, MO, USA
(KFS) data to investigate loan accessibility among urban
Corresponding Author:
small businesses, our analysis utilizes firm and owner char- Timothy Bates, Wayne State University, 42 W Warren Avenue, Detroit,
acteristics, controls for neighborhood economic well-being, MI 48202, USA.
and firm-specific geographic location identifiers. Email: tbates7893@gmail.com
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 237 of 801
160 Economic Development Quarterly 30(2)
enforcement by regulatory authorities began only in the that a $1 billion increase in bank assets was associated with
1990s, these actions did ultimately pressure banks to reex- a 5% increase in a high CRA rating likelihood. Large banks
amine their lending policies, leading gradually to greater are a growing presence in many minority neighborhoods and
loan availability for minority neighborhood small businesses their branches sometimes actively target minority clients.
(Bates & Robb, 2015; Immergluck, 2004). Complementary strategies entail staffing branches in minor-
Since the CRA’s passage, the banking industry has under- ity communities with minority employees and monitoring
gone profound changes, including the increasing concentra- branches to ensure employees are sensitive to the banking
tion of assets into a small number of institutions, each holding needs of minority clients (Lubin, 2008).
hundreds of billions in assets. Small banks in urban America Certain facts, nonetheless, are clear: MBEs account for
were often displaced by branches of these huge institutions. nearly half of all small firms located in urban minority neigh-
In the process, lending policies evolved for a variety of rea- borhoods nationwide (Bates & Robb, 2014). MBE owners,
sons. Community activism was certainly a catalyst. Effective second, view their access to financing as severely con-
regulation under the CRA was often instigated by neighbor- strained. The U.S. Chamber of Commerce surveyed 1,080
hood associations, civil rights groups, community develop- small business owners: 19% of responding Whites identified
ment corporations, and coalitions of these groups (Bates & access to credit as their top problem, compared with 54% for
Robb, 2015; Immergluck, 2004). minority owners (U.S. Chamber of Commerce, 2005, cited
Under the CRA, activist groups like the Woodstock in Blanchflower, 2009). Influential studies have documented
Institute of Chicago could intervene when banks requested small-business financing scarcity in inner-city minority com-
permission to expand operations by acquiring another bank munities (Immergluck, 2002, 2004; Smith, 2003), while
or opening a branch. This was done by opposing banker complementary studies documented the higher loan-applica-
requests on the grounds of their failure to meet the credit tion rejection rates of MBEs, in comparison to equally cred-
needs of local communities (Casey, 2009). The threat of for- itworthy White-owned firms (Blanchflower, Levine, &
mal challenges often motivated large banks to seek relation- Zimmerman, 2003; Mitchell & Pierce, 2011). When MBEs
ships with activists. CRA agreements thus emerged whereby do receive bank financing, they pay higher interest rates and
activist groups pressured banks and the latter agreed to alter receive smaller loans than their White counterparts (Bates &
lending policies to avoid challenges (Immergluck, 2004). Robb, 2013).
Bank treatment of minority clients was typically the issue of A common initial objective of firm owners seeking busi-
contention, and CRA challenges were concentrated in cities ness loans is to identify bank-lending criteria. In their audit
where residents were predominantly minorities. Presence of study of small business owners seeking bank loans, Bone,
a CRA agreement in a metropolitan area, according to Casey, Christensen, and Williams (2014) focused directly on this
Glasberg, and Beeman (2011), encouraged Black and Latino inquiry stage and found that Black and Latino owners were
residents—but not Whites—to submit more loan applica- treated differently than matched Whites. Typifying audit
tions to banks, compared with residents in areas lacking studies, the White and minority testers were matched regard-
agreements. “Community activism forced regulators to cre- ing age, gender, credit history, personal net worth, character-
ate institutional processes such as public hearings and the istics of the loans being sought, and other traits, and their
consideration of CRA agreements in assessing CRA perfor- differential treatment was strongly consistent with minority
mance under the law” (Immergluck, 2004, p. 167). owners being treated worse than Whites.
The large banks currently predominant in many urban A key objective of this study was to determine whether
regions are intensely concerned about their public images the testers encountered similar scrutiny from loan officers
and therefore seek outstanding CRA ratings. Regulatory when applying for loans. In comparison to White testers,
authorities have assigned to them, without exception, the minorities were more often asked to provide business finan-
highest possible ratings (Immergluck, 2004). Focusing pub- cial statements (83% vs. 50%), income tax returns (86% vs.
lic attention on bank-lending practices in minority neighbor- 52%), bank account information (25% vs. 0%), personal
hoods has encouraged banks to adopt more liberal lending financial asset details (60% vs. 22%), and credit card debt
policies (Bates & Robb, 2015; Immergluck, 2004). Whether (42% vs. 13%). Additionally, minorities were offered less
the public spotlight is focused on bank-lending policies by frequent assistance than Whites in completing loan applica-
community activists, the mass media, articles in scholarly tions (18% vs. 59%), and loan officers offered business cards
journals, bank regulatory authorities, Michael Porter, the to minority testers less often (43% vs. 82%; Bone et al.,
Initiative for a Competitive Inner City (ICIC), or other activ- 2014). Overall, minorities were consistently offered less
ist groups, the consequence appears to be wider credit avail- assistance and subjected to greater scrutiny, in comparison
ability in minority neighborhoods. Among banks sensitive with the White testers.
about their public images, high CRA ratings convey multiple Bates and Robb (2015) analyzed KFS data describing
advantages, including the immunity it provides from activist firm and owner traits to determine both the extent of unmet
challenges. Stegman, Cochran, and Farris (2002) estimated small-business credit needs and outcomes of their
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 238 of 801
Bates and Robb 161
applications for bank loans over the 2008 to 2011 period. issue emphasized by SSBF-based studies is whether Black
Nationwide, Black-, Latino-, and Asian-owned firms, com- firms (and MBEs generally) possessing identical risk-related
pared with Whites, were more likely to be discouraged bor- traits (other than race) have less access to bank credit than
rowers and more likely to have their applications denied by Whites. The consistent finding is that Black (and Latino)
banks, controlling for credit-risk factors. Examining firms in applicants experience a higher incidence of loan denials and
urban areas only, they next compared these same outcomes pay higher interest rates than Whites (Blanchflower, 2009;
in differing geographic contexts—minority neighborhoods Blanchflower et al., 2003; Cavalluzzo & Wolken, 2005).
and other urban areas—while controlling for firm and owner Furthermore, among SSBF firms needing credit, owners
characteristics, neighborhood traits, and other risk-related often report not applying for loans because they fear rejec-
factors. Regarding unmet credit needs, minority neighbor- tion. After controlling statistically for the presumed greater
hood location had no impact on outcomes, although owner credit risks posed by MBEs needing but not applying for
race/ethnicity powerfully explained the incidence of discour- credit, Blanchflower et al. (2003) found remaining gaps of
aged borrowers (those with unmet credit needs). Among oth- 26 and 15 percentage points, respectively, for Blacks and
erwise equally creditworthy firms, those owned by African Hispanics (relative to Whites).2
Americans, Asian Americans, and Latinos had significantly Bostic and Lampani (1999), like other SSBF-based stud-
higher unmet credit needs compared with Whites, and these ies, found statistically significant disparities in Black/White
outcomes characterized firms operating in both minority loan approvals after controlling for risk factors, but they
neighborhoods and other urban environs. found, as well, that firm location was a determinant of loan
Among small firms actually applying for bank loans, denials. The economic and demographic characteristics of a
however, very different findings emerged. Logistic regres- firm’s local geography, they concluded, was important:
sion analysis revealed that MBEs located in minority neigh- Black applicants were rejected disproportionately, in part,
borhoods were actually more likely to have their applications because their firms were often located in economically
approved by banks than minority businesses located else- depressed African American residential areas.
where in urban America (Bates & Robb, 2015). They attrib- Pursuing small-firm location as a determinant of appli-
uted this unexpected finding to banker sensitivity about cant outcomes, Smith (2003) analyzed CRA data describing
meeting their CRA obligations. In the minority neighbor- bank lending in Chicago to determine if loan availability dif-
hood context, CRA requirements are apparently being taken fered in predominantly minority neighborhoods and nonmi-
seriously by bankers. This finding is, nonetheless, counterin- nority areas. Smith disaggregated the city into 77
tuitive, in light of companion findings of higher unmet credit neighborhoods (zip codes) and examined loan availability
needs among MBEs. Findings of the Bone et al. (2014) audit relative to the number of small businesses. He found that
study, however, provide a logical explanation of this appar- loan availability and size in lower income minority areas
ent paradox. The harsh scrutiny potential MBE applicants were relatively lower than in higher income, predominantly
are subjected to at the loan inquiry stage, combined with loan White areas. Loan availability was highest in Forest Glen, an
officers offering less assistance to minority owners, com- affluent White neighborhood, and lowest in Washington
pared with Whites, appears to suppress the incidence of loan Park, a low-income African American neighborhood. A firm
applications actually being filed by MBEs. MBE-discouraged in Forest Glen was 3.04 times more likely to receive bank
borrowers are numerous, in part because the applicant review credit than one in Washington Park; average loan size was
practices of loan officers actively discourage potential MBE $39,600 in the former and $9,500 in the latter. Short-term,
applicants from actually applying for loans.1 expensive credit-card borrowings were predominant among
Older studies based on nationwide SSBF data consis- Washington Park firms. Neighborhoods lowest in loan avail-
tently find large denial-rate differences across business ability and size rankings were all low-income minority
applicant groups of different race/ethnicities. Examining neighborhoods (Smith, 2003).
their most recent loan application, only 9% of White appli- Immergluck’s (2002) analysis of small-business lending
cants for non-line-of-credit bank loans were denied versus in Philadelphia found similar patterns. Small firms in pre-
56% of Blacks and 63% of Hispanics (Mitchell & Pierce, dominantly White census tracts received, on average, 11.0
2011). Over the time frame covered by 2003 SSBF data, the bank loans per 100 businesses (excluding credit card bal-
average loan size was $89,833 (White-owned firms) and ances), versus 1.2 loans in African American tracts.
$53,843 for minorities among firms with annual sales under Controlling for family income, business credit score, and
$500,000. Corresponding loan mean interest rates were other characteristics, going from an all-White to an all-Black
6.9% and 9.1%, respectively, for White borrowers and neighborhood resulted in a drop of 6.8 loans per 100 firms.
MBEs (Bates & Robb, 2013). Complementary findings of Immergluck (2002), Smith
Black-owned firms have been more widely studied than (2003), and Bostic and Lampani (1999), in summary, all por-
other MBEs and their difficulty accessing financing is not tray minority neighborhoods as areas where credit availabil-
controversial (see Bates, 1973; Blanchflower, 2009). The ity is constrained in comparison with loan access elsewhere.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 239 of 801
162 Economic Development Quarterly 30(2)
Discriminatory Processes: Redlining, conceivably make sense, if it was otherwise difficult for
Racial Discrimination, or Both? bankers to assess credit risks efficiently.
Bankers may also practice statistical discrimination in
Mainstream finance theory studies loan access by assuming their dealings with MBE clients. Past studies demonstrate
that debt providers have complete, accurate information that in markets where shoppers incur substantial search costs,
about businesses seeking loans (Parker, 2009). In reality, minorities often fare badly, relative to Whites. Their dispa-
small businesses are opaque: Banks financing these ventures rate treatment is caused in part by sellers’ inferences about
have imperfect information about entrepreneurs’ abilities racial differences in customer reservation prices—higher
and prospects for profitable operation. Lenders, additionally, among Blacks than Whites (Salop & Stiglitz, 1977).
are vulnerable to opportunistic defaults, often rooted in Reservation prices are, by definition, the prices (and terms)
moral-hazard issues (Aghion & Bolton, 1997). Business customers are willing to pay for specific products (bank
owners best positioned to attract bank loans are high-net- loans, new cars). Among well-defined customer groups dif-
worth individuals: Abundant personal wealth serves as col- fering in one trait (Whites, minorities), one group accepting
lateral, alleviating banker moral-hazard concerns, less favorable prices/terms than the other implies that reser-
encouraging loan approval, and larger loans at lower interest vation price differentials exist. Stated differently, sellers may
rates (Aghion & Bolton, 1997; Cagnetti & De Nardi, 2006). decide to exploit racial differentials in reservation prices by
Although existing evidence suggests that owner race is an charging minority customers higher prices than otherwise
independent factor shaping banker decisions about business identical White customers because it is profitable to do so.
lending, key questions remain. If owner race, other factors In their classic audit study of new-car buying, for exam-
being equal, truly affects loan availability and lending terms, ple, Ayres and Siegelman (1995) concluded that the higher
then understanding the how and why of these impacts raises new car prices quoted to African American shoppers were
deeper questions: Why does owner race matter? Anchoring not based on a psychological distaste for dealing with Blacks
banker risk aversion in uncertainty about applicant profit- but on their relatively high reservation prices: “Even a mar-
ability, entrepreneurs’ abilities, and moral-hazard concerns ket in which no participants are prejudiced might exhibit dis-
seems straightforward, but how is owner race causally related crimination if the dealers use buyers’ race or gender to make
to these factors? Is “minority owner” a proxy for inadequate statistical inferences about the expected profitability of sell-
expertise? Moral-hazard concerns? Gunnar Myrdal in the ing to them” (p. 304). Dealers sought higher prices from
1940s observed Black shoppers because it was profitable to do so. Auto deal-
erships owned by African Americans quoted the same higher
the Negro businessman encounters greater difficulties than prices as White dealers, consistent with the higher average
Whites in securing credit. This is partially due to the marginal
reservation prices of Black shoppers, compared with Whites.
position of Negro business. It is also partially due to the
Audit study findings of Bone et al. (2014) indicate that
prejudicial opinions among Whites concerning the business
ability and personal reliability of Negroes. (Myrdal, 1944, starkly differential treatment is real in the experiences of
p. 308) minorities investigating small-business financing sources.
By themselves, these findings provide no direct evidence of
The rationale for banker aversion to lending in minority racial reservation price differentials regarding loan terms.
neighborhoods is altogether different than motivations for What they do provide is audit study evidence of minorities
discriminating against MBE borrowers. In his classic study, being treated badly, compared with Whites. In this sense,
Jackson (1985) traced the historical origins of bank redlin- they confirm, with control-group precision, past findings that
ing. Institutionalized in the 1930s was the notion that neigh- banks treat MBEs badly, relative to equally creditworthy
borhoods where “racially inharmonious groups” resided Whites. Studies using regression analysis to demonstrate dis-
were risky places for banks to lend. Color-coded maps were proportionate bank rejection of minority loan applicants,
used subsequently by banks and government agencies to their unfavorable loan terms, and high discouraged-borrower
rank neighborhoods according to their degree of lending risk. incidence are all subject to omitted variable-bias criticisms
Racially mixed (and African American only) areas were (Becker, 1993). No such criticisms apply to the audit study
shaded in red, signifying the highest risk: They were “red- findings of Bone et al. (2014).
lined” and loan availability was minimal, irrespective of Five types of evidence, in summary, suggest that MBEs
other borrower credit-risk factors. expect to encounter difficulty when seeking bank financing.
Providing a theoretical framework explaining why First, surveys indicate that minority owners are more likely
redlining might be practiced by profit-oriented banks, than Whites to view credit access as their top problem.
Stiglitz and Weiss (1981) argued, “It is difficult for banks Second, studies measuring discouraged-borrower incidence
to identify ‘good’ borrowers, and to do so requires a vari- indicate that minority owners, fearing denial, are less likely
ety of screening devices” (p. 393). Using the racially than equally creditworthy Whites to seek loans. Third, the
mixed neighborhood trait as a screening device could higher denial rates among MBEs applying for bank financing,
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 240 of 801
Bates and Robb 163
in comparison to otherwise identical Whites, confirm dis- Recall that discouraged borrowers are those needing loans but
couraged-borrower suspicions that banks penalize them for not applying because they feared rejection. We chose to nar-
being minorities. Fourth, compared with equally creditworthy row this definition. Owners self-identifying as discouraged
White-owned firms, Black and Hispanic borrowers pay borrowers sometimes apply for loans anyway, despite their
higher interest rates. Fifth, audit study evidence of owners rejection fears. We define discouraged borrowers narrowly as
inquiring about bank loans reveals that Blacks and Latinos those meeting two conditions: (a) they explicitly state their
receive greater scrutiny and are offered less assistance than firms need credit but nonetheless do not seek bank loans and
matched Whites.3 Expecting their search for financing to be (b) they do not subsequently apply for bank loans in the cur-
difficult shapes reservation prices of MBE borrowers. rent calendar year. Relative discouraged-borrower frequen-
Questions about underlying causes of limited MBE loan cies were 22.2% among minority neighborhood businesses
access nonetheless persist. and 14.8% for firms located in other urban environs. Among
firms actually submitting loan applications to financial insti-
tutions, 56.7% of minority area applicants and 62.0% of
Analyzing Credit Demand and Loan applicants in other urban areas reported their applications
Terms: Database, Summary Statistics, were always approved. These applicant, loan recipient, and
Hypotheses, and Econometric Findings discouraged-borrower percentages are annual percentages.
Among discouraged borrowers, average net profits were
The KFS provides the only nationally representative small- negative, minority ownership was relatively frequent, and
business database with firm-specific information on geo- nearly 38% had low credit scores (Table 1). Nearly 27% of
graphic location, loan application outcomes, and debt discouraged-borrower firm owners, however, reported per-
sources. KFS tracks firms starting operations in 2004; their sonal wealth of at least $250,000 and 48% were college grad-
strength is inclusion of annual follow-up information on uates. Breaking discouraged borrowers into minority
financing raised subsequent to start up. Tracking a cohort of neighborhood and White-neighborhood subsets, the former
new businesses through year-end 2011,4 KFS collected infor- reported lower profits than the latter, as well as lower credit
mation annually on 4,928 firms, including industry, location, scores, personal wealth holdings, and annual sales. Judged
employee numbers, sales, profits, owner characteristics, by conventional credit-risk standards, minority neighbor-
equity and debt capital, and other traits.5 Each calendar year hood firms are collectively less creditworthy than small busi-
is treated as a separate observation in the sense, for example, nesses located in other urban areas.
that a firm receiving a loan in 2008 is a different observation Compared with discouraged borrowers, bank loan recipi-
than the same firm self-identifying as a discouraged bor- ents were better credit risks, reporting higher average profits
rower in 2011. Because firm, owner, and environmental traits and sales; over 57% of the owners were college graduates,
are recorded annually, traits of the same firm often vary, owner high-wealth incidence was noteworthy, and White
depending on the period under consideration. Additional ownership incidence was high (Table 2). Bank loan recipi-
KFS details appear in Robb, Ballou, DesRoches, Potter, and ents in minority neighborhoods reported higher average loan
Zhao (2009). sizes ($138,300) in comparison to borrowers located else-
To investigate loan access in different geographic con- where ($115,000). The minority neighborhood borrowers,
texts, we sorted firms, based on zip code, into businesses in furthermore, were larger, more profitable firms, on average,
(a) predominantly minority residential areas and (b) all other than borrowing firms located in other urban environs.
urban regions. We analyzed small-business credit needs and Looking solely at MBEs irrespective of geographic loca-
borrower loan terms. Our analyses are conducted separately tion, their discouraged-borrower incidence was higher com-
for firms in (a) inner-city neighborhoods, (b) minority neigh- pared with White-owned ventures, the percentage of
borhoods, and (c) firms located anywhere in urban America. approved loan applicants was lower, and borrower loan sizes
When analyzing loan size, we focused solely on bank loan were lower. Among firms in minority neighborhoods, loan
recipients. Credit-card borrowing was excluded, as were size differentials are striking: Average loan size was $167,100
loans from informal sources, including family, friends, and/ for White-owned firms and $97,200 among MBEs.
or nonbank firms. We have not reported descriptive statistics explicitly for
firms in inner-city areas. According to an ICIC study, over
Statistics Describing Small-Business Borrowers in 80% of inner-city residents are minorities (ICIC, 2005).
There is, however, a major difference because inner-city
Urban America areas, by definition, are characterized by a high incidence of
Among the firms under consideration, those in minority poverty and relatively low household incomes. Predominantly
neighborhoods were more likely to self-identify as discour- minority neighborhoods, in contrast, may or may not fit this
aged borrowers and less likely to have their bank loan appli- high poverty/low household income profile. This distinction
cations accepted compared with firms located elsewhere. is important because small businesses—MBEs and
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 241 of 801
164 Economic Development Quarterly 30(2)
Table 1. Firms Needing Credit That Choose Not to Apply for Bank Loans: Statistics Describing Firm, Owner, and Neighborhood
Traits.
All urban firms Minority neighborhood firms White neighborhood firms
A. Firm traits
Annual sales, 1,000 (M) $396.7 $377.1 $409.1
Net profit lagged 1 year, 1,000 (M) −$11.6 −$19.7 −$6.7
Minority neighborhood 39.0% 100% 0%
B. Owner characteristics
Black 20.6% 37.9% 9.3%***
Latino 8.6% 14.9% 4.8%***
Asian 6.2% 7.3% 5.4%
College graduate 48.0% 44.6% 49.9%***
Wealth $250,000+ 26.9% 24.0% 28.9%***
Wealth $50,000-250,000 32.2% 27.9% 35.1%**
Wealth under $50,000 41.0% 48.1% 36.0%***
Credit score high 25.4% 17.0% 30.9%***
Credit score medium 36.8% 40.1% 34.4%
Credit score low 37.8% 43.0% 34.7%***
C. Neighborhood traits
Household poverty rate 9.6% 14.8% 6.3%***
N 814 301 513
Note. One, two, and three asterisks denote significance at the 10%, 5%, and 1% levels, respectively.
Source. Kauffman Firm Survey (KFS).
*p < 1.
Table 2. Bank Loan Recipients: Statistics Describing Loan Size, Firm, and Owner Traits.
All urban firms Minority neighborhood firms White neighborhood firms
A. Firm traits (M)
Annual sales, 1,000 $581.3 $667.8 $548.4***
Net profit lagged 1 year, 1,000 $34.9 $39.7 $30.9***
B. Owner characteristics (%)
Minority 20.1 40.3 11.2***
White 79.5 58.8 88.6***
College graduate 57.1 56.0 57.8
Wealth $250,000+ 44.9 38.5 47.7**
Wealth $50,000-250,000 34.0 33.3 34.6
Wealth under $50,000 21.1 28.2 17.8**
C. Loan size, $ amount
All owners 121.2 138.3 115.0**
Minority owners 100.0 97.2 108.6
White owners 126.3 167.1 115.8
N 657 173 484
Note. One, two, and three asterisks denote significance at the 10%, 5%, and 1% levels, respectively.
Source. Kauffman Firm Survey (KFS).
*p < 1.
White-owned ventures—concentrate disproportionately in of statistics. Breaking inner-city businesses into minority
minority neighborhoods where poverty rates are below the neighborhood and all-others subsets is unproductive, further-
thresholds identifying inner-city areas, and household more, because the latter is too small to support meaningful
incomes are higher. Our KFS business sample size is thus statistics. Inner-city area impacts on discouraged-borrower
considerably smaller for inner-city firms compared with incidence and borrower loan sizes are investigated in the
minority neighborhood ventures, thus reducing the reliability econometric findings section of this study.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 242 of 801
Bates and Robb 165
Hypotheses America, other factors being equal, then bankers would logi-
cally protect themselves from these risks by offering less
Our primary objective is to investigate small-business racial attractive terms to loan recipients in these areas. These terms
and spatial differences in (a) discouraged-borrower inci- partially compensate bankers for their increased risk expo-
dence and (b) differentials in loan terms of bank borrowers. sure. Smaller loans lessen potential losses in the event of
Past evidence suggests that owner race and business place do default since fewer loan dollars are at risk and borrower col-
affect business access to bank loans, yet key questions remain lateral coverage is greater, relative to loan size, thus increas-
unanswered. Which among these factors really matters— ing expected recoupment of the loan balance in the event of
business location, owner race, both, neither? Although banks borrower default.
face regulatory agency requirements to serve the credit needs
of minority neighborhood firms and activist demands to ful- Hypothesis 4: Among equally creditworthy firms receiv-
fill their CRA obligations, their lending decisions are none- ing bank loans, MBEs are likely to receive loans smaller
theless shaped by risk aversion reflecting uncertainty about in dollar amount than White-owned firms.
borrower repayment capacity and owner intent to repay.
How do these diverse factors play out in practice? Assuming minority ownership is an independent factor
heightening default risk, offering less attractive loan terms
Hypothesis 1: Among equally creditworthy firms, busi- partially compensates bankers for their greater risk, as
nesses in minority neighborhoods are more likely than explained in Hypothesis (3). Alternatively, bankers may
those located elsewhere to be discouraged borrowers. engage in statistical discrimination—providing smaller
loans—to MBE borrowers because they perceive the credit
Traits often associated with urban minority neighborhoods market offers them fewer borrowing options, compared with
include low, fluctuating household incomes, high crime rates, equally creditworthy Whites. In this environment of fewer
inferior public services, outdated infrastructure, and other choices, providing MBEs with relatively inferior loan terms
disamenities (Bates, 2010). Because minority neighborhood is consistent with banker perceptions that minority owner
firms needing credit assume bankers will penalize them for reservation prices differ from those of Whites. Stated differ-
these disadvantages, they are presumably hesitant to incur the ently, MBE borrowers are expected to be willing to accept
time costs and application fees required to prepare loan appli- smaller loans than equally creditworthy Whites, effectively
cations. Consistent with Bostic and Lampani’s findings lessening bank risk exposure without driving away custom-
(1999), Hypothesis 1 assumes that minority neighborhood ers. As Ayres and Siegelman (1995) observed, bankers are
location is an independent risk factor. using borrower’s race “to make statistical inferences about
the expected profitability of selling to them” (p. 304).
Hypothesis 2: Among equally creditworthy firms, busi- Bankers, seeking maximum profits, are presumably setting
nesses owned by minorities are more likely than White- loan terms for MBEs based on what the market will bear.
owned firms to be discouraged borrowers. We have not specified hypotheses regarding impacts of
inner-city firm location regarding either discouraged-bor-
As Myrdal (1944) observed, bankers may view minority rower incidence or borrower loan size because we have no
ownership as a proxy for moral hazard. Alternatively, they empirical or theoretical basis for doing so. Applicable theo-
may simply be averse to dealing with minorities. MBE own- ries and findings of past studies refer to neighborhoods in
ers seeking financing incur substantial search costs and most terms of residential racial composition, covered already
expect the process to be difficult. Discouraged borrowers are (Hypotheses 1 and 3). We could perhaps hypothesize that
unwilling to incur the costs of actually applying for loans— neighborhood poverty rates affect discouraged-borrower
which, by their own admission, they need—because they incidence and loan terms, but we did not because adequate
expect to be turned down. Because MBE owners needing theoretical underpinnings are lacking for such hypotheses.
credit often assume bankers will penalize them for being We have controlled for zip-code-level poverty rates in our
minorities, many are hesitant to incur the time costs and fees regression analyses.
required to complete loan applications.
Econometric Findings
Hypothesis 3: Among equally creditworthy firms receiv-
ing bank loans, businesses in minority neighborhoods are One trait distinguishes minority neighborhood KFS firms
likely to receive loans smaller in dollar amount than firms from others: over 48% are MBEs, whereas 14.5% of the
located elsewhere. urban firms located elsewhere are minority owned. Marion
(2009) has documented similar patterns of MBE geographic
If minority neighborhoods are indeed riskier locations for concentration. Analyzing construction firms, he found that
operating small businesses than other locations in urban MBEs “are disproportionately located in zip codes with the
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 243 of 801
166 Economic Development Quarterly 30(2)
Table 3. Logistic Regression—Delineating Firms Needing Credit but Not Applying for Bank Loans From Others.
Regression coefficient Standard error
Minority area 0.204 0.162
Black 1.271* 0.231
Latino 0.824* 0.258
Asian 0.295 0.247
Other minority −0.135 0.376
Minority area * minority owner −0.324 0.273
Industry experience −0.006 0.006
Owner hours worked 0.014* 0.002
College graduate −0.056 0.121
Wealth high −0.929* 0.152
Wealth medium −0.731* 0.139
# Employees 0.001 0.003
Net profit −0.092* 0.003
Credit score high −0.826* 0.152
Credit score medium −0.888* 0.137
Household poverty rate −0.035* 0.011
Constant 0.632 1.173
N 4,332
Log likelihood (significant level) −(0.000)
Efron’s R2 .111
Cragg–Uhler (Nagelkerke) R2 .170
Source. Kauffman Firm Survey database.
*Significant 5% level.
greatest concentration of own-race residents” (p. 441). This time. Interestingly, firms in neighborhoods characterized by
locational pattern highlights the issue of whether the numer- high household poverty rates are no more likely to be dis-
ous discouraged borrowers in minority communities reflect couraged borrowers than those operating in more affluent
the high incidence of MBEs, banker aversion to lending in areas (Table 3). Bank regulatory authorities often stress that
these neighborhoods, or both. To clarify this issue, we utilize CRA enforcement focuses on bank lending adequacy in low-
logistic regression analysis to delineate discouraged borrow- income areas (Immergluck, 2004).
ers from others, controlling for credit-risk factors. The Black- and Latino-owned firms, according to our regres-
dependent variable equals 1 for discouraged borrowers (oth- sion findings, recognize that bankers view minority owner-
erwise 0): Positive regression coefficients indicate a greater ship negatively. They are therefore less likely to submit loan
likelihood of being a discouraged borrower; results are applications, compared with equally creditworthy White
reported in Table 3. firms, because they expect to be penalized by bankers for
Regression analysis results indicate that minority neigh- being minority owners. We therefore accept Hypothesis 2.
borhood firm location is unrelated to discouraged-borrower Race, not space, emerges as a key factor explaining the high
likelihood, other factors being constant. We therefore reject discouraged-borrower incidence typifying urban minority
Hypothesis 1: discouraged borrowers apparently recognize communities.
that minority neighborhood location is not, by itself, penal- We estimated a variant of Table 3’s analysis to test whether
ized by banks. Instead, strong discouraged-borrower identi- inclusion of a variable identifying inner-city firms was useful
fiers include (a) owner wealth of $50,000 or less, (b) firm for delineating discouraged borrowers from others (results
ownership by African Americans or Latinos, (c) low profits, not reported). Previously included variables identifying
and (d) low credit scores (Table 3).6 The high discouraged- minority neighborhoods and neighborhood poverty rates
borrower incidence in minority neighborhoods is apparently were excluded from this exercise. The inner-city variable had
rooted in conventional borrower credit-risk measures and no explanatory power: Neither inner-city nor minority neigh-
owner race/ethnicity, not firm location per se. The low-net- borhood location, per se, appears to influence owner deci-
worth owner operating an unprofitable firm, in other words, sions about whether to apply for loans.
correctly recognizes that banker loan-application approval is Ordinary least squares regression analysis is next utilized
unlikely, particularly if the business has a history of credit to explain the loan sizes received by small firms receiving
problems. Filing a loan application is therefore a waste of bank loans. Two variants are reported: In Model 1, a
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 244 of 801
Bates and Robb 167
Table 4. Regression Model Explaining Loan Size Among Small Firms Receiving Bank Loans.
Model 1 Model 2
Regression coefficient Standard error Regression coefficient Standard error
Minority area 0.047 0.244 —
Black owner −0.618* 0.172 −0.676* 0.272
Latino owner −0.551 0.403 −0.370 0.375
Asian owner 0.200 0.270 0.294 0.321
Other minority owner 0.217 0.305 0.169 0.320
Industry experience −0.007 0.008 −0.006 0.008
Owner hours worked 0.003 0.004 0.002 0.004
College graduate 0.373* 0.169 0.392* 0.187
Wealth high 0.716* 0.244 0.696* 0.252
Wealth medium −0.266 0.255 −0.398 0.257
# Employees 0.022* 0.006 0.034* 0.007
Net profit 0.018 0.030 0.023 0.035
Credit score high 0.380 0.231 0.320 0.250
Credit score medium 0.343 0.241 0.324 0.269
Inner-city area — 0.343 0.442
Household poverty rate −0.002 0.012 —
Constant 11.33* 1.499 11.33* 1.422
N 656 549
R2 .278 .308
Source. Kauffman Firm Survey database.7
*Significant 5% level.
locational variable identifying minority neighborhood firm borrower variable is similarly large and negative (−0.551 in
location is included, while in Model 2, an inner-city identi- Model 1), but the corresponding standard error (0.403)
fier is employed (Table 4). The issues of interest are (a) negates statistical significance and suggests widely varying
whether minority neighborhood or inner-city location or (b) banker treatment of Latino-owned firms in determining bank
minority ownership are statistically significant determinants loan amounts. Although these findings are provocative, they
of loan size, after controlling for credit-risk factors. We are are insufficient to justify acceptance of Hypothesis 4 in the
testing Hypothesis 3 that minority neighborhood (inner city) case of Latino-owned borrowers.8
firm location shapes loan amounts and Hypothesis 4 that
owner race/ethnicity influences loan amounts.
Conclusion
Consistent regression findings emerge in the sense that
firm location, per se, has no impact on loan size once credit- The high discouraged-borrower incidence documented in
risk factors and owner race/ethnicity are controlled for (Table our study confirms the consensus view that business owners
4). Hypothesis 3 is therefore rejected. Firm and owner traits needing credit often do not seek bank financing. Discouraged-
related to larger borrower loan size include owner wealth of borrower firms are particularly numerous in minority neigh-
$250,000 or more, college graduate owners, and, to a small borhoods, where the percentage of small businesses needing
degree, being a large firm, measured by employee numbers. credit, yet not applying, far exceeds the incidence of firms
Owner race/ethnicity, additionally, stands out as important, actually applying for bank loans. Numerous discouraged
but less so than in our previous analysis of discouraged bor- borrowers typifying minority neighborhoods are a reflection
rowers; we therefore accept Hypothesis 4 with reservations. of their bleak credit-risk profiles and the high incidence of
Determinants of the loan amounts banks provide to small- Black- and Latino-owned small businesses, not firm loca-
business borrowers include owner race in one instance: tion, per se. The confirmed absence of a direct causal link
Black firm ownership, other factors being equal, is associ- between being located in a minority neighborhood and the
ated with smaller loan amounts. Racial differences in firm likelihood of being a discouraged-borrower firm advances
owner reservation prices are consistent with this pattern of our understanding of why credit is scarce for small busi-
Black-owned firms receiving smaller loans than other bor- nesses in inner city, predominantly minority areas.
rowers. Importantly, Table 4’s regression exercise controls In light of these widespread unmet credit needs, it is
for firm industry. The coefficient attached to the Latino important to understand why inner-city firms are so often
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 245 of 801
168 Economic Development Quarterly 30(2)
discouraged borrowers. The mere fact of being owned by an additional independent factor limiting MBE access to bank
African American or Latino individual explains why many loans. By identifying owner race/ethnicity as an important
suffer from limited loan access. Firms benefitting from fac- causal factor, a remedial policy is obvious. Banker treatment
tors like the abundant, low-cost local labor supply may thrive of potential MBE clients at the applicant screening stage
in the same inner-city environs where others suffer, and these stands out when one considers policy options useful for
locational advantages often lead to solidly profitable busi- increasing business loan access, both in minority neighbor-
nesses with high credit scores and owners with rising per- hoods and throughout urban America. Relatively harsh
sonal wealth. Our findings suggest that banks readily assist banker screening of potential MBE loan applicants (Bone
these firms, particularly if their owners are White. et al., 2014) discourages some from completing the applica-
Other small firms, meanwhile, cope with very real opera- tion process. Regarding marketing their lending services, we
tional disadvantages, often rooted in inner-city neighborhood have no direct measures of the differing degrees to which
disamenities. Excessive competition among small firms banks target MBEs and White-owned firms. Both the audit
catering to clients in overcrowded product niches is a related study findings reported by Bone et al. (2014) and the higher
problem (Bates & Robb, 2014). These in turn affect firm risk-adjusted incidence of discouraged MBE borrowers doc-
credit histories and profitability negatively, eventually reduc- umented in this study suggest that bankers more proactively
ing owner wealth and producing high credit-risk profiles, seek White-owned businesses as clients than MBEs.
which in turn minimize business prospects for obtaining Our finding that banks are providing smaller loans to
needed financing. Bankers, our findings suggest, focus heav- Black-owned small businesses (in comparison to equally
ily on these conventional credit-risk measures (and minority creditworthy White ventures) is speculative. Whereas racial
ownership), not firm location, when deciding which firms to differences in borrower reservation prices provide a logical
cultivate as clients. Client screening and loan-application explanation for these smaller loans, other explanations are
evaluation practices assigning greater risk to all minority possible. Our present policy implication is that bank provi-
neighborhood firms in an across-the-board fashion harm sion of smaller loans to Black-owned firms is discriminatory,
stronger and weaker area businesses alike, but we have no but this assumes supply-side constraints limit loan size, not
clear evidence that banks pursue such strategies (Bates & demand-side considerations. Although Black-owned firms
Robb, 2015). We found no evidence that bankers are discrim- collectively may possibly seek smaller loans than equally
inating against local firms solely because of locational fac- creditworthy Whites, we doubt that such demand-side con-
tors. Focusing narrowly on minority neighborhood/inner-city siderations determine their smaller loan amounts.
location while ignoring conventional credit-risk indicators Undercapitalization of Black-owned firms, importantly, is
describing those same firms is not useful if one’s objective is widely recognized as a major determinant of their lower
to delineate small businesses having greater access to bank profits and higher closure rates, in comparison to White-
loans from those with less access. owned businesses (see, e.g., Bates, 1997; Fairlie & Robb,
Although our findings may appear to be inconsistent with 2008). The lower levels of financial capitalization typifying
those of previous studies documenting bank loan scarcity in Black-owned firms are “the single most important fact
minority neighborhoods, in fact they are not. Immergluck explaining racial differences in business outcomes” (Fairlie
(2002), for example, documented the relatively low level of & Robb, 2008, p. 130). The Black-owned business borrow-
loan availability typifying Black neighborhoods in ers receiving the larger loans are more profitable and less
Philadelphia, but he did not investigate impacts of individual likely to go out of business than others, controlling for firm
firm characteristics or owner traits on loan availability and owner traits. Stated differently, seeking loans equal in
because the CRA data he analyzed lacked these variables. dollar amount to those sought by equally creditworthy
Our contrasting findings, based on KFS data more com- White-owned firms is a sound business strategy. A possible
pletely describing business/owner risk profiles, clarify the demand-side constraint hypothesis worthy of investigation is
findings of Immergluck (2002), Smith (2003), and others. that owners fearing rejection because of who they are possi-
We are not suggesting Immergluck’s findings of low Black- bly request smaller loans to enhance their odds of having
neighborhood loan frequency would disappear had he based bankers approve their loan applications.
his analysis on better data. We found a high incidence of dis- We believe, finally, that racial differences in owner reser-
couraged MBE borrowers in minority neighborhoods, one vation prices are an important cause of the differing loan
that persisted after controlling for risk factors. This high terms received by MBEs and White business borrowers.
discouraged-borrower incidence, heavily rooted in wide- Applicable loan terms include not only loan dollar amount
spread MBE presence, is a likely cause of the low bank loan but loan maturity and interest rate as well. Our operating
frequency documented by Immergluck (2002). hypothesis is that the restricted borrowing choices faced by
Regarding widespread unmet small-business credit needs, minority business owners produce observed loan-term dif-
low firm profitability, credit scores, and owner wealth levels ferentials. Debt providers recognizing racial differences in
are important causal factors, but owner race stands out as an these reservation prices are able to provide less favorable
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 246 of 801
Bates and Robb 169
loan terms, on average, to MBE owners than to White loan Notes
recipients. To test this hypothesis, an empirical analysis of 1. High-level bank officials often appear to encourage nondis-
factors explaining the higher interest rates paid by MBEs, in criminatory lending practices, but this sentiment may or may
comparison with White borrowers, needs to be conducted. not filter down to the branch bank level. For this reason, the
Although Blacks may conceivably prefer smaller loans or type of audit study described by Bone et al. (2014) is sometimes
loans with shorter maturities than equally creditworthy commissioned by banks seeking to monitor the loan evaluation
White owners, they certainly do not want to pay higher inter- policies pursued by their individual branches (Lubin, 2008).
est rates. As MBEs become a growing presence among bank 2. Asian-owned firms are routinely identified as less likely than
clients, understanding their higher average borrowing costs Whites to have their loan applications approved, but owner race
emerges inconsistently as a significant determinant of applica-
and smaller loan amounts may eventually become a greater
tion rejection, perhaps because of sample-size constraints.
focus than our traditional concerns with understanding racial 3. Owners were asked about loans each year and asked to include
differentials in unmet credit needs and loan approval rates. new as well as renewal applications.
A useful present-day strategy for increasing small busi- 4. The KFS sampling frame is based on Dun & Bradstreet data.
ness loan availability in inner-city minority neighborhoods The KFS excludes branches/subsidiaries owned by existing
involves having bank regulators actually enforce the Equal businesses. A “business,” by definition, was created by one or
Opportunity Credit Act (EOAC) prohibiting banks from dis- more people and includes purchasing an existing business or a
criminating against minority applicants. The Federal Reserve franchise. Firms generating trivial annual sales were excluded.
Board is responsible for EOAC enforcement; an effective 5. Because certain firm subgroups were oversampled, we
strategy would entail conducting the type of audit study weighted the KFS data to compensate.
described in Bone et al. (2014). Effective EOAC enforce- 6. Alternative specifications of Table 3’s regression model gen-
erate results consistent with the findings discussed here; see
ment would ideally motivate bankers to subject potential
Bates and Robb (2015). Definitions of explanatory variables
MBE applicants to the same degree of scrutiny as White cli- used in regression analyses, often self-explanatory, require
ents seeking loans. A complementary strategy would be to some clarification. Net profit is measured in hundreds of thou-
assist business owners in a color-blind manner. sands (e.g., $20,000 = 0.2).
Demand-side strategies possibly useful for increasing 7. Year dummy variables, industry, owner age, and owner gender
MBE access to bank loans include encouraging discouraged also controlled for in Tables 3 and 4 regressions.
minority borrowers to go ahead and actually apply for bank 8. Another interesting variant of our loan-size analysis entails
loans. Organizations offering assistance to minority entre- delineating White Latino owners from non-Whites. Roughly
preneurs should actively encourage creditworthy discour- 30% of Latino firm owners self-identify as White, under 4%
aged borrowers to complete the application process. Equal identify themselves as Black, and the rest self-identify as other
treatment of equally creditworthy firms, irrespective of race/ minorities. White Latinos may be treated differently than other
Latino owners when bankers set loan terms. We are unaware
ethnicity, at all stages of the loan application and approval
of any studies testing this possibility.
process would help alleviate the higher barriers constraining
minority entrepreneurs seeking financing, thus allowing
References
more to enter into and operate successful small businesses,
whether those firms are located in minority communities or Aghion, P., & Bolton, P. (1997). A theory of trickle-down growth
elsewhere in America. and development. Review of Economic Studies, 64, 151-172.
Ayres, I., & Siegelman, P. (1995). Race and gender discrimination
in bargaining for a new car. American Economic Review, 85,
Authors’ Note 304-321.
All errors are the authors’ responsibility. Certain data included Bates, T. (1973). An econometric analysis of lending to Black-owned
herein are derived from the Kauffman Firm Survey release 4.0. Any businesses. Review of Economics and Statistics, 55, 272-283.
opinions, findings, and conclusions or recommendations expressed Bates, T. (1997). Race, self employment, and upward mobility: An
in this material are those of the authors and do not necessarily illusive American dream. Baltimore: Johns Hopkins University
reflect the views of the Ewing Marion Kauffman Foundation. Press.
Bates, T. (2010). Alleviating the financial capital barriers impeding
Declaration of Conflicting Interests business development in inner cities. Journal of the American
Planning Association, 76, 349-362.
The author(s) declared no potential conflicts of interest with respect
Bates, T. (2011). Minority entrepreneurship. Foundations and
to the research, authorship, and/or publication of this article.
Trends in Entrepreneurship, 7, 151-311.
Bates, T., & Robb, A. (2013). Greater access to capital is
Funding needed to unleash the local economic development poten-
The author(s) disclosed receipt of the following financial support tial of minority-owned businesses. Economic Development
for the research, authorship, and/or publication of this article: Quarterly, 27, 250-259.
Research reported in this study was generously supported by the Bates, T., & Robb, A. (2014). Small-business viability in America’s
Ewing Marion Kauffman Foundation. urban minority communities. Urban Studies, 51, 2844-2862.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 247 of 801
170 Economic Development Quarterly 30(2)
Bates, T., & Robb, A. (2015). Has the Community Reinvestment Jackson, K. (1985). Crabgrass frontier: The suburbanization of
Act increased loan availability among small businesses operat- America. New York, NY: Oxford University Press.
ing in minority neighborhoods? Urban Studies, 52, 1702-1721. Lubin, P. (2008). Fair lending testing: Best practices, trends, and
Becker, G. (1993). Nobel lecture: The economic way of looking at training. Cambridge, MA: Joint Center for Housing Studies,
behavior. Journal of Political Economy, 101, 385-389. Harvard University.
Blanchflower, D. (2009). Minority self-employment in the United Marion, J. (2009). Firm racial segregation and affirmative action in
States and the impact of affirmative action programs. Annals of the highway construction industry. Small Business Economics,
Finance, 5, 361-396. 14, 441-453.
Blanchflower, D., Levine, P., & Zimmerman, D. (2003). Mitchell, K., & Pearce, D. (2011). Lending technologies, lending
Discrimination in the small business credit markets. Review of specialization, and minority access to small-business loans.
Economics and Statistics, 85, 930-943. Small Business Economics, 37, 277-304.
Bone, S., Christensen, G., & Williams, J. (2014). Rejected, shack- Myrdal, G. (1944). An American dilemma: The Negro problem and
led, and alone: The impact of systematic restricted consumer modern democracy. New York, NY: Harper & Row.
choice among minority customers’ construction of self. Journal Parker, S. (2009). The economics of entrepreneurship. Cambridge,
of Consumer Research, 41, 451-474. England: Cambridge University Press.
Bostic, R., & Lampani, P. (1999). Racial differences in patterns of Robb, A., Ballou, J., DesRoches, D., Potter, F., & Zhao, Z. (2009).
small business finance. In J. Blanton, A. Williams & S. Rhine Overview of the Kauffman firm survey. Kansas City, MO:
(Eds.), Business access to capital and credit (pp. 253-276). Ewing Marion Kauffman Foundation.
Washington, DC: Federal Reserve System. Salop, S., & Stiglitz, J. (1977). Bargains and rip-offs: A model
Cagnetti, M., & De Nardi, M. (2006). Entrepreneurship, frictions, of monopolistically competitive price dispersion. Review of
and wealth. Journal of Political Economy, 114, 835-870. Economic Studies, 44, 493-510.
Casey, C. (2009). Linking social capital and indirect policy tools: Smith, G. (2003). Small business lending in the Chicago region,
Fostering equitable community investment responses. Journal 2001. Reinvestment Alert, 23, 1-17.
of Planning Education and Research, 28, 413-425. Stegman, M., Cochran, K., & Farris, R. (2002). Toward a more
Casey, C., Glasberg, D., & Beeman, A. (2011). Racial disparities performance-driven service test: Strengthening basic banking
in access to mortgage credit: Does governance matter? Social services under the Community Reinvestment Act. Georgetown
Science Quarterly, 92, 783-806. Journal of Poverty Law & Policy, 9, 405-452.
Cavalluzzo, K., & Wolken, J. (2005). Small business loan turn- Stiglitz, J., & Weiss, A. (1981). Credit rationing in markets with
downs, personal wealth and discrimination. Journal of imperfect information. American Economic Review, 71,
Business, 78, 2153-2178. 393-410.
Fairlie, R., & Robb, A. (2008). Race and entrepreneurial success:
Black-Asian- and White-owned businesses in the United States. Author Biographies
Cambridge: MIT Press.
Timothy Bates is a distinguished professor at Wayne State
Immergluck, D. (2002). Redlining redux: Black neighborhoods,
University. Previously, he was chair of the graduate program of urban
Black-owned firms, and the regulatory cold shoulder. Urban
policy analysis at the New School for Social Research. He is cur-
Affairs Review, 38, 22-44.
rently writing a book, “Black Detroit and the Promise of America,”
Immergluck, D. (2004). Credit in the community. Armonk, NY: M.
coauthored by Beth Bates.
E. Sharpe.
Initiative for a Competitive Inner City. (2005). State of the Alicia Robb is a senior fellow with the Ewing Marion Kauffman
inner city economies: Small businesses in the inner city Foundation and the founder of Next Wave Ventures, which is
(Research Contract SBAHQ03-M-056). Washington, DC: focused on driving diversity in angel investing. She is also a visiting
U.S. Small Business Administration. scholar with the University of California, Berkeley.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 248 of 801
Exhibit H
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 249 of 801
Journal of Economic Perspectives—Volume 28, Number 3—Summer 2014—Pages 25–48
Entrepreneurship as Experimentation†
William R. Kerr, Ramana Nanda, and Matthew
Rhodes-Kropf
A
lthough the fundamental importance of entrepreneurship has been gener-
ally recognized by economists since at least Schumpeter (1911 [1934]; 1943),
it has not been the subject of substantial research and is more-or-less absent
from standard textbooks. Over the last decade, however, entrepreneurship has flour-
ished as a research topic within economics. As entrepreneurship emerges from the
shadows, many central and unresolved questions linger: How should we define entre-
preneurship? What are its key aspects? Is entrepreneurship about the next Skype or
the next self-employed accountant? We assert in this paper that effective entrepreneur-
ship—especially among high-growth ventures and for the economy as a whole—builds
upon a process of experimentation in deep and nuanced ways.
Entrepreneurship is fundamentally about experimentation because the knowl-
edge required to be successful cannot be known in advance or deduced from some
set of first principles. As Hayek (1948) put it, “the solution of the economic problem
of society is . . . always a voyage of exploration into the unknown.” For entrepre-
neurs, it can be virtually impossible to know whether a particular technology or
product or business model will be successful, until one has actually invested in it.
Two interesting examples help frame the discussion. In 1999, the venture
capital firms Sequoia Capital and Kleiner Perkins Caufield & Byers each invested
$12.5 million in Google, a brand new startup that claimed to have a superior search
engine. By the time Sequoia sold its stake in 2005, that investment was worth over
■ William Kerr, Ramana Nanda, and Matthew Rhodes-Kropf are Associate Professors at
Harvard Business School, Harvard University, Boston, Massachusetts, and Faculty Research
Fellows at the National Bureau of Economic Research, Cambridge, Massachusetts. Their
email addresses are wkerr@hbs.edu, RNanda@hbs.edu, and mrhodeskropf@hbs.edu.
†
To access the data Appendix, visit
http://dx.doi.org/10.1257/jep.28.3.25 doi=10.1257/jep.28.3.25
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 250 of 801
26 Journal of Economic Perspectives
$4 billion, returning 320 times the initial cost (Sahlman 2010). The staggering
success should not obscure a key point—Google was by no means a sure-shot invest-
ment in 1999. The search algorithm space was already crowded and dominated by
other players such as Yahoo! and Altavista. Google, founded by two students, might
well have turned out to just be a “me too” investment. In fact, other venture capi-
talists had turned down the opportunity to invest in Google at the time. One such
firm was Bessemer Ventures. Partner David Cowan, on being asked by a friend to
meet with the two Google founders who had rented space in her garage, is believed
to have quipped, “How can I get out of this house without going anywhere near
your garage?” (Bessemer Venture Partners, no date). Over the years, Bessemer has
been so successful that they playfully mock at their “anti-portfolio” of deals that they
passed on (at http://www.bvp.com/portfolio/antiportfolio).
Not many investors are as candid about great opportunities that they have
passed on, but more recently, Fred Wilson of Union Square Ventures has simi-
larly written about his regret at passing on Airbnb, a startup that lets people rent
rooms or homes to prospective guests online. Started in 2008, Airbnb currently lists
500,000 properties in more than 34,000 cities across 192 countries, far more than
any hotel chain—thereby making it the largest lodging company and brand in the
world (Levere 2013). Wilson (2011) explains Union Square Venture’s decision to
pass on the deal in 2009: “we couldn’t wrap our heads around air mattresses on the
living room floors as the next hotel room. . . . Others saw the amazing team that we
saw, funded them, and the rest is history.”
These two examples highlight several challenges associated with commercial-
ization of new ideas, products, and technologies. First, the actual distribution of
returns in such ventures has a low median value but very high variance (Scherer
and Harhoff 2000; Hall and Woodward 2010). Most new ventures fail badly, but
some turn out to be wildly successful. Second, even for professional investors or
managers making resource allocation decisions, it is impossible to know in advance
which ideas will work. As we describe in greater detail below, venture capital inves-
tors make their returns on the one investment out of many that turns out to be a
wild success like Google or Airbnb. The vast majority of venture capital investments,
however, return less than the face value of the investment.
How entrepreneurs and investors respond to these inherent challenges has
important implications for their own success and also for the broader economy in
terms of the “best ideas” being commercialized. To flesh out this point, it is important
to separate two frames of reference regarding experimentation. The first relates to
economic experimentation in a Darwinian sense, which is the natural starting point
for most economists. In this conceptual model, new ventures compete with existing
products and technologies, and the ensuing competition leads to the survival of
the fittest, just as Google surpassed its early rivals due to its superior technology.
This competition can be described as experimentation at the level of the economy.
In settings where the best approach among several options is unknown, a great
benefit of market-based economies is that winners are often chosen by consumers
and competition. As Rosenberg (1994) has argued, one of the defining features
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 251 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 27
Table 1
US Venture Capital Statistics for 2010
Number of active venture capital firms 462
Amount invested by venture capital firms $22 billion
Number of companies receiving venture capital investment 2,749
Number of companies receiving funding for first time 1,001
Share of dollars invested in Information Technology firms 51%
Share of dollars invested in Life Science firms 27%
Source: Data from National Venture Capital Association, nvac.org.
of capitalism is the freedom it provides entrepreneurs to pursue novel approaches
to value creation in the pursuit of economic gain. The promise of large rewards
drives entrepreneurs to experiment with new ideas, helping to create a dynamic
and growing economy. An institutional environment that facilitates experimenta-
tion is thus central to maintaining a vibrant entrepreneurial ecosystem (Dosi and
Nelson 2010). This experimentation, combined with a willingness to let losing
incumbents fail, is the underlying notion behind Schumpeter’s (1943) process of
“creative destruction.”
While this Darwinian depiction is important and a natural starting point, a
second reference point emphasizes that we should be cautious of assuming that
market-based mechanisms can always serve as a guiding hand with respect to experi-
mentation. As in the Google example, it is often very difficult to know whether a
particular technology or venture will succeed until one has made an investment.
Moreover, the investments to obtain this vital information can be nontrivial: in
Google’s case, a total of $25 million was invested in the first round. Table 1 docu-
ments some basic facts about the venture capital industry, including the number
of active investors. Given the relatively small number of financiers that invest in
this sector, it is easy to imagine scenarios where Google or other highly productive
investments fail to receive the required funding.
In an entrepreneurial setting, where the benefit of pursuing different
approaches is not clear and the costs of tests are expensive, each individual endeavor
is also engaged in a process of experimentation. As these experiments provide
information about the likelihood of ultimate success, entrepreneurs and investors
gain information about whether to continue the project. However, the investment
and continuation decisions for entrepreneurs are often not made in a competitive
Darwinian contest, because the decisions to invest further or shut down a firm are
often made by only a few investors, well before startups can compete in the product
market or have positive cash flow. Moreover, the decisions are made by discrete indi-
viduals, often in venture capital firms or other early-stage financing vehicles, whose
actions are impacted by a myriad of incentive, agency, and coordination problems.
Thus, the extent to which the best idea goes forward may depend on factors such
as the organizational structure or incentive system of the firm where the investor
is based, available information sets (for example, access to certain networks),
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 252 of 801
28 Journal of Economic Perspectives
coordination costs, and other such frictions. Taken together, these factors affect
how much experimentation is undertaken in the economy and also the trajectory of
experimentation, with potentially very deep economic consequences.
The lens of experimentation has implications for what types of innovations will
occur, who will pursue them and when. As Stern (2005) argued, “a favorable envi-
ronment for entrepreneurship and a high level of economic experimentation go
hand in hand.” Although experiments can be conducted in large companies or in
the public sector, new technologies and innovative products are often commercial-
ized by entrepreneurs and often cluster at particular times. We argue that the costs
and constraints on the ability to experiment alter the type of organizational form
surrounding innovation and influence when innovation is more likely to occur.
This article considers these costs and constraints to experimentation by inves-
tors and the implications that these issues have on the type of entrepreneurship
across time and across economies. We start with a micro-level examination of experi-
mentation and focus on the processes used by investors to experiment and the issues
that individual firms face. We then turn to the macro perspective of the economy
and the policies that shape entrepreneurship and its effectiveness.
A Closer Look at the Process of Experimentation
High-impact entrepreneurship requires, almost by definition, going against
the grain. Rajan (2012) argues that an entrepreneur “must be willing to strike
out, largely on the basis of intuition, on courses of action in direct opposition to
the established settled patterns.” A consequence of this setting is extreme uncer-
tainty about whether a particular technology, product, or business model will be
successful. This uncertainty is fundamentally different from risk, as Knight (1921)
stressed. With risk, such as placing money on a spin of a roulette wheel, one can
define exact probabilities and expected values. With Knightian uncertainty, these
probabilities are not known, and even the form of the potential outcomes may be
unclear. For example, looking forward from today, at what rate will electric cars, if
at all, replace traditional automobiles and how will the supporting infrastructure
for battery recharging be designed? What will be the impact of nascent augmented
reality technologies for how humans interact? Which, if any, of the several current
ideas to cure cancer will be successfully commercialized?
In this environment of tremendous uncertainty, experimentation allows entre-
preneurs and investors to assess and commercialize projects without investing the
full amount. Crucially, experimentation offers more than just a possibility of higher
returns—it also allows entrepreneurs and investors to pursue projects that are not
feasible in an all-or-nothing bet. For example, consider a project that requires $110
to commercialize and will be worth $0 with 99 percent probability or $10,000 with
1 percent probability. This project will not be pursued, because its expected value is
negative (−$10). But imagine we can conduct an experiment that will reveal if the
project has a 10 percent chance of working. (Suppose further that the probability
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 253 of 801
Entrepreneurship as Experimentation 29
of the experiment having a positive outcome is only 10 percent, so that in this
example the 1 percent chance of overall success is unchanged.) If the experiment
gives a positive signal, the project has a larger expected value of $890. Thus, as long
as the experiment costs less than $89 (10 percent × $890), the experiment should
be conducted, with the project then being either shut down or commercialized
based upon the results.
This simple example provides insights into many features of the entrepreneurial
landscape, as we elaborate below. The test resolves uncertainty about the project’s
potential and creates a real option value for further pursuit. Thus, experimentation
is particularly valuable in cases where initial information can be especially informa-
tive about the overall quality of the project and this information is cost effective
to obtain.1
The example also illustrates that the costs of experimenting are important.
Our simple example was phrased in terms of the direct costs of conducting the
test. More broadly, important indirect costs must be borne by either the investor
or entrepreneur when they fail. Even if direct costs are small, significant indirect
costs like a stigma of failure (Landier 2005) may still prevent entrepreneurs from
pursuing otherwise valuable tests if they only have a 10 percent chance to succeed.
Moreover, the ability to document and transmit the results of experiments is impor-
tant, because the information from the experiment is only valuable if it can be acted
upon. Can an investor use the information to decide to provide the follow-on invest-
ment required to realize the project? The framework also begins to highlight the
distinct roles of the entrepreneur and the investor.
We trace these features out next by first looking more closely at the economics
of a venture capital firm. This setting provides a powerful archetype of high-impact
entrepreneurship more broadly and the role of experimentation. We then discuss
frictions to experimentation that are observed in the economy and the extent to
which agents are able to address them.
The Economics of Experimentation in a Venture Capital Firm
The economics and structure of a venture capital firm provide insights into
entrepreneurship as experimentation (Sahlman 1990). A crucial starting point is
the recognition that venture capital firms do not simply act as a portfolio of risky
startups. While the portfolio nature of a venture capital firm is important, it is
quite distinct from what is observed in stock market mutual funds, for example,
where the portfolio is designed to smooth out the idiosyncrasies of individual
stocks. Venture capital firms are better thought of as conducting a portfolio of
tests across a number of highly uncertain ideas with skewed economics (similar
to the example described above). In most cases, the tests come out negative. In
fact, the majority of venture-capital-backed firms fail, despite the investments
made into them by highly skilled investors. Once venture capital firms identify the
1
Weitzman (1979) introduces and analyzes the optimal search behaviors under these uncertainties.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 254 of 801
30 Journal of Economic Perspectives
cases that deliver positive test results, however, they pursue them aggressively and
invest ever larger amounts on the promising candidates to scale them up. Whereas
mutual fund investors do not need to be an early investor in a company in order to
be allowed to invest more later, this is precisely the way that venture capital firms
aim to make money—by starting small and owning a larger share of the firms that
turn out to be successful, while attempting to cut their losses on the unsuccessful
ones as early as possible.2
Some simple calculations document these skewed returns using data on
all venture capital investments in the United States. Using data from Thompson
Venture Economics, we identify all startup ventures that received their first round
of early-stage financing for the years between 1985 and 2009. We calculate both
the total investment made by venture capital investors in each of these firms, as
well as the gross return at the time of exit—either through an acquisition deal, an
initial public offering, or bankruptcy. We then group each of these startups based
on their economic return to investors at the time of exit—that is, the ratio of the
gross return at the point of exit divided by the total amount invested in the startup.
About 55 percent of startups that received venture capital over this period were
terminated at a loss, and only 6 percent of them returned more than five times their
investment. This 6 percent group, however, was extremely successful and together
accounted for about 50 percent of the gross return generated over the period.3 The
fact that the returns are so skewed across the portfolio is prima facie evidence that
the (quite knowledgeable) investors cannot distinguish in advance the next Google
from the other cases.
A second example provides more systematic evidence on the difficulty of
predicting outcomes, even conditional on a venture capital firm making an invest-
ment. For a single large and successful venture capital firm that has invested more
than $1 billion over the last decade, we have access to both the outcomes of indi-
vidual investments and the scores that partners assigned to each venture at the time
of their first investment. We place the investments of this firm into buckets according
2
Gompers and Lerner (2004) and Da Rin, Hellman, and Puri (2013) provide a comprehensive over-
view of these intermediaries. Chemmanur, Krishnan, and Nandy (2011) and Puri and Zarutskie (2012)
provide recent evidence on the performance of venture-capital-backed startups. Parker (2004) provides
a broader review of the literature on entrepreneurship.
3
Total gross returns need to be estimated due to the fact that a venture’s value at exit is missing for a
large number of acquisitions, for firms that went bankrupt, and for the “living dead”—firms that are
coded as still private but have not received follow-on financing for several years. Our research suggests
that most of the acquisitions without data are fire sales and that firms that are coded as “alive” but have
not received a follow-on round of financing within three years of the last financing are likely to be bank-
rupt. In the calculation reported above, we have conservatively assumed that unreported acquisitions
were undertaken at 1.5 times cost and that firms that did not receive subsequent rounds of financing
for three years, but were coded as still private, were liquidated at 0.25 times cost. The results are robust
to assuming a wide range of possible values for these outcomes, including an (implausible) average
of two times the dollars invested in the firm. These numbers correspond closely with numbers from
Sahlman (2010) who uses data from eleven early-stage venture capital firms to show that 64 percent of
their investments were terminated at a loss, while 8 percent of the investments, those that returned over
five times the investment, generated about 60 percent of the gross return of the portfolio.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 255 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 31
Figure 1
Total Cost and Total Return for a Venture Capital Firm
70%
Share of total cost in bucket
Share of gross return from bucket
60%
50%
40%
30%
20%
10%
0%
≤1 >1 and ≤3 >3 and ≤5 >5 and ≤10 >10
Gross Return/Total Investment
Source: Authors using proprietary data from an anonymous venture capital firm that has invested more
than $1 billion over the last decade.
Notes: Investments are placed into buckets according to their return multiple. For each bucket, we show
the total cost and total return for that group of investments.
to their return multiples, and in Figure 1, show the total cost and total return for
each bucket of investments. The distribution of outcomes for this firm follows the
same pattern as the aggregate data on venture capital (with about 60 percent of
investments terminated at a loss and 10 percent generating a return more than
five times the capital invested).
The firm has an excellent track record in the industry, which implies a well-
functioning algorithm for choosing and managing investments. Conditional on
making an investment, however, Figures 2A and 2B show the limited ability of
the venture capital investors to tell the most successful investments from the less
successful ones—even for these experienced investors. Figure 2A highlights that the
distribution of scores assigned to investments that ended up performing extremely
well is statistically no different from the distribution of scores assigned to investments
that ultimately failed or had mediocre returns.4 (The labels on the horizontal axis
4
The Kolmogorov–Smirnov test for equality of distribution functions could not reject the hypothesis that
these were the same.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 256 of 801
32 Journal of Economic Perspectives
Figure 2
Scores Assigned to Investments at Time of First Investment and the Ultimate
Returns of Those Investments, for One Venture Capital Firm
A: Distribution of Scores by Outcome
.6 Ultimate return ≤ 1X
Ultimate return 1X to 3X
Ultimate return > 3X
.4
Density
.2
0
Average score from partners at time of first investment
B: Correlation between Scores and Outcomes
5
4
Log (1+ Return multiple)
3
2
1
0
Average score from partners at time of first investment
Note: The labels on the horizontal axis have been suppressed to maintain the confidentiality of the
investor’s rating scale, but lower predictions were on the left and higher predictions were on the right.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 257 of 801
Entrepreneurship as Experimentation 33
have been suppressed to maintain the confidentiality of the investor’s rating scale,
but lower predictions were on the left and higher predictions were on the right.)
Figure 2B is a scatterplot showing, for each of the firm’s investments, the average
score assigned at the time of first investment on the x-axis versus a function of the
return multiple on the y-axis. The raw correlation between scores and the return
multiple from the investments is 0.1.
This example shows that even conditional on making an investment, it is hard
for the investors to predict which firm will be successful. Using a similar framework
from a sample of early-stage “angel investors” in Kerr, Lerner, and Schoar (2014),
the correlation among the interest levels assigned to funded deals and their ulti-
mate success is less than 0.1. Again, these examples help illustrate the high degree
of uncertainty present in these investments.
Thus, venture capital firms need to experiment, and many aspects of their busi-
ness model facilitate or emphasize this experimentation. One example is their focus
on sectors that are capital-efficient for both experimentation and subsequent scaling
and that can generate large returns for the successful investments in a short period
of time. Under these conditions, most notably associated with information tech-
nology investments, venture capital firms can run initial experiments of manageable
financial sizes and then fund the winners to completion. The corollary to this, which
is further elaborated upon below, is that venture capital activity is concentrated in a
narrow range of technological opportunities. Some sectors, like renewable energy
production, need to be proven at large scale to demonstrate technical feasibility
and unit economics. Commercializing such ventures requires building large manu-
facturing plants and hence is significantly more capital intensive, and takes much
longer. Following a brief period where venture capitalists invested heavily in biofuel
and solar technologies only to learn these lessons the hard way, they have largely
shied away from funding renewable energy production startups, instead devoting
their attention within clean energy to startups commercializing energy efficiency,
smart grid, and other software technologies (Nanda, Younge, and Fleming, forth-
coming). Fleming (2001) and Fleming and Sorenson (2004) discuss the aggregate
implications of a narrower range of search or recombination.
In addition to focusing on specific sectors, venture capital structures and
contractual choices also address experimental challenges. For example, in a way that
is similar to our earlier numerical example, venture capital investors provide staged
financing to startup companies that tie each financial infusion to milestones—points
at which information is revealed about the quality of the project. This structured
financing builds real options (for example, Gompers 1995; Bergemann and Hege
2005; Bergemann, Hege, and Peng 2008), by matching the amount of money raised
in each round to the specific uncertainty that needs to be resolved with that round
of funding: for example, proof that the technology works, that consumers will buy
the product, and so on. The most successful investors and entrepreneurs are able
to identify the most important uncertainties facing a new idea and experiment in a
way that resolves the greatest proportion of the uncertainty around them effectively
and quickly.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 258 of 801
34 Journal of Economic Perspectives
The consequence of this staging approach is that an important role of venture
capital firms is to shut down ventures for which they are receiving bad signals.
Investors exercise the abandonment options in several ways. Across financing
rounds, the investors can choose to not reinvest in the company; within each
financing round, venture capital firms also structure contracts to provide them
with control and cash flow rights. Control rights allow venture capital firms to
replace underperforming chief executive officers, and the investors can also push
the company to reposition itself if the chosen path turns sour. These rights appear
to be exercised often (for example, Hellman and Puri 2000; Kaplan, Sensoy, and
Strömberg 2009; Tian 2011).5
The available evidence suggests that this process of venture capital experimen-
tation does lead to successful economic outcomes. Kortum and Lerner (2000) and
Samila and Sorenson (2011) document spillover effects from venture capital activity
within industries and cities, respectively. Using Census Bureau data, we provide
descriptive evidence of how this sort of experimentation is associated with outcomes
for ventures. We use name and address matching to identify new entrants that
received venture capital investments during the 1986–1997 period. Looking ahead
to 2007, 75 percent of these venture-capital-backed ventures have shut down, but the
remaining firms have grown to a level where they combine to equal 364 percent of
the total employment of the original investments (including those that failed). For
comparison, we generated a second sample of non-venture-capital-backed entrants
that have the same four-digit industry, year, and starting employment level as the
investment sample. In 2007, a greater share of this second group remains alive—only
66 percent have shut down. However, those that are alive only account for two-thirds
(67 percent) of the total employment of their original sample. These matched
ventures have thus grown, but not nearly to the extent of the venture-capital-backed
group. While our simple matching procedure may leave residual differences across
the groups and also does not account for either the role of selection or the effect of
venture capital on startups’ outcomes, it should not obscure the broad overall point:
the venture-capital-backed ventures experienced a higher if similar rate of failure
(75 versus 66 percent) but also experienced far greater employment gains (364
versus 67 percent of employment in their samples). High-growth ventures typically
constitute a small proportion of new startups but are responsible for a dispropor-
tionate share of employment and GDP. As a rough benchmark, about 1,000 of the
over 500,000 firms founded each year in the United States obtain venture capital
finance, but these firms account for about 40 percent of new US publicly-listed
companies over the last three decades (Ritter 2014).
5
Many other aspects of the venture capital model reflect these cash-flow and control rights (for example,
Kaplan and Strömberg 2003, 2004). One prominent example is convertible preferred stock that protects
venture capital firms in bad outcomes by allowing them to redeem the full value of their investment,
thereby treating the investment as debt to the extent that the failed venture has any residual value. On
the other hand, in positive outcomes, the venture capital firm can choose to convert the preferred to
common stock, allowing them to share in the upside when things go well.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 259 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 35
Frictions for the Process of Experimentation—Costs of Experiments
The costs of running experiments play a big role in entrepreneurship. Techno-
logical change in the last decade has dramatically lowered these costs, particularly
in industries that have benefited from the emergence of the Internet, including
trends like open-source software and cloud computing. Industry observers suggest
that firms in these sectors that would have cost $5 million to set up a decade ago
can be done for under $50,000 today. For example, open-source software lowers
the costs associated with hiring programmers. In addition, fixed investments in
high-quality infrastructure, servers, and other hardware are no longer necessary at
the birth of many software firms because they can be rented in tiny increments from
cloud computing providers and efficiently scaled up as demand for their products
increases (Blacharski 2013; Palmer 2012). This reduced entry barrier has led to an
explosion of experimentation with new entrepreneurial ideas in this area. Funding
sources have also proliferated as the ability to finance these smaller and cheaper
experiments is possible for individuals and groups beyond the traditional venture
capital investors with large pools of capital.
One prominent example of this shift is the Lean Startup methodology advo-
cated by serial entrepreneur Steve Blank and popularized by Eric Ries’s (2011)
book, “The Lean Startup.” Many startup founders have shaped their efforts using
the Lean Startup approach, and the ideas have reached the leadership of large
companies like Jeff Immelt, the chief executive officer of General Electric. The core
of the management style is a focus on identifying and developing “minimal viable
products” (MVPs) that reveal how well the overall opportunity will fare. An example
of an MVP is the release of a consumer website that has only 10 percent of the
functionality that the founders ultimately envision for the product. However, by
quickly building a workable version with only the bare essentials, the MVP approach
seeks to validate as many assumptions as possible about the viability of the final
product before expending enormous effort and financial resources. From the infor-
mation collected during these experiments, the venture adjusts its course in a way
that may involve pivoting from the original agenda. The followers of this method-
ology frequently discuss how to make their experiments ever more cost-effective,
in large part so that they do not need to raise as much money to pursue a range of
possible ideas.
This approach to building companies has coincided with the rapid rise of angel
investors and crowd-funding platforms,6 particularly for consumer Internet startups.
6
Angel investors are defined as individuals investing their own capital in startup ventures. Given
the decentralized nature of such investments, it is extremely difficult to accurately quantify the size
of the market. The Angel Capital Association (2012) estimates that its membership consists of about
8,000 accredited investors (who are individuals allowed by the Securities and Exchange Commission to
make investments in risky, private startups) who invest in approximately 800 companies per year. The
size of the broader angel market has been estimated to be as large as $20 billion dollars, but the vast
majority of this is tiny investments made by individuals in firms that are already cash-flow positive (Shane
2009). The total does, however, provide a sense of how large the crowd-funding market could be given
recent changes in regulation through the Jumpstart Our Business Startups ( JOBS) Act of 2012, which
allow a larger number of individuals to invest legally in private companies (Nanda and Kind 2013).
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 260 of 801
36 Journal of Economic Perspectives
For example, AngelList, a platform dedicated to matching startups with angel inves-
tors experienced rapid growth since its founding in 2007, and in June 2013 had
approximately 100,000 startups and 18,000 accredited investors on its platform, of
which 2,000 had been funded (Nanda and Kind 2013). The lower costs associated
with initial experiments implies that those with pools of capital smaller than typical
venture capitalists can more easily become involved in these financing decisions.
Because positive information from experiments leads the valuation of startups in
subsequent rounds to increase substantially, this leads to less dilution for early-stage
financiers and makes it even more attractive for investors with smaller pools of capital
to get involved with financing decisions. Equally important, however, is the fact that
as these experiments get better at quickly telling apart projects with high and low
potential, they allow startups that would probably not have received funding in the
past due to their uncertain outcomes to now receive financing. Ewens, Nanda, and
Rhodes-Kropf (2014) show that these are particularly likely to be startups that had
a very small chance of being extremely successful, such as those founded by young,
inexperienced, first-time entrepreneurs. While these individuals would not have
received capital in the past, the financing environment has changed in recent years,
creating room for cohort-based accelerator programs with an educational compo-
nent, such as Y-Combinator, that aim to provide mentoring, networks, and capital to
such founders (Graham 2008). Fehder and Hochberg (2014) document that there
were no such programs in the United States in 2004, but by 2013 over 40 such accel-
erators were functioning across the United States.
While a trend toward lower start-up costs is particularly true in software and
information technology businesses, where the cost of experimentation has declined
dramatically and the frequency with which one learns new information about the
product is very high, the same forces exist in very capital-intensive industries as well.
An example is Terrapower, a startup trying to commercialize a new way of producing
nuclear energy (Sahlman, Nanda, Lassiter, and McQuade 2012). Bill Gates, an early
backer of Terrapower, noted that rapid advances in computing power provide this
startup with significant advances toward experimentation. In the past, Terrapower
would have had to construct an entire nuclear power plant to test whether its new
technology would work in practice, costing several billion dollars and taking years
to complete, making it impossible to finance in the first place. Now, with the intro-
duction of supercomputers, Terrapower’s engineers can simulate the inside of a
nuclear reactor, learn whether its technology will work, and make rapid (and much
cheaper) iterations to gain more confidence about the potential of the technology
before ever constructing the physical nuclear power plant. This in turn makes it
viable to finance an initial exploration for an early-stage investor.
These examples reinforce two points about the cost of experimentation. First,
differences across industries in the ability and cost to learn about the final outcome
have a huge bearing on the degree of experimentation that we see in early stages—
the very long time frames and costs for learning about potential technologies in
renewable energy or different approaches to curing cancer create a dearth of exper-
imentation, despite intense societal interest (Fernandez, Stein, and Lo 2012). Even
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 261 of 801
Entrepreneurship as Experimentation 37
for venture capital firms, the levels of investment required for clean energy efforts
or certain healthcare investments tend to be too high per project for them to bear.
Second, the pace of technological progress has a feedback effect on new inno-
vations. As technological advances such as supercomputers or cloud computing
diffuse, they affect the cost of experimentation in completely different sectors
like nuclear energy and consumer Internet. The advances mean that projects that
had a negative expected value in the past become viable. Referring to our earlier
numerical example, they may drop the cost of the hypothetical test from several
hundred dollars to less than $89. This spillover effect helps sustain and potentially
even increase high rates of technological progress.
Frictions for the Process of Experimentation—Organizational
A key part of experimentation is the ability to terminate projects, but organiza-
tions differ in their ability and willingness to terminate underperforming ventures.
The venture capital industry has long used the expression “throwing good money
after bad money” to warn against continual investment in startups for whom the
initial experiments reveal poor information. Venture capital firms that shut down
more ventures do not necessarily perform worse, and, in fact, some of the best
venture capital firms have among the highest rates of abandoning projects—in part
because their skill at designing experiments and acting upon the results allows them
to pursue more aggressive strategies and enter into more uncertain domains.7
This capacity to terminate projects, while certainly not exclusive to the venture
capital community, is often difficult to replicate within large corporations or
through equity markets. Large corporations, for example, often find it difficult
to terminate experiments that aren’t working out, due in part to career concerns
of the managers in charge of the effort, the “soft budget constraints” that arise
when a large corporation can keep providing funding for a time, and similar traits.
The initial experiment may also provide ambiguous or slightly negative informa-
tion that can be interpreted in different ways depending upon the biases of parties.
In several ways, the venture capital system is designed to counteract these tenden-
cies, beginning with an industry structure where venture capital firms are separate
from the startup itself but still capable of intensive monitoring. Another mechanism
often used by venture capital firms is to bring in new investors with each financing
round to ensure that the deal is evaluated by outsiders. In academia, the academic
tenure process that includes external review after a specific time horizon generates
comparable features.
Of course, because venture capital investors consider each investment as
contributing to overall portfolio returns, their incentives are not always aligned with
entrepreneurs in terms of strategic decisions such as when to shut down or exit
7
While the higher rate of abandoning projects does not jump out from the statistics we provided above, it
is worth noting that venture capital investors also likely select better projects, so that a simple comparison
of realized failure rates does not provide a complete picture of how much more they abandon projects
compared to non-venture-capital-backed firms.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 262 of 801
38 Journal of Economic Perspectives
an investment. For example, Gompers (1996) has documented the incentives of
venture capital firms to engage in grandstanding (taking companies public earlier
than may be optimal) to develop a reputation that enables them to raise follow-on
funds. Venture capital investors may also choose to shut down an underperforming
firm if it is unlikely to generate high gross returns, or they may continue to reinvest
if they want to build a reputation as entrepreneur-friendly. These decisions may not
always be optimal for the firm.
These examples highlight a related tension with organizational design,
however. Innovation requires running experiments that will often fail, and such
failures will occur even if the idea looked promising at the start and all parties
acted in good faith. These features may require an organizational tolerance for
failure in the short-term, with compensation rewards over the long-run based upon
success (Manso 2011). This tolerance allows opportunities for early experimenta-
tion to identify the best course of action, while the long-term compensation tied to
performance combats the “moral hazard” concern that decision makers might take
excessive risks, and it solicits their best efforts.
How organizations balance the termination of poor experiments with main-
taining a tolerance for failure in turn affects the overall set of projects that they
choose to start. Nanda and Rhodes-Kropf (2012) show that the inability or the
unwillingness to terminate projects when intermediate information is negative
leads organizations to start projects that are less experimental in the first place.
In other words, organizations can pick a radical range of projects at the start and
terminate a high percentage of them, or pick a more limited range of projects at the
start and terminate fewer of them. Large bureaucratic organizations often struggle
with terminating projects, and these differences provide a rationale for why large
companies are comparatively better at pursuing incremental improvements than
radical innovations, although this is of course not true in every case. March (1991)
and Akcigit and Kerr (2010) provide greater discussion on firm sizes and explora-
tion versus exploitation choices. Lerner (2012) provides a discussion of the relative
benefits of the traditional model of research and development within firms versus
the model of venture capital investors, and Thomke (2003) discusses how large
companies can take insights from the model of experimentation to improve the way
in which they develop and advance innovations.
Frictions for the Process of Experimentation—Continuation and Financing Risk
Successful experimentation requires being able to capitalize on experiments
that reveal positive outcomes, and these upside scenarios can be as tricky as termi-
nation decisions. Financing projects in stages requires that startups return to the
capital markets at regular intervals for more capital. Thus far, we have emphasized
positive aspects of multiple financing rounds (for example, the advantages of having
an outside investor not previously involved with the company value the venture).
A negative aspect is that entrepreneurs and early-stage investors feel vulnerable to
the state of the financial markets at each round of financing. At times, financial
capital is freely available, while in other instances it may be hard to come by, even
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 263 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 39
for otherwise sound projects. These financing constraints can be marketwide or
they can be specific to sectors due to the degree to which investors herd around
certain hot sectors.
Venture capital investors routinely refer to “financing risk” (similar to rollover risk)
to describe how otherwise sound projects may not obtain capital for the next experi-
ment. One way to protect against financing risk is to go less frequently to the capital
markets, by taking larger chunks of money at each stage. This, however, reduces the
value of the abandonment options for venture capital firms. Nanda and Rhodes-Kropf
(2010, 2013) argue that hot markets—times when financing risk is low—allow proj-
ects with the highest real option values to be funded, because the continuation risk
is lower for all projects in the economy. Empirically, venture-capital-backed firms that
are first funded during hot markets have the highest failure rates, but conditional
on being successful, startups funded in more-active periods were valued higher at
initial public stock offering or acquisition compared to startups funded in less-active
periods exiting at the same time. Financing risk is most salient for projects that need
to go back to the capital markets many times, and thus projects that are worst hit
by this risk are experimental new technologies that tend to have the highest option
value. In fact, a possible implication of this work, which is related to work on the role
of stock markets in financing innovation (Brown, Fazzari, and Petersen 2009), is that
the most innovative projects in the economy may need times of low financing risk
(“hot financial markets”) to drive their initial commercialization.
Financing risk is representative of a broader class of continuation features
that govern the degree to which experimentation can be pursued. To illustrate,
consider some of the differences between clean energy and biotech; biotech also
requires substantial investment and very long horizons but is the beneficiary of
substantially more venture capital investment. Biotech startups and the pharma-
ceutical industry are part of a vibrant “market for ideas” (Gans, Hsu, and Stern
2002) that allows biotech companies to sell the results of their experiments at
milestones. That is, a biotech company can undertake a first experiment to show
that a project that originally seemed to have had a 0.1 percent chance of becoming
a blockbuster drug has in fact a 5 percent chance. This experiment has generated
substantial information value, and pharmaceutical companies are willing to buy
the remaining 1-in-20 opportunity for their drug development portfolios. This
market works because the science used in the first experiment is observable and
verifiable, a patent system protects the startup from intellectual theft, and the
long-term opportunity is more protected compared to clean energy innovations
(which can face competition from all alternative sources of energy, including
the opening up of strategic oil reserves). Such a “market for ideas” is much less
developed for clean energy, and as a consequence entrepreneurs and investors
who are considering an experiment in this area face additional continuation risks
(Nanda, Younge, and Fleming, forthcoming). Relatedly, competition to acquire
new technologies among pharmaceutical firms combined with a relatively robust
market for initial public offerings have meant that the upside from biotechnology
innovation, if successful, is sufficiently high to warrant the financing of upstream
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 264 of 801
40 Journal of Economic Perspectives
experimentation. We return to the role of the markets for initial public offerings
in greater detail in the next section.
Discrete Choices and Individuals—A Process of Experimentation
We close this section by emphasizing its most prominent feature—the role of
humans, often just a few humans, in making these decisions. Economic models
often focus on the fact that in well-functioning economies, good ideas are rewarded
and flourish relative to worse ideas. But they are silent on how these ideas emerge.
A central theme of this section is that entrepreneurs and investors must make
discrete choices, often with quite incomplete information and uncertainty about
the future, about which ideas to progress and which to shut down. Financiers rather
than markets dictate which projects are realized, as they choose which experiments
to attempt, how to interpret the results, and whether to continue or abandon the
investment. The entrepreneurial landscape at the micro-level is thus characterized
by individuals choosing the fate of a venture based on what they have learned about
what is likely to work before it competes in the product market.
This section has emphasized how constraints on the ability of investors to
experiment efficiently can shape which industries, organizations, and time periods
see the most radical innovations. It also sets the framework for understanding where
barriers to experimentation may lead to market failures. For example, intermediate
experiments may reveal very little about the ultimate success of some projects (such
as building a particle collider or the nuclear power startup before the advent of
simulation), requiring extremely large investments before one can learn about
their viability. Alternatively, the horizon for commercialization may be extremely
uncertain and distant, such as in the case of research in basic science. Institutional
regimes such as academia (and at times the government) may thus be critical to
enable experimentation in areas that are of importance to society but where a
process of serial experimentation by profit-seeking investors is unlikely to provide
a set of stepping-stones to the technologies behind disruptive innovation.
Economic Experiments at the Society Level
While we have focused our attention on the difficulty in predicting outcomes
for new technologies, many of these challenges also exist for firms in the broader
economy. For example, Shane (2009) documents that of the approximately
500,000 startups that were founded in the United States in 1996, over half had failed
by 2002 and only about 3,500 achieved sales of greater than $10 million. Haltiwanger,
Jarmin, and Miranda (2013) note the difficulty in predicting upfront which firms
will succeed even among producers of homogeneous goods such as cement. The
laws and institutions at the regional and national level can therefore play a critical
role in driving productivity growth—often through the incentives they create or the
costs that they impose on the ability to effectively experiment. For example, Glaeser,
Kerr, and Kerr (forthcoming) trace the effects of entrepreneurship on long-run
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 265 of 801
Entrepreneurship as Experimentation 41
urban growth by documenting how cities with industrial and mining legacies in
1900 found it harder to create the up-or-out dynamic associated with productivity
growth. They argue that the institutions (industrial structure, culture, and skill
base) created by the predominantly large mining firms in the 1900s hindered entre-
preneurship in the modern era, leading to systematically lower urban growth in
these regions post-1970.
Although some institutions are extremely persistent and hard to change, we
outline some types of policies that seem to be particularly effective in promoting
“economic experiments” in the form of entrepreneurship.
Democratizing Entry and Facilitating Efficient Failure
Bank finance and debt financing more generally is an extremely important
source of capital for young companies. The US Small Business Administration has
estimated that there were 21 million small business commercial and industrial loans
outstanding in 2010 (defined as loans being below $1 million) that were valued at
$310 billion (Office of Advocacy, US SBA 2013). While many small businesses are
not young, Robb and Robinson (2014) document that debt finance is important
even among young firms. They look at the sample of startups in the Kauffman Firm
Survey and find that within the first three years of founding, 40 percent of the
funding source for these startups is constituted by outside debt, over and above
the 4 percent of debt that consists of the owner’s credit cards and personal loans.
Changes in the availability and terms of bank finance can therefore have
important implications for entrepreneurship. For example, Kerr and Nanda (2009)
consider how the banking competition fostered by the state-level US banking
deregulations that occurred from the 1970s through the 1990s affected entrepre-
neurship and the economy. These deregulations facilitated greater competition
by allowing out-of-state banks to enter local markets, thereby increasing access to
credit and lowering the cost of external finance. Using micro-level data from the
US Census Bureau, Kerr and Nanda (2009) find evidence for the standard story of
creative destruction—a few strong entrants challenging and later replacing incum-
bent firms. However, the mechanism through which this was achieved was through
widespread entry that largely resulted in failure, so that the most pronounced impact
was a massive increase in churning among new entrants.8 Similarly, Chava, Oettle,
Subramanian, and Subramanian (2013) examine innovation outcomes following
the banking deregulations and find that banking deregulation facilitated greater
risk taking and experimentation by small firms. Because it is hard to know before-
hand which projects are going to be successful, “democratizing entry” seems to be
an important trait of well-functioning capital markets.
Efficient experimentation implies that institutional environments facilitating
exit are as important as those that facilitate entry. Two areas that have received
8
Prior work by Black and Strahan (2002) and Cetorelli and Strahan (2006) connected the deregulations
to higher entry rates, but it was quite puzzling that the elasticity for entrepreneurship was about ten times
higher than the response observed on any other dimension like productivity change.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 266 of 801
42 Journal of Economic Perspectives
significant attention in recent years are the role of bankruptcy law and employment
protection laws and tradeoffs involved as they relate to experimentation.
On the one hand, bankruptcy laws that favor creditors allow them to recoup as
much capital as they can from the startup—encouraging them to lend in the first
place (Berkowitz and White 2004; Cumming forthcoming; Cerqueiro, Hegde, Penas,
and Seamans 2013). Such laws also discourage the moral hazard that might otherwise
arise among entrepreneurial agents if they faced little need to repay creditors. On
the other hand, entrepreneurial failure allows society to test uncertain projects. This
process requires separating the entrepreneur from the firm, as is accomplished to
some extent by the limited liability provision that entrepreneurs are not personally
liable for the debts of a firm, and thus provides entrepreneurs with the ability to termi-
nate projects and move on. As an example, Eberhart, Eesley, and Eisenhardt (2013)
find that a bankruptcy reform in Japan that reduced the consequences of closing
a firm encouraged greater levels of entrepreneurship and risk taking following the
reform. The best policy appears to involve striking a balance between a good measure
of limited liability for entrepreneurs, allowing them to transition across projects
without severe and lasting penalties, along with some protections or restrictions for
small or unsophisticated investors who may not comprehend fully the low likelihood
of the entrepreneur’s success. While legal factors play a role in reducing the down-
side from failure, cultural factors such as a stigma of failure can also play a role in
hindering entry. In some regions, such as Silicon Valley, past failure can even be seen
as a badge of honor, making it much easier for individuals to take big risks in terms
of startup ventures.
A related set of issues can arise from strict employment protection laws that
limit the ability of firms to adjust their workforce rapidly and in that way act as an
effective tax on employment adjustments (Autor, Kerr, and Kugler 2007). Many
ultimately successful startups have had to undertake one or more critical changes
in their business model, which in turn require substantial adjustments in their
workforce. One of the tradeoffs of legal mandates for employment is that they
make it much more difficult for high-growth startups to experiment and for inno-
vative, volatile sectors to form (for example, Saint-Paul 2002; Samaniego 2006).
Indeed, venture capital investment is lower in countries with stringent employ-
ment protection laws, and the more-volatile sectors are the most affected (Bozkaya
and Kerr forthcoming). In this sense, the flexibility of labor markets governs the
types of projects that entrepreneurs can undertake (Fallick, Fleischman, and
Rebitzer 2006).
Appropriating Value in Successful States of the World
Beyond the policies that shape an entrepreneur’s ability to experiment when
things are going poorly, policies or conditions that limit the value that entrepre-
neurs or investors receive in good states of the world are also important. Some
conditions like the extent of patent protection were noted earlier in our compar-
ison of biotech and clean energy investments. More generally, the investors and
state-contingent financial contracts that we describe require strong property rights
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 267 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 43
and a sound rule of law (for example, La Porta, Lopez-de-Silanes, Shleifer, and
Vishny 1997). Similarly, public equity markets are critical to rewarding startups
that are successful, by allowing investors and entrepreneurs to cash out on the
expectation of future growth. When equity markets do not work well or there are
limited opportunities to exit and retrieve at least a portion of the earlier invest-
ment, it becomes much harder to experiment (for example, Black and Gilson
1998; Michelacci and Suarez 2004).
Policy Implications
How should policymakers interested in promoting entrepreneurship think
about the role of experimentation? Clearly, difficult entry regulations suppress the
experimentation undertaken by startups (Klapper, Laeven, and Rajan 2006) and
often are difficult to justify on any grounds. As one example, Fairlee, Kapur,
and Gates (2011) describe the extent to which the prevalence of employer-provided
health insurance in the US economy serves as an implicit entry barrier for potential
founders of new firms, making it more difficult for them to leave current jobs.
In addition, taking the experimental perspective seriously suggests that it is a
poor idea for government to seek to pick and promote individual firms. After all,
even the most-experienced venture capital firms have substantial success in only
one of every ten investments they pick, so we shouldn’t expect inexperienced and
possibly not-very-objective politicians to do better. Indeed, one of the features that
make market-based economies better at commercializing radical innovations is the
decentralized and parallel nature with which new ideas are tested (Rosenberg 1994;
Qian and Xu 1998; Scherer and Harhoff 2000). Politicians also have greater diffi-
culty terminating projects—that is, telling taxpayers that legislative decisions have
spent their money with little or no return. In our experience, most economists buy
into this wariness of policymakers acting as a venture capital firm, and many go a
step further and caution against picking an individual sector or industry to support.
Virtually every state has a biotech initiative, but few will be successful. Lerner (2009)
emphasizes the dismal performance of policies attempting to seed specific ventures
or industries.
Governments are more likely to facilitate effective entrepreneurship if they
work to reduce the costs of experimentation in general. What we have in mind here
is not so much government programs that seek to target start-ups with particular
benefits, but instead a careful consideration of the broader regulatory framework,
including labor laws and requirements with which new entrants need to comply,
with a focus on how they affect incentives for entry. These efforts to structure a
better playing field are admittedly less glamorous than announcing a new biotech
cluster initiative, but they are far more likely to have sustained effects.
As alluded to in the previous section, the experimentation view also suggests
that there may be systematic market failures when the costs associated with experi-
mentation are too high or the returns are too uncertain and far into the future. This
creates a framework for thinking about which sectors, such as say basic science, may
warrant sustained government support.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 268 of 801
44 Journal of Economic Perspectives
Finally, government should be cautious about industrial policies that seek to
minimize business failures, as such policies may only be propping up firms that need
to fail. For example, Acemoglu, Akcigit, Bloom, and Kerr (2013) build a model of
firm-level innovation, productivity growth, and reallocation featuring endogenous
entry and exit, and calibrate it using firm-level micro data from the Census. Their
results highlight a key role for new entrants in policies to promote innovation and
growth. Acemoglu et al. (2013) find that while policies like research and develop-
ment tax credits to entrants can help and may encourage growth, their impact pales
in comparison to removing artificial support for inefficient incumbents.
Conclusions
Picking up a quotation originally made by Michael Jordan to describe missing
basketball shots, a prominent venture capitalist, Vinod Khosla, emphasizes, “our
willingness to fail gives us the ability and opportunity to succeed where others may
fear to tread” (Khosla Ventures, no date; Khosla 2013). This paper has outlined our
perspective on entrepreneurship that emphasizes this fundamental role for experi-
mentation. Viewing entrepreneurship as experimentation allows individuals and
societies to evaluate businesses and technologies in domains with greater uncer-
tainty than otherwise possible, unlocking deep growth opportunities.
Economists should bear in mind two very different forms of experimentation
that are associated with entrepreneurship. One form is the economic experimenta-
tion that takes place in market-based economies when several new ideas, products
and technologies are continually tested and either displace existing technologies
or more likely fail themselves. A second less-appreciated form of experimentation
happens at the micro-level, before these ideas compete, and relates to the process
of bringing new ideas to the market. There are too many opportunities to pursue
with scarce resources, and the best-designed experiments can come back with
misleading or ambiguous results. Under these conditions, specific choices made
by discrete individuals—especially founders of firms and early-stage investors—
become very important. Recent research in financial economics, organizational
design, and related fields highlight how the costs associated with the ability of inves-
tors to experiment can alter the nature of entrepreneurship and help to explain
why entrepreneurship is more prevalent in certain industries, regions, or periods
of time.
Less than a decade ago, research on entrepreneurship within economics
had little internal cohesion; instead, researchers working on entrepreneurship
were spread out within broader fields like finance, labor economics, or macroeco-
nomics. Today, while most researchers in this area continue to keep one foot firmly
planted in a traditional field, the internal cohesion for entrepreneurial research has
formed and is rapidly obtaining scale. Researchers on entrepreneurship are them-
selves experimenting with new ideas and new directions, and we expect this area of
research to develop rapidly in the coming years.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 269 of 801
Entrepreneurship as Experimentation 45
■ We appreciate comments and editorial direction by David Autor, Chang-Tai Hsieh, and
Timothy Taylor. The research in this paper was conducted while the authors were Special
Sworn Status researchers of the US Census Bureau at the Boston Census Research Data Center
(BRDC). Support for this research from the NSF grant ITR-0427889 [BRDC], the Division
of Research and Faculty Development at Harvard Business School, and from the Kaufman
Foundation is gratefully acknowledged. Research results and conclusions expressed are the
authors and do not necessarily reflect the views of the Census Bureau or the NSF. This paper
has used proprietary data from a venture capital firm that requests it remain anonymous, and
it has screened the paper to ensure that no confidential data are revealed.
References
Acemoglu, Daron, Ufuk Akcigit, Nicholas Has Changed How Startups Launch.” Techie.com,
Bloom, and William R. Kerr. 2013. “Innovation, November, 26. http://techie.com/top-ten-ways-the
Reallocation and Growth.” NBER Working Paper -cloud-has-changed-how-startups-launch/.
18993. Black, Bernard S., and Ronald J. Gilson. 1998.
Akcigit, Ufuk, and William Kerr. 2010. “Growth “Venture Capital and the Structure of Capital
through Heterogeneous Innovations.” NBER Markets: Banks versus Stock Markets.” Journal of
Working Paper 16443. Financial Economics 47(3): 243–77.
Angel Capital Association. 2012. “ACA, Angel Black, Sandra E., and Philip E. Strahan. 2002.
Groups and Angel-Backed Companies.” Slides, “Entrepreneurship and Bank Credit Availability.”
September. http://www.angelcapitalassociation Journal of Finance 57(6): 2807–33.
.org/data/Documents/Resources/ACAandAngel Bozkaya, Ant, and William Kerr.Forthcoming.
GroupBackground09-12.pdf. “Labor Regulations and European Venture
Autor, David H., William R. Kerr, and Adriana Capital.” Journal of Economics and Management
D. Kugler. 2007. “Does Employment Protection Strategy.
Reduce Productivity? Evidence from US States.” Brown, James R., Steven M. Fazzari, and
Economic Journal 117(521): 189–217. Bruce C. Petersen. 2009. “Financing Innovation
Bergemann, Dirk, and Ulrich Hege. 2005. “The and Growth: Cash Flow, External Equity and
Financing of Innovation: Learning and Stopping.” the 1990s R&D Boom.” Journal of Finance 64(1):
RAND Journal of Economics 36(4): 719–52. 151–85.
Bergemann, Dirk, Ulrich Hege, and Liang Cerqueiro, Geraldo, Deepak Hegde, María
Peng. 2008. “Venture Capital and Sequential Fabiana Penas, and Robert Seamans. 2013.
Investments.” Working paper. “Debtor Rights, Credit Supply, and Innovation.”
Berkowitz, Jeremy, and Michelle J. White. 2004. Unpublished paper.
“Bankruptcy and Small Firms’ Access to Credit.” Cetorelli, Nicola, and Philip Strahan. 2006.
Rand Journal of Economics 35(1): 69–84. “Finance as a Barrier to Entry: Bank Competition
Bessemer Venture Partners.No date. “Anti- and Industry Structure in Local U.S. Markets.”
portfolio.” http://www.bvp.com/portfolio/anti Journal of Finance 61(1): 437–61.
portfolio. Chava, Sudheer, Alexander Oettl, Ajay
Blacharski, Dan. 2013. “Top Ten Ways the Cloud Subramanian, and Krishnamurthy Subramanian.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 270 of 801
46 Journal of Economic Perspectives
2013. “Banking Deregulation and Innovation.” and Urban Growth: An Empirical Assessment
Journal of Financial Economics 109(3): 759–74. with Historical Mines.” Review of Economics and
Chemmanur, Thomas J., Karthik Krishnan, Statistics.
and Debarshi K. Nandy. 2011. “How Does Venture Gompers, Paul A. 1995. “Optimal Investment,
Capital Financing Improve Efficiency in Private Monitoring, and the Staging of Venture Capital.”
Firms? A Look beneath the Surface.” Review of Journal of Finance 50(5): 1461–89.
Financial Studies 24(12): 4037–90. Gompers, Paul A. 1996. “Grandstanding in
Cumming, Douglas.Forthcoming. “Measuring the Venture Capital Industry.” Journal of Financial
the Effect of Bankruptcy Laws on Entrepreneur- Economics 42(1): 133–56.
ship across Countries.” Journal of Entrepreneurial Gompers, Paul, and Josh Lerner. 2004. The
Finance. Venture Capital Cycle. Cambridge, MA: MIT Press.
Da Rin, Marco, Thomas Hellmann, and Manju Graham, Paul. 2008. “A New Venture
Puri. 2013. “A Survey of Venture Capital Research.” Animal.” March, http://paulgraham.com
Chap. 8 in Handbook of the Economics of Finance, /ycombinator.html.
Vol. 2, Part A, edited by George Constantinides, Grenadier, Steven R., and Andrey Malenko.
Milton Harris, and René M. Stulz. Amsterdam: 2011. “Real Options Signaling Games with Applica-
North Holland. tions to Corporate Finance.” Review of Financial
Dosi, Giovanni, and Richard R. Nelson. 2010. Studies 24(12): 3993–4036.
“Technical Change and Industrial Dynamics as Hall, Robert E., and Susan E. Woodward.
Evolutionary Processes.” In Handbook of the Economics 2010. “The Burden of the Nondiversifiable Risk
of Innovation, Vol. 1 edited by B. H. Hall and of Entrepreneurship.” American Economic Review
N. Rosenberg, 51–127. Amsterdam: North Holland. 100(3): 1163–94.
Eberhart, Robert, Charles E. Eesley, and Haltiwanger, John, Ron S. Jarmin, and Javier
Kathleen M. Eisenhardt. 2013. “Failure is an Miranda. 2013. “Who Creates Jobs? Small versus
Option: Failure Barriers and New Firm Perfor- Large versus Young.” Review of Economics and
mance.” Stanford University Working Paper 111. Statistics 95(2): 347–61.
Ewens, Michael, Ramana Nanda, and Matthew Hayek, Friedrich A. 1948. Individualism and
Rhodes-Kropf. 2014. “Entrepreneurship and the Economic Order. University of Chicago Press.
Cost of Experimentation.” Unpublished paper. Hellmann, Thomas, and Manju Puri. 2000.
Fairlee, Robert W., Kanika Kapur, and Susan “The Interaction between Product Market and
Gates. 2011. “Is Employer-based Health Insurance Financing Strategy: The Role of Venture Capital.”
a Barrier to Entrepreneurship?” Journal of Health Review of Financial Studies 13(4): 959–84.
Economics 30(1): 146–62. Kaplan, Steven N., Berk A. Sensoy, and Per
Fallick, Bruce, Charles A. Fleischman, and Strömberg. 2009. “Should Investors Bet on the
James Rebitzer. 2006. “Job-hopping in Silicon Jockey or the Horse? Evidence from the Evolu-
Valley: Some Evidence Concerning the Microfoun- tion of Firms from Early Business Plans to Public
dations of a High-Technology Cluster.” Review of Companies.” Journal of Finance 64(1): 75–115.
Economics and Statistics 88(3): 472–81. Kaplan, Steven N., and Per Strömberg. 2003.
Fehder, Daniel C., and Yael V. Hochberg. 2014. “Financial Contracting Theory Meets the Real
“Accelerators and the Regional Supply of Venture World: An Empirical Analysis of Venture Capital.”
Capital Investment.” Unpublished paper. Review of Economic Studies 70(2): 281–315.
Fernandez, Jose-Maria, Roger M. Stein, and Kaplan, Steven N., and Per Strömberg. 2004.
Andrew W. Lo. 2012. “Commercializing Biomed- “Characteristics, Contracts, and Actions: Evidence
ical Research through Securitization Techniques.” from Venture Capitalist Analyses.” Journal of
Nature Biotechnology 30(10): 964–75. Finance 59(5): 2177–2210.
Fleming, Lee. 2001. “Recombinant Uncertainty Kerr, William R., Josh Lerner, and Antoinette
in Technological Search.” Management Science Schoar. 2014. “The Consequences of Entrepre-
47(1): 117–32. neurial Finance: Evidence from Angel Financings.”
Fleming, Lee, and Olav Sorenson. 2004. Review of Financial Studies 27(1): 20–55.
“Science as a Map in Technological Search.” Stra- Kerr, William R., and Ramana Nanda. 2009.
tegic Management Journal 25(5–9): 909–28. “Democratizing Entry: Banking Deregulations,
Gans, Joshua, David H. Hsu, and Scott Stern. Financing Constraints, and Entrepreneurship.”
2002. “When Does Start-up Innovation Spur the Journal of Financial Economics 94(1): 124–49.
Gale of Creative Destruction?” RAND Journal of Klapper, Leora F., Luc Laeven, and Raghuram
Economics 33(4): 571–86. G. Rajan. 2006. “Entry Regulation as a Barrier to
Glaeser, Edward L., Sari Pekkala Kerr, and Entrepreneurship.” Journal of Financial Economics
William R. Kerr.Forthcoming. “Entrepreneurship 82(3): 591–625.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 271 of 801
William R. Kerr, Ramana Nanda, and Matthew Rhodes-Kropf 47
Knight, Frank H. 1921. Risk, Uncertainty, and Office of Advocacy, US Small Business Asso-
Profit. Boston, MA: Houghton Mifflin. ciation. 2013. Small Business Lending in the United
Kortum, Samuel S., and Josh Lerner. 2000. States 2012. http://www.sba.gov/sites/default
“Assessing the Contribution of Venture Capital /files/files/sbl_12study.pdf.
to Innovation.” RAND Journal of Economics 31(4): Palmer, Maija. 2012. “Cloud Computing Cuts
674–92. Start-up Costs.” Financial Times, February 29.
Khosla Ventures.No date. Company website Parker, Simon C. 2004. “The Economics of Self-
mainpage, http://www.khoslaventures.com/ Employment and Entrepreneurship.” Cambridge
(accessed 5/14/14). University Press.
Khosla, Vinod. 2013. “Fail as Often as You Want, Puri, Manju, and Rebeca Zarutskie. 2012. “On
And Fail Well.” Video of a speech posted March 10, the Lifecycle Dynamics of Venture-Capital- and
2013, at Wunderman Reports.com. http://www Non-Venture-Capital-Financed Firms.” Journal of
.wundermanreports.com/fail-as-often-as-you-want Finance 67(6): 2247–93.
-and-fail-well. Qian, Yingyi, and Chenggang Xu. 1998. “Inno-
Landier, Augustin. 2005. “Entrepreneurship vation and Bureaucracy under Soft and Hard
and the Stigma of Failure.” Unpublished paper. Budget Constraints.” Review of Economic Studies
La Porta, Rafael, Florencio Lopez-de-Silanes, 65(1): 151–64.
Andrei Shleifer, and Robert Vishny. 1997. “Legal Rajan, Raghuram G. 2012. “Presidential
Determinants of External Finance.” Journal of Address: The Corporation in Finance.” Journal of
Finance 52(3): 1131–50. Finance 67(4): 1173–1217.
Lerner, Josh. 2009. Boulevard of Broken Dreams: Ries, Eric. 2011. The Lean Startup: How Today’s
Why Public Efforts to Boost Entrepreneurship and Entrepreneurs Use Continuous Innovation to Create
Venture Capital Have Failed—And What to Do about Radically Successful Businesses. Crown Business
It. Princeton University Press. Publishing.
Lerner, Josh. 2012. The Architecture of Innovation: Ritter, Jay R. 2014. “Initial Public Offerings:
The Economics of Creative Organizations. Boston, MA: Updated Statistics.”
Harvard Business School Press. Robb, Alicia M., and David T. Robinson. 2014.
Levere, Jane L. 2013. “Airbnb Campaign “The Capital Structure Decisions of New Firms.”
Uses Birdhouses to Widen Its Reach.” New Review of Financial Studies 27(1): 153–79.
York Times, December 15. http://www.nytimes Rosenberg, Nathan. 1994. “Economic
.com/2013/12/16/business/media/airbnb Experiments.” In Chap. 5 in Exploring the Black
-campaign-uses-birdhouses-to-widen-its-reach.html. Box: Technology, Economics, and History. Cambridge
Manso, Gustavo. 2011. “Motivating Innova- University Press.
tion.” Journal of Finance 66(5): 1823–69. Sahlman, William A. 1990. “The Structure and
March, James G. 1991. “Exploration and Exploi- Governance of Venture-Capital Organizations.”
tation in Organizational Learning.” Organizational Journal of Financial Economics 27(2): 473–521.
Science 2(1): 71–87. Sahlman, William A. 2010. “Risk and Reward
Michelacci, Claudio, and Javier Suarez. 2004. in Venture Capital.” Harvard Business School
“Business Creation and the Stock Market.” Review Background Note 811-036.
of Economic Studies 71(2): 459–81. Sahlman, William, Ramana Nanda, Joseph B.
Nanda, Ramana, and Liz Kind. 2013. Lassiter III, and James McQuade. 2012. “Terra-
“AngelList.” Harvard Business School Case 814-036. Power.” Harvard Business School Case Study
Nanda, Ramana, and Matthew Rhodes-Kropf. 813-108.
2010. “Financing Risk and Innovation.” Harvard Saint-Paul, Gilles. 2002. “Employment
Business School Working Paper 11–013. Protection, International Specialization, and
Nanda, Ramana, and Matthew Rhodes-Kropf. Innovation.” European Economic Review 46(2):
2012. “Innovation and the Financial Guillotine.” 375–95.
Harvard Business School Working Paper 13–038. Samaniego, Roberto M. 2006. “Employment
Nanda, Ramana, and Matthew Rhodes-Kropf. Protection and High-Tech Aversion.” Review of
2013. “Investment Cycles and Startup Innovation.” Economic Dynamics 9(2): 224–41.
Journal of Financial Economics 110(2): 403–18. Samila, Sampsa, and Olav Sorenson. 2011.
Nanda, Ramana, Ken Younge, and Lee Fleming. “Venture Capital, Entrepreneurship and Economic
Forthcoming. “Innovation and Entrepreneurship Growth.” Review of Economics and Statistics 93(1):
in Renewable Energy.” In The Changing Frontier: 338–49.
Rethinking Science and Innovation Policy edited by Scherer, F. M., and Dietmar Harhoff. 2000.
Adam Jaffe and Ben Jones. National Bureau of “Technology Policy for a World of Skew-distributed
Economic Research. Outcomes.” Research Policy 29(4–5): 559–66.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 272 of 801
48 Journal of Economic Perspectives
Schumpeter, Joseph. 1911 [1934]. The Theory Policy Report, edited by Carl J. Schramm, 16–20.
of Economic Development: An Inquiry into Profits, Ewing Marion Kauffman Foundation.
Capital, Credit, Interest and the Business Cycle. Thomke, Stefan H. 2003. Experimentation Matters:
Cambridge, MA: Harvard University Press. Unlocking the Potential of New Technologies for Innova-
Schumpeter, Joseph. 1943. Capitalism, Socialism, tion. Boston, MA: Harvard Business School Press.
and Democracy. Harper Brothers, New York, NY. Tian, Xuan. 2011. “The Causes and Conse-
Shane, Scott A. 2009. Fool’s Gold? The Truth quences of Venture Capital Stage Financing.”
behind Angel Investing in America. Oxford Univer- Journal of Financial Economics 101(1): 132–59.
sity Press. Weitzman, Martin. 1979. “Optimal Search for
Stern, Scott. 2005. “Economic Experiments: The the Best Alternative.” Econometrica 47(3): 641–54.
Role of Entrepreneurship in Economic Prosperity.” Wilson, Fred. 2011. “Airbnb.” March 16. http://
In Understanding Entrepreneurship: A Research and www.avc.com/a_vc/2011/03/airbnb.html.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 273 of 801
This article has been cited by:
1. Hyeonsuh Lee. 2023. Unraveling the effect of pre-entry knowledge of founders on experimentation
in nascent industries: A configurational approach. Journal of Business Research 167, 114155. [Crossref]
2. Gerardo Reyes Ruiz. Aversion of a Person Facing the Risk of Failure When Starting a Business in
Mexico 50-81. [Crossref]
3. Pontus Braunerhjelm, Roger Svensson. 2023. Inventions, commercialization strategies, and knowledge
spillovers in SMEs. Small Business Economics 32. . [Crossref]
4. MATTHEW DENES, SABRINA T. HOWELL, FILIPPO MEZZANOTTI, XINXIN WANG,
TING XU. 2023. Investor Tax Credits and Entrepreneurship: Evidence from U.S. States. The Journal
of Finance 1. . [Crossref]
5. Indu Khurana, Alexis Habiyaremye, Veysel Avsar, Siri Terjesen. 2023. The impact of policy
uncertainty on entrepreneurial activity: a cross-country analysis. Entrepreneurship & Regional
Development 35:7-8, 593-616. [Crossref]
6. Thomas Zellweger, Todd Zenger. 2023. Entrepreneurs as Scientists: A Pragmatist Approach to
Producing Value Out of Uncertainty. Academy of Management Review 48:3, 379-408. [Crossref]
7. Shad Morris, Chad Carlos, Geoffrey M. Kistruck, Robert B. Lount, Tumsifu Elly Thomas. 2023.
The impact of growth mindset training on entrepreneurial action among necessity entrepreneurs:
Evidence from a randomized control trial. Strategic Entrepreneurship Journal 38. . [Crossref]
8. Yuan Shi. 2023. A Change of Tune: The Democratization of Market Mediation and Crossover
Production in the U.S. Commercial Music Industry. Administrative Science Quarterly 68:2, 319-354.
[Crossref]
9. Silvia Sanasi. 2023. Entrepreneurial experimentation in business model dynamics: Current
understanding and future opportunities. International Entrepreneurship and Management Journal 19:2,
805-836. [Crossref]
10. Frank P. Balz, Florian Brinkmann, Dominik K. Kanbach. 2023. The impact of independent and
heterogeneous corporate venture capital on firm efficiency. Journal of Business Venturing Insights 19,
e00384. [Crossref]
11. Rebecca Karp. 2023. Gaining Organizational Adoption: Strategically Pacing the Position of
Digital Innovations. Academy of Management Journal 66:3, 773-796. [Crossref]
12. Som Sekhar Bhattacharyya. 2023. Elucidating technology-based social entrepreneurs’ scale and
scope perspectives: a study of environmental and organizational variables. International Journal of
Organizational Analysis 31:4, 1001-1023. [Crossref]
13. Jade Y. Lo, Lei Xu, Haemin Dennis Park. 2023. Hot Markets, Sociocognitive Cues, and New
Market Entry in the U.S. Venture Capital Industry. Entrepreneurship Theory and Practice 37,
104225872311736. [Crossref]
14. Pravin Krishna, Andrei A. Levchenko, Lin Ma, William F. Maloney. 2023. Growth and risk: A view
from international trade. Journal of International Economics 142, 103755. [Crossref]
15. Jianhong Cao, Siong Hook Law, Abdul Rahim Abdul Samad, Wan Norhidayah W. Mohamad. 2023.
Internal mechanism analysis of the financial vanishing effect on green growth: Evidence from China.
Energy Economics 120, 106579. [Crossref]
16. Ben Spigel, Fizza Khalid, David Wolfe. 2023. Alacrity: a new model for venture acceleration.
International Entrepreneurship and Management Journal 19:1, 237-259. [Crossref]
17. Andrés Redchuk, Federico Walas Mateo, Guadalupe Pascal, Julian Eloy Tornillo. 2023. Adoption Case
of IIoT and Machine Learning to Improve Energy Consumption at a Process Manufacturing Firm,
under Industry 5.0 Model. Big Data and Cognitive Computing 7:1, 42. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 274 of 801
18. Dante Ayaviri-Nina, Jessica Cáceres-Guzmán, Gabith Miriam Quispe Fernández, Alba Isabel
Maldonado-Nuñez. 2023. The Determinants of Success in Entrepreneurship: A Study in the Urban
Area of Ecuador. Sustainability 15:6, 5277. [Crossref]
19. Christoph Carnehl, Johannes Schneider. 2023. on Risk and Time Pressure: When to Think and
When to Do. Journal of the European Economic Association 21:1, 1-47. [Crossref]
20. Nataliya Langburd Wright, Rembrand Koning, Tarun Khanna. 2023. Judging foreign startups.
Strategic Management Journal 58. . [Crossref]
21. Qianqian Zhang, Chunzi Jiang, Xiaomei Zhang. 2023. Exploration or exploitation? A study on equity
incentive design, dynamic decision making, and economic consequences. PLOS ONE 18:1, e0277965.
[Crossref]
22. Joshua S. Gans. 2023. Experimental Choice and Disruptive Technologies. Management Science 82. .
[Crossref]
23. Qingjie Zhou, Dongyao Yu, Feng Xu, Jiamin Sun. 2023. The Impact of Institutional Friction Cost
on Economic Growth: Evidence from OECD Countries. Sustainability 15:1, 427. [Crossref]
24. Pravin Krishna, Andrei A. Levchenko, Lin Ma, William F. Maloney. 2023. Growth and Risk: A View
from International Trade. SSRN Electronic Journal 59. . [Crossref]
25. Silvia Dalla Fontana, Ramana Nanda. 2023. Innovating to Net Zero: Can Venture Capital and Start-
Ups Play a Meaningful Role?. Entrepreneurship and Innovation Policy and the Economy 2, 79-105.
[Crossref]
26. Maxime Bonelli. 2023. The Adoption of Artificial Intelligence by Venture Capitalists. SSRN Electronic
Journal 3. . [Crossref]
27. Ramana Nanda, Matthew Rhodes-Kropf. Financing Booms in Venture Capital 1-5. [Crossref]
28. Yongwook Paik, Sunu Kim. 2023. How Incumbent Firms Cope with Economic Policy Uncertainty:
The Case of Corporate Venture Capital Investments in Innovative Startups as a Real Option. SSRN
Electronic Journal 73. . [Crossref]
29. Sylvia Hubner-Benz, Michael Frese. Creativity in entrepreneurship: Dancing between nothing and
structure 377-392. [Crossref]
30. Nataliya Wright. 2023. Where Strategy Matters: Evidence from a Global Startup Field Study. SSRN
Electronic Journal 39. . [Crossref]
31. Juanita Gonzalez-Uribe, Robyn Klingler-Vidra, Su Wang, Xiang Yin. 2023. The Broader Impact
of Venture Capital on Innovation: Reducing Entrepreneurial Capability Constraints Through Due
Diligence. SSRN Electronic Journal . [Crossref]
32. Jorge Guzman, Fiona E. Murray, Scott Stern, Heidi Williams. 2023. Accelerating Innovation
Ecosystems: The Promise and Challenges of Regional Innovation Engines. SSRN Electronic Journal
113. . [Crossref]
33. Yang Pan, Wei-Ling Song. 2023. Tech Giants and New Entry Threats. SSRN Electronic Journal 80. .
[Crossref]
34. Victor M. Bennett, Aaron K. Chatterji. 2023. The entrepreneurial process: Evidence from a nationally
representative survey. Strategic Management Journal 44:1, 86-116. [Crossref]
35. Anne Tryba, Holger Patzelt, Nicola Breugst. 2023. Knowledge Diversity and Venture Growth: The
Contingent Effects of Early Planning and Experimentation. British Journal of Management 34:1,
343-362. [Crossref]
36. Silvia Sanasi, Jacopo Manotti, Antonio Ghezzi. 2022. Achieving Agility in High-Reputation Firms:
Agile Experimentation Revisited. IEEE Transactions on Engineering Management 69:6, 3529-3545.
[Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 275 of 801
37. Suting Hong, Guangwei Li, Wangshuai Wang, Zhiqi Zhao. 2022. Does winning a venture competition
encourage entrepreneurial exploration? Evidence from China. China Economic Review 76, 101876.
[Crossref]
38. Mike G. Tsionas, Pankaj C. Patel. 2022. An entrepreneur's dilemma: An optimal stopping rule in
pivoting. Managerial and Decision Economics 43:8, 3498-3515. [Crossref]
39. Yves Guéron, Jihong Lee. 2022. Learning by Selling, Knowledge Spillovers, and Patents*. The Journal
of Industrial Economics 70:4, 867-912. [Crossref]
40. Ari Hyytinen, Petri Rouvinen, Mika Pajarinen, Joosua Virtanen. 2022. Ex Ante Predictability of Rapid
Growth: A Design Science Approach. Entrepreneurship Theory and Practice 30, 104225872211282.
[Crossref]
41. Shai Bernstein. 2022. The Effects of Public and Private Equity Markets on Firm Behavior. Annual
Review of Financial Economics 14:1, 295-318. [Crossref]
42. Randall Morck. 2022. Kindleberger Cycles: Method in the Madness of Crowds?. Annual Review of
Financial Economics 14:1, 563-585. [Crossref]
43. Thomas Zellweger, Todd Zenger. 2022. Entrepreneurs as scientists, Bayesian inference, and belief
revision. Journal of Business Venturing Insights 18, e00350. [Crossref]
44. Sandy Yu, Lee Fleming. 2022. Regional crowdfunding and high tech entrepreneurship. Research Policy
51:9, 104348. [Crossref]
45. Minsi Liu, Kevin Lo. 2022. Eco-cities as urban laboratories of Chinese ecological modernization: eco-
experimentation and the modernization of urban governance. Urban Geography 185, 1-25. [Crossref]
46. Olivia Aronson, Irene Henriques. 2022. Shared Value Creation in Equivocal CSR Environments: A
Configuration Approach. Journal of Business Ethics 38. . [Crossref]
47. Frank P. Balz, Benjamin M. Bugl, Dominik K. Kanbach. 2022. New Venture Value Creation in
Syndicates Between Independent and Corporate Investors. International Journal of Innovation and
Technology Management 19:06. . [Crossref]
48. Antonella Zucchella, Pietro Previtali, Roger Strange. 2022. Proactive and reactive views in the
transition towards circular business models. A grounded study in the plastic packaging industry.
International Entrepreneurship and Management Journal 18:3, 1073-1102. [Crossref]
49. Rembrand Koning, Sharique Hasan, Aaron Chatterji. 2022. Experimentation and Start-up
Performance: Evidence from A/B Testing. Management Science 68:9, 6434-6453. [Crossref]
50. Andrea Contigiani, Trevor Young‐Hyman. 2022. Experimentation, planning, and structure in early‐
stage ventures: Evidence from pitch decks. Strategic Entrepreneurship Journal 16:3, 425-459. [Crossref]
51. Joshua D Gottlieb, Richard R Townsend, Ting Xu. 2022. Does Career Risk Deter Potential
Entrepreneurs?. The Review of Financial Studies 35:9, 3973-4015. [Crossref]
52. Christian Hopp, Stefan Rose, Jermain Kaminski. 2022. The Pervasive Role of Campaign and Product-
Related Uncertainties in Inhibiting Crowdfunding Success. Journal of Risk and Financial Management
15:8, 370. [Crossref]
53. Diego Zunino, Gary Dushnitsky, Mirjam van Praag. 2022. How Do Investors Evaluate Past
Entrepreneurial Failure? Unpacking Failure Due to Lack of Skill versus Bad Luck. Academy of
Management Journal 65:4, 1083-1109. [Crossref]
54. Kathryn Shaw, Anders Sørensen. 2022. Coming of age: Watching young entrepreneurs become
successful. Labour Economics 77, 102033. [Crossref]
55. John S. Chen, Daniel W. Elfenbein, Hart E. Posen, Ming zhu Wang. 2022. The Problems
and Promise of Entrepreneurial Partnerships: Decision-Making, Overconfidence, and Learning in
Founding Teams. Academy of Management Review 47:3, 489-520. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 276 of 801
56. Michael Wessel, Ferdinand Thies, Alexander Benlian. 2022. The role of prototype fidelity in
technology crowdfunding. Journal of Business Venturing 37:4, 106220. [Crossref]
57. Timo Ehrig, Jens Schmidt. 2022. Theory‐based learning and experimentation: How strategists
can systematically generate knowledge at the edge between the known and the unknown. Strategic
Management Journal 43:7, 1287-1318. [Crossref]
58. Randall E. Westgren, Travis L. Holmes. 2022. Entrepreneurial Beliefs and Agency under Knightian
Uncertainty. Philosophy of Management 21:2, 199-217. [Crossref]
59. Jose P. Arrieta, Yash R. Shrestha. 2022. On the strategic value of equifinal choice. Journal of
Organization Design 11:2, 37-45. [Crossref]
60. Jose-Luis Hervas-Oliver, Carles Boronat-Moll, Marta Peris-Ortiz, Ronald Rojas-Alvarado. 2022.
Understanding spatial networking and industrial district evolution from firms’ strategies. European
Planning Studies 1, 1-29. [Crossref]
61. Jorge Melegati, Henry Edison, Xiaofeng Wang. 2022. XPro: A Model to Explain the Limited
Adoption and Implementation of Experimentation in Software Startups. IEEE Transactions on
Software Engineering 48:6, 1929-1946. [Crossref]
62. Shashi Bhushan Kumar, Nandan Sudarsanam. 2022. A Decision-Making Framework for
Entrepreneurial Venture in Emerging Economies. International Journal of Global Business and
Competitiveness 17:1, 11-23. [Crossref]
63. Wang Chengbin, Wang Hongbin, Dai Min, Fang Yongyan. 2022. Lean Startup Approaches(LSAs):
Convergence, Integration and Improvement. Technological Forecasting and Social Change 179, 121640.
[Crossref]
64. Simon Mayer. 2022. Financing breakthroughs under failure risk. Journal of Financial Economics 144:3,
807-848. [Crossref]
65. Daniel S. Hamermesh, Andrew K. Leigh. 2022. “Beauty too rich for use”: Billionaires’ assets and
attractiveness. Labour Economics 76, 102153. [Crossref]
66. Luís Guimarães, Pedro Mazeda Gil. 2022. Explaining the Labor Share: Automation Vs Labor Market
Institutions. Labour Economics 75, 102146. [Crossref]
67. Jorge Melegati, Eduardo Guerra, Xiaofeng Wang. 2022. HyMap: Eliciting hypotheses in early-
stage software startups using cognitive mapping. Information and Software Technology 144, 106807.
[Crossref]
68. Lauren Lanahan, Daniel Erian Armanios, Amol M. Joshi. 2022. Inappropriateness Penalty,
Desirability Premium: What Do More Certifications Actually Signal?. Organization Science 33:2,
854-871. [Crossref]
69. Arti Grover, Somik V. Lall, William F. Maloney. Local Economic Development Policies 187-220.
[Crossref]
70. Margarida Madaleno, Max Nathan, Henry Overman, Sevrin Waights. 2022. Incubators, accelerators
and urban economic development. Urban Studies 59:2, 281-300. [Crossref]
71. Jeroen Mahieu, Francesca Melillo, Peter Thompson. 2022. The long‐term consequences of
entrepreneurship: Earnings trajectories of former entrepreneurs. Strategic Management Journal 43:2,
213-236. [Crossref]
72. Mirjana Grčić Fabić. Perspectives and Challenges in the Development of the Croatian Digital Startup
Sector 139-156. [Crossref]
73. Alexander Styhre. Status Relations and Associations in Life Science Venturing 145-166. [Crossref]
74. Ufuk Akcigit, Douglas Hanley, Stefanie Stantcheva. 2022. Optimal Taxation and R&D Policies.
Econometrica 90:2, 645-684. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 277 of 801
75. Vladimir A. Gatchev, Christo A. Pirinsky, Buvaneshwaran Venugopal. 2022. A language-based
approach to measuring creative exploration. Research Policy 51:1, 104426. [Crossref]
76. Mari Sako, Matthias Qian, Mark Verhagen. 2022. Knowledge Similarity Among Founders and Joiners:
Impact on Venture Scaleup in Fintech and Lawtech. SSRN Electronic Journal 20. . [Crossref]
77. William F. Maloney, Andres Zambrano. 2022. Learning to Learn: Experimentation, Entrepreneurial
Capital, and Development. SSRN Electronic Journal 41. . [Crossref]
78. Simone Santamaria. 2022. Portfolio Entrepreneurs’ Behavior and Performance: A Resource
Redeployment Perspective. Management Science 68:1, 333-354. [Crossref]
79. Joshua S. Gans. 2022. A Theory of Visionary Disruption. SSRN Electronic Journal 57. . [Crossref]
80. Joshua Gans. 2022. A Theory of Visionary Disruption. SSRN Electronic Journal 57. . [Crossref]
81. Simon Rudat. 2022. Philanthropic Investments in Deep Tech Start-Ups: An Exploratory Study.
SSRN Electronic Journal 139. . [Crossref]
82. Marco Da Rin, Marina Di Giacomo, Alessandro Sembenelli. 2022. Leverage at Entry and the Growth
of Newly Incorporated Companies. SSRN Electronic Journal 108. . [Crossref]
83. Louis Kaplow. 2022. Optimal Income Taxation. SSRN Electronic Journal 44. . [Crossref]
84. Rebecca Janßen, Reinhold Kesler, Michael Kummer, Joel Waldfogel. 2022. Gdpr and the Lost
Generation of Innovative Apps. SSRN Electronic Journal 54. . [Crossref]
85. Mario Daniele Amore, Annamaria Conti, Valerio Pelucco. 2022. Micro VC. SSRN Electronic Journal
35. . [Crossref]
86. Stephen Michael Impink. 2022. Outsourcing IT and Technological Differentiation: Evidence from
Digital Startups. SSRN Electronic Journal 72. . [Crossref]
87. William F. Maloney, Andres Zambrano. Learning to Learn: Experimentation, Entrepreneurial
Capital, and Development 120, . [Crossref]
88. Roberto Pessoa de Queiroz Falcão, Michel Mott Machado, Eduardo Picanço Cruz, Robson
Moreira Cunha. 2021. TRAJETÓRIAS EMERGENTES DE STARTUPS BRASILEIRAS-
CANADENSES À LUZ DO MODELO DE UPPSALA, EMPREENDEDORISMO DE
IMIGRANTES E DA EFFECTUATION. REAd. Revista Eletrônica de Administração (Porto Alegre)
27:3, 835-869. [Crossref]
89. Mujtaba Ahsan, Martina Musteen. 2021. International opportunity development on crowdfunding
platforms: A spatial, temporal, and structural framework. International Business Review 30:6, 101912.
[Crossref]
90. Jeroen Mahieu, Francesca Melillo, Toke Reichstein, Peter Thompson. 2021. Shooting stars?
Uncertainty in hiring entrepreneurs. Strategic Entrepreneurship Journal 15:4, 526-567. [Crossref]
91. JASON RODERICK DONALDSON, GIORGIA PIACENTINO, ANJAN THAKOR. 2021.
Intermediation Variety. The Journal of Finance 76:6, 3103-3152. [Crossref]
92. Elena Cefis, Franco Malerba, Orietta Marsili, Luigi Orsenigo. 2021. Time to exit: “revolving door
effect” or “Schumpeterian gale of creative destruction”?. Journal of Evolutionary Economics 31:5,
1465-1494. [Crossref]
93. Irfan Qureshi, Donghyun Park, Gustavo Atilio Crespi, Jose Miguel Benavente. 2021. Trends and
determinants of innovation in Asia and the Pacific vs. Latin America and the Caribbean. Journal of
Policy Modeling 43:6, 1287-1309. [Crossref]
94. William H. Janeway, Ramana Nanda, Matthew Rhodes-Kropf. 2021. Venture Capital Booms and
Start-Up Financing. Annual Review of Financial Economics 13:1, 111-127. [Crossref]
95. Lin William Cong, Sabrina T. Howell. 2021. Policy Uncertainty and Innovation: Evidence from Initial
Public Offering Interventions in China. Management Science 67:11, 7238-7261. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 278 of 801
96. Geerten van de Kaa, Linda Kamp. 2021. Exploring design dominance in early stages of the dominance
process: The case of airborne wind energy. Journal of Cleaner Production 321, 128918. [Crossref]
97. Hong Luo, Jeffrey Macher, Michael Wahlen. 2021. Judgment Aggregation in Creative Production:
Evidence from the Movie Industry. Management Science 67:10, 6358-6377. [Crossref]
98. Jillian Grennan, Roni Michaely. 2021. FinTechs and the Market for Financial Analysis. Journal of
Financial and Quantitative Analysis 56:6, 1877-1907. [Crossref]
99. Ajay Agrawal, Joshua S. Gans, Scott Stern. 2021. Enabling Entrepreneurial Choice. Management
Science 67:9, 5510-5524. [Crossref]
100. Joshua L. Krieger. 2021. Trials and Terminations: Learning from Competitors’ R&D Failures.
Management Science 67:9, 5525-5548. [Crossref]
101. Ziliang Liu, Shengjun Zhu. 2021. Changing institutional context and regional industrial dynamics:
New evidence from the establishment of administrative approval centers in China. Growth and Change
52:3, 1271-1294. [Crossref]
102. Rafael Fazzi Bortolini, Marcelo Nogueira Cortimiglia, Angela de Moura Ferreira Danilevicz, Antonio
Ghezzi. 2021. Lean Startup: a comprehensive historical review. Management Decision 59:8, 1765-1783.
[Crossref]
103. Chia-Hui Chen, Junichiro Ishida, Wing Suen. 2021. Reputation Concerns in Risky Experimentation.
Journal of the European Economic Association 19:4, 1981-2021. [Crossref]
104. Takao Fujiwara. Star Biotech Start-ups' Past R&D Investment in the Financial Crisis Time 1-9.
[Crossref]
105. Leonardo Mazzoni, Luciana Lazzeretti, Niccolò Innocenti. 2021. Entrepreneurship, complexity and
the emergent order in the techno-economic scenario of the twenty-first century. Evidence from a field
study in Tuscany. Industry and Innovation 28:5, 570-593. [Crossref]
106. Ximena Alejandra Flechas Chaparro, Leonardo Augusto de Vasconcelos Gomes. 2021. Pivot decisions
in startups: a systematic literature review. International Journal of Entrepreneurial Behavior & Research
27:4, 884-910. [Crossref]
107. Sabrina T Howell. 2021. Learning from Feedback: Evidence from New Ventures*. Review of Finance
25:3, 595-627. [Crossref]
108. Onesun Steve Yoo, Tingliang Huang, Kenan Arifoğlu. 2021. A Theoretical Analysis of the Lean
Start-up Method. Marketing Science 40:3, 395-412. [Crossref]
109. FELIX SCHEUENSTUHL, PETER M. BICAN, ALEXANDER BREM. 2021. HOW CAN
THE LEAN STARTUP APPROACH IMPROVE THE INNOVATION PROCESS OF
ESTABLISHED COMPANIES? AN EXPERIMENTAL APPROACH. International Journal of
Innovation Management 25:03, 2150029. [Crossref]
110. Antonella Zucchella. 2021. International entrepreneurship and the internationalization phenomenon:
taking stock, looking ahead. International Business Review 30:2, 101800. [Crossref]
111. Stephen X. Zhang, Renfei Gao, Nicolás Odeh, Michael Leatherbee. 2021. A microfoundational model
of real options reasoning: The roles of individual search propensity and perceived uncertainty. Strategic
Entrepreneurship Journal 15:1, 98-120. [Crossref]
112. Nicolai J. Foss, Peter G. Klein, Lasse B. Lien, Thomas Zellweger, Todd Zenger. 2021. Ownership
competence. Strategic Management Journal 42:2, 302-328. [Crossref]
113. Gilles Chemla, Katrin Tinn. How Wise are Crowds on Crowdfunding Platforms? 397-406. [Crossref]
114. Michael Fritsch, Michael Wyrwich. Entrepreneurship, Gründungen, Marktdynamik 5-19. [Crossref]
115. Martin Andersson, Johan P. Larsson. Geography and Entrepreneurship 1361-1373. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 279 of 801
116. Ryan Shuwei Hsu, Aichia Chuang, An-Chih Wang. 2021. Business founders' work design and new
venture development. Journal of Business Venturing 36:1, 106000. [Crossref]
117. Simon Neugebauer, Lukas Rippitsch, Florian Sobieczky, Manuela Geiβ. 2021. Explainability of AI-
predictions based on psychological profiling. Procedia Computer Science 180, 1003-1012. [Crossref]
118. Adrian L. Merida, Vera Rocha. 2021. It's about time: The timing of entrepreneurial experience and
the career dynamics of university graduates. Research Policy 50:1, 104135. [Crossref]
119. Sasan Mansouri, Paul P. Momtaz. 2021. Financing Sustainable Entrepreneurship: ESG Measurement,
Valuation, and Performance in Token Offerings. SSRN Electronic Journal . [Crossref]
120. Nuri Ersahin, Ruidi Huang, Naveen Khanna. 2021. The Achilles Heel of Reputable VCs. SSRN
Electronic Journal 119. . [Crossref]
121. Xiqian Cai, Lata Gangadharan, Yi Lu, Xiaojian Zhao. 2021. Does a Sea Fishing Legacy Explain
Differences in Risk Attitudes and Entrepreneurship?. SSRN Electronic Journal 95. . [Crossref]
122. Randall K. Morck. 2021. Kindleberger Cycles & Economic Growth: Method in the Madness of
Crowds?. SSRN Electronic Journal 124. . [Crossref]
123. Nataliya Wright, Rembrand Koning, Tarun Khanna. 2021. Judging Foreign Startups. SSRN Electronic
Journal 47. . [Crossref]
124. Yifei Zhang. 2021. Corporate Venture Capital and Firm Scope. SSRN Electronic Journal 26. .
[Crossref]
125. Andrey Malenko, Ramana Nanda, Matthew Rhodes-Kropf, Savitar Sundaresan. 2021. Investment
Committee Voting and the Financing of Innovation. SSRN Electronic Journal 128. . [Crossref]
126. Joshua S. Gans. 2021. A Synthetic Model of Disruption and Experimentation. SSRN Electronic Journal
129. . [Crossref]
127. Elena Novelli, Chiara Spina. 2021. When do Entrepreneurs Benefit from Acting Like Scientists? A
Field Experiment in the UK. SSRN Electronic Journal 1. . [Crossref]
128. Joshua Gans. 2021. A Synthetic Model of Disruption and Experimentation. SSRN Electronic Journal
129. . [Crossref]
129. Jorge Guzman. 2021. Treatment Effects in Strategic Management: With an Application to Choosing
Early Stage Venture Capital. SSRN Electronic Journal 21. . [Crossref]
130. Suting Hong, Pierre Mella-Barral. 2021. Heterogeneous Venture Capitalists and Syndicate Switching.
SSRN Electronic Journal 138. . [Crossref]
131. Vladimir A. Gatchev, Christo Angelov Pirinsky, Buvaneshwaran Venugopal. 2021. A Language-Based
Approach to Measuring Creative Exploration. SSRN Electronic Journal 60. . [Crossref]
132. Michael Leatherbee, Riitta Katila. 2020. The lean startup method: Early‐stage teams and hypothesis‐
based probing of business ideas. Strategic Entrepreneurship Journal 14:4, 570-593. [Crossref]
133. Jan Konietzko, Brian Baldassarre, Phil Brown, Nancy Bocken, Erik Jan Hultink. 2020. Circular
business model experimentation: Demystifying assumptions. Journal of Cleaner Production 277,
122596. [Crossref]
134. Ehsan Azarmsa, Lin William Cong. 2020. Persuasion in relationship finance. Journal of Financial
Economics 138:3, 818-837. [Crossref]
135. H.R. Ganesh, P. S. Aithal. 2020. Extending the Concept of Delayed Gratification to Retail Start-
ups in India: An Imperative Strategy for the Success, Long-Term Sustainability, and Protection of
Founding Members’ Majority Shareholding. International Journal of Management, Technology, and
Social Sciences 252-265. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 280 of 801
136. Jorge Guzman, Scott Stern. 2020. The State of American Entrepreneurship: New Estimates of the
Quantity and Quality of Entrepreneurship for 32 US States, 1988–2014. American Economic Journal:
Economic Policy 12:4, 212-243. [Abstract] [View PDF article] [PDF with links]
137. Ahmad Beltagui, Achilleas Sesis, Nikolaos Stylos. 2020. A bricolage perspective on democratising
innovation: The case of 3D printing in makerspaces. Technological Forecasting and Social Change
120453. [Crossref]
138. Javier Monllor, Aracely Soto-Simeone. 2020. The impact that exposure to digital fabrication
technology has on student entrepreneurial intentions. International Journal of Entrepreneurial Behavior
& Research 26:7, 1505-1523. [Crossref]
139. Jermain C. Kaminski, Christian Hopp. 2020. Predicting outcomes in crowdfunding campaigns with
textual, visual, and linguistic signals. Small Business Economics 55:3, 627-649. [Crossref]
140. Jorge Melegati, Eduardo Guerra, Xiaofeng Wang. 2020. Understanding Hypotheses Engineering in
Software Startups through a Gray Literature Review. Information and Software Technology 51, 106465.
[Crossref]
141. Suting Hong, Konstantinos Serfes, Veikko Thiele. 2020. Competition in the venture capital market
and the success of startup companies: Theory and evidence. Journal of Economics & Management
Strategy 29:4, 741-791. [Crossref]
142. Doan Winkel, Justin Wilcox, Atul Teckchandani. 2020. The 60-Minute MVP. Entrepreneurship
Education and Pedagogy 3:4, 371-386. [Crossref]
143. Anna Goldstein, Claudia Doblinger, Erin Baker, Laura Díaz Anadón. 2020. Patenting and business
outcomes for cleantech startups funded by the Advanced Research Projects Agency-Energy. Nature
Energy 5:10, 803-810. [Crossref]
144. Robin De Cock, Johan Bruneel, Annelies Bobelyn. 2020. Making the lean start-up method work:
The role of prior market knowledge. Journal of Small Business Management 58:5, 975-1002. [Crossref]
145. Kieron J. Meagher, Arlene Wong, Klaus G. Zauner. 2020. A competitive analysis of fail fast: Shakeout
and uncertainty about consumer tastes. Journal of Economic Behavior & Organization 177, 589-600.
[Crossref]
146. Kellilynn M. Frias, Deidre L. Popovich, Dale F. Duhan, Robert F. Lusch. 2020. Perceived Market
Risk in New Ventures: A Study of Early-Phase Business Angel Investment Screening. Journal of
Macromarketing 40:3, 339-354. [Crossref]
147. Jorge Melegati, Rafael Chanin, Afonso Sales, Rafael Prikladnicki, Xiaofeng Wang. MVP and
experimentation in software startups: a qualitative survey 322-325. [Crossref]
148. Ross Brown, Augusto Rocha, Marc Cowling. 2020. Financing entrepreneurship in times of crisis:
Exploring the impact of COVID-19 on the market for entrepreneurial finance in the United Kingdom.
International Small Business Journal: Researching Entrepreneurship 38:5, 380-390. [Crossref]
149. Alexander Styhre. 2020. Thinly and Thickly Capitalized Projects: Theorizing the Role of the Finance
Markets and Capital Supply in Project Management Studies. Project Management Journal 51:4,
378-388. [Crossref]
150. Tom Vanacker, Daniel P. Forbes, Mirjam Knockaert, Sophie Manigart. 2020. Signal Strength,
Media Attention, and Resource Mobilization: Evidence from New Private Equity Firms. Academy of
Management Journal 63:4, 1082-1105. [Crossref]
151. Yaroslav Rosokha, Kenneth Younge. 2020. Motivating Innovation: The Effect of Loss Aversion on
the Willingness to Persist. The Review of Economics and Statistics 102:3, 569-582. [Crossref]
152. Rory M. McDonald, Kathleen M. Eisenhardt. 2020. Parallel Play: Startups, Nascent Markets, and
Effective Business-model Design. Administrative Science Quarterly 65:2, 483-523. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 281 of 801
153. Ewa Sońta-Drączkowska, Matthias Mrożewski. 2020. Exploring the Role of Project Management in
Product Development of New Technology-Based Firms. Project Management Journal 51:3, 294-311.
[Crossref]
154. Giulio Bosio, Federica Origo. 2020. Who gains from active learning in higher education?. Education
Economics 28:3, 311-331. [Crossref]
155. Ploypailin Kijkasiwat, Pongsutti Phuensane. 2020. Innovation and Firm Performance: The
Moderating and Mediating Roles of Firm Size and Small and Medium Enterprise Finance. Journal of
Risk and Financial Management 13:5, 97. [Crossref]
156. Gilles Chemla, Katrin Tinn. 2020. Learning Through Crowdfunding. Management Science 66:5,
1783-1801. [Crossref]
157. Kim Claes, Balagopal Vissa. 2020. Does Social Similarity Pay Off? Homophily and Venture Capitalists’
Deal Valuation, Downside Risk Protection, and Financial Returns in India. Organization Science 31:3,
576-603. [Crossref]
158. Ratan J.S. Dheer, Tomasz Lenartowicz. 2020. Effect of generational status on immigrants’ intentions
to start new ventures: The role of cognitions. Journal of World Business 55:3, 101069. [Crossref]
159. Alessandro Spiganti. 2020. Can Starving Start‐ups Beat Fat Labs? A Bandit Model of Innovation with
Endogenous Financing Constraint*. The Scandinavian Journal of Economics 122:2, 702-731. [Crossref]
160. Pierre Azoulay, Benjamin F. Jones, J. Daniel Kim, Javier Miranda. 2020. Age and High-Growth
Entrepreneurship. American Economic Review: Insights 2:1, 65-82. [Abstract] [View PDF article] [PDF
with links]
161. Sandeep D. Pillai, Brent Goldfarb, David A. Kirsch. 2020. The origins of firm strategy: Learning by
economic experimentation and strategic pivots in the early automobile industry. Strategic Management
Journal 41:3, 369-399. [Crossref]
162. Erin L. Scott, Pian Shu, Roman M. Lubynsky. 2020. Entrepreneurial Uncertainty and Expert
Evaluation: An Empirical Analysis. Management Science 66:3, 1278-1299. [Crossref]
163. Arnaldo Camuffo, Alessandro Cordova, Alfonso Gambardella, Chiara Spina. 2020. A Scientific
Approach to Entrepreneurial Decision Making: Evidence from a Randomized Control Trial.
Management Science 66:2, 564-586. [Crossref]
164. 2020. Innovation: Organization & Management Special Issue on Experimentation for innovation
What’s now and what’s next. Innovation 22:1, 91-96. [Crossref]
165. Alexander Styhre. Governing Innovation-Led Economies: The Role of Business Creation and
Creativity 37-77. [Crossref]
166. Jorge Melegati, Xiaofeng Wang. Hypotheses Elicitation in Early-Stage Software Startups Based on
Cognitive Mapping 211-220. [Crossref]
167. Martin Andersson, Johan P. Larsson. Geography and Entrepreneurship 1-13. [Crossref]
168. Sanjay Goel, Ranjan Karri. 2020. Entrepreneurial aspirations and poverty reduction: the role of
institutional context. Entrepreneurship & Regional Development 32:1-2, 91-111. [Crossref]
169. Yves Guéron, Jihong Lee. 2020. Learning by Selling, Knowledge Spillovers, and Patents. SSRN
Electronic Journal . [Crossref]
170. Ganesha H. R., P. S. Aithal. 2020. Extending the Concept of Delayed Gratification to Retail Start-
ups in India: An Imperative Strategy for the Success, Long-Term Sustainability, and Protection of
Founding Members’ Majority Shareholding. SSRN Electronic Journal . [Crossref]
171. John S. Chen, Daniel W. Elfenbein, Hart E. Posen, Ming zhu Wang. 2020. Pivot Rules for
(Overconfident) Entrepreneurs. SSRN Electronic Journal 47. . [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 282 of 801
172. Ruiqing Cao. 2020. Crowd-Based Rankings and Frictions in New Venture Finance. SSRN Electronic
Journal . [Crossref]
173. Ernad Kahrović. 2020. Entrepreneurial universities and intermediary organizations as a success factor
in SMEs: Literature review. Ekonomika preduzeca 68:3-4, 229-247. [Crossref]
174. Christian Granz, Marisa Henn, Eva Lutz. Research on Venture Capitalists’ and Business Angels’
Investment Criteria: A Systematic Literature Review 105-136. [Crossref]
175. Matthew Denes, Sabrina Howell, Filippo Mezzanotti, Xinxin Wang, Ting Xu. 2020. Investor Tax
Credits and Entrepreneurship: Evidence from U.S. States. SSRN Electronic Journal 2. . [Crossref]
176. Michael Ewens, Ramana Nanda, Christopher Stanton. 2020. The Evolution of CEO Compensation
in Venture Capital Backed Startups. SSRN Electronic Journal 59. . [Crossref]
177. Snehal Banerjee, Martin Szydlowski. 2020. Friends Don’t Lie: Monitoring and Communication With
Risky Investments. SSRN Electronic Journal 62. . [Crossref]
178. Ruiqing Cao, Rembrand Koning, Ramana Nanda. 2020. Biased Sampling of Early Users and the
Direction of Startup Innovation. SSRN Electronic Journal 5. . [Crossref]
179. Steven Chong Xiao, Serena Wenjing Xiao. 2020. Renter Protection and Entrepreneurship. SSRN
Electronic Journal 117. . [Crossref]
180. Orie Shelef, Robert Wuebker, Jay B. Barney. 2020. Heisenberg Effects On Business Ideas. SSRN
Electronic Journal 29. . [Crossref]
181. Liinus Hietaniemi, simone santamaria, Aleksandra Kacperczyk, Juhana Peltonen. 2020.
Entrepreneurial Hiring Strategies in Standalone and Portfolio Ventures. SSRN Electronic Journal 10. .
[Crossref]
182. Elżbieta Pohulak-Żołędowska. 2019. Changes of Venture Capital Financing in the USA and in Europe.
Studia i Materiały Wydziału Zarządzania UW 2/2018 cz.2:29, 41-46. [Crossref]
183. Nirosha Hewa Wellalage, Viviana Fernandez. 2019. Innovation and SME finance: Evidence from
developing countries. International Review of Financial Analysis 66, 101370. [Crossref]
184. Åsa Lindholm-Dahlstrand, Martin Andersson, Bo Carlsson. 2019. Entrepreneurial experimentation:
a key function in systems of innovation. Small Business Economics 53:3, 591-610. [Crossref]
185. Carolin Bock, Martin Watzinger. 2019. The Capital Gains Tax: A Curse but Also a Blessing for
Venture Capital Investment. Journal of Small Business Management 57:4, 1200-1231. [Crossref]
186. Horia Tigau. 2019. The Importance of FDI on Stimulating Entrepreneurship – A Regional Study
in the Case of Romania. Proceedings of the International Conference on Applied Statistics 1:1, 484-495.
[Crossref]
187. Sari Pekkala Kerr, William R. Kerr, Margaret Dalton. 2019. Risk attitudes and personality traits
of entrepreneurs and venture team members. Proceedings of the National Academy of Sciences 116:36,
17712-17716. [Crossref]
188. Vedapradha. R, Hariharan Ravi. 2019. Challenges or survival instinct of Tibetan entrepreneurs. Asia
Pacific Journal of Innovation and Entrepreneurship 13:2, 203-213. [Crossref]
189. Thomas F. Hellmann, Paul H. Schure, Chloe Tergiman, Dan H. Vo. 2019. Ownership dynamics
within founder teams: The role of external financing. Strategic Entrepreneurship Journal 13:3, 256-287.
[Crossref]
190. Ian Appel, Joan Farre-Mensa, Elena Simintzi. 2019. Patent trolls and startup employment. Journal of
Financial Economics 133:3, 708-725. [Crossref]
191. Antonio Ghezzi. 2019. Digital startups and the adoption and implementation of Lean Startup
Approaches: Effectuation, Bricolage and Opportunity Creation in practice. Technological Forecasting
and Social Change 146, 945-960. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 283 of 801
192. Ethan Gifford, Maureen McKelvey. 2019. Knowledge-Intensive Entrepreneurship and S3:
Conceptualizing Strategies for Sustainability. Sustainability 11:18, 4824. [Crossref]
193. Sharon Belenzon, Victor Manuel Bennett, Andrea Patacconi. 2019. Flexible Production and
Entry: Institutional, Technological, and Organizational Determinants. Strategy Science 4:3, 193-216.
[Crossref]
194. Christina Lüthy, Chris Steyaert. 2019. The onto-politics of entrepreneurial experimentation:
re-reading Hans-Jörg Rheinberger’s understanding of ‘experimental systems’. Entrepreneurship &
Regional Development 31:7-8, 652-668. [Crossref]
195. Simon Quinn, Christopher Woodruff. 2019. Experiments and Entrepreneurship in Developing
Countries. Annual Review of Economics 11:1, 225-248. [Crossref]
196. Pamela Adams, Patrick Burd. 2019. Classic soccer shoes: From an informal to a formal business start-
up. The International Journal of Entrepreneurship and Innovation 20:3, 220-226. [Crossref]
197. Jonathan T. Eckhardt. Chapter 2 The Distinctive Domain as a Durable Way Forward 21-33.
[Crossref]
198. David B. Audretsch, Erik E. Lehmann, Nikolaus Seitz. 2019. Amenities, subcultures, and
entrepreneurship. Small Business Economics 28. . [Crossref]
199. Alexander Konon, Alexander S. Kritikos. 2019. Prediction based on entrepreneurship-prone
personality profiles: sometimes worse than the toss of a coin. Small Business Economics 53:1, 1-20.
[Crossref]
200. Diego Comin, Ramana Nanda. 2019. Financial Development and Technology Diffusion. IMF
Economic Review 67:2, 395-419. [Crossref]
201. Elena Cefis, Orietta Marsili. 2019. Good times, bad times: innovation and survival over the business
cycle. Industrial and Corporate Change 28:3, 565-587. [Crossref]
202. Joshua S. Gans, Scott Stern, Jane Wu. 2019. Foundations of entrepreneurial strategy. Strategic
Management Journal 40:5, 736-756. [Crossref]
203. Russell E. Browder, Howard E. Aldrich, Steven W. Bradley. 2019. The emergence of the maker
movement: Implications for entrepreneurship research. Journal of Business Venturing 34:3, 459-476.
[Crossref]
204. Eric Christian Brun. 2019. Understanding a Business Incubator as a Start-Up Factory: A Value Chain
Model Perspective. International Journal of Innovation and Technology Management 16:03. . [Crossref]
205. David W. Williams, Matthew S. Wood, J. Robert Mitchell, Diemo Urbig. 2019. Applying
experimental methods to advance entrepreneurship research: On the need for and publication of
experiments. Journal of Business Venturing 34:2, 215-223. [Crossref]
206. Abdel Malik Ola, Catherine Deffains-Crapsky, Régis Dumoulin. 2019. Vers une nouvelle approche de
l’investissement en amorçage : un raisonnement à travers la théorie de l’alignement structurel. Finance
Contrôle Stratégie :NS-5. . [Crossref]
207. Andrin Spescha. 2019. R&D expenditures and firm growth – is small beautiful?. Economics of
Innovation and New Technology 28:2, 156-179. [Crossref]
208. Arti Grover Goswami, Denis Medvedev, Ellen Olafsen. Searching for Winners 117-163. [Crossref]
209. Markus Grillitsch, Torben Schubert, Martin Srholec. 2019. Knowledge base combinations and firm
growth. Research Policy 48:1, 234-247. [Crossref]
210. Jermain Kaminski, Christian Hopp, Tereza Tykvová. 2019. New technology assessment in
entrepreneurial financing – Does crowdfunding predict venture capital investments?. Technological
Forecasting and Social Change 139, 287-302. [Crossref]
211. Alexander Styhre. Passion: The Motivation Behind Venture Work 105-144. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 284 of 801
212. Werner Bönte, Diemo Urbig. Connecting People and Knowledge: Knowledge Spillovers, Cognitive
Biases, and Entrepreneurship 385-397. [Crossref]
213. Jorge Melegati, Rafael Chanin, Xiaofeng Wang, Afonso Sales, Rafael Prikladnicki. Enablers and
Inhibitors of Experimentation in Early-Stage Software Startups 554-569. [Crossref]
214. Michael Fritsch. Entrepreneurship, Gründungen, Marktdynamik 5-19. [Crossref]
215. Mika Maliranta, Satu Nurmi. 2019. Business owners, employees, and firm performance. Small Business
Economics 52:1, 111-129. [Crossref]
216. David de Meza, Christopher Dawson, Andrew Henley, G. Reza Arabsheibani. 2019. Curb your
enthusiasm: Optimistic entrepreneurs earn less. European Economic Review 111, 53-69. [Crossref]
217. Godfrey Keller, Vladimír Novák, Tim Willems. 2019. A note on optimal experimentation under risk
aversion. Journal of Economic Theory 179, 476-487. [Crossref]
218. Raymond J. Jones, Anat Barnir. 2019. Properties of opportunity creation and discovery: Comparing
variation in contexts of innovativeness. Technovation 79, 1-10. [Crossref]
219. Yujin Kim, Chirantan Chatterjee, Matthew John Higgins. 2019. Moving Beyond the Valley of Death:
Regulation and Venture Capital Investments in Early-Stage Biopharmaceutical Firms. SSRN Electronic
Journal . [Crossref]
220. Kenneth A. Younge, Jeffrey M. Kuhn. 2019. First Movers and Follow-on Invention: Evidence from a
Vector Space Model of Invention. SSRN Electronic Journal . [Crossref]
221. Sabrina Howell, Ramana Nanda. 2019. Networking Frictions in Venture Capital, and the Gender Gap
in Entrepreneurship. SSRN Electronic Journal . [Crossref]
222. William R. Kerr, Frédéric Robert‐Nicoud. 2019. Tech Clusters. SSRN Electronic Journal . [Crossref]
223. Sanjay Goel, Raymond J. Jones, Ranjan Karri. Conceptualizing and Investigating Entrepreneurial
Action in Family Firms: A Few Promising Directions 873-907. [Crossref]
224. Simon Mayer. 2019. Financing Breakthroughs under Failure Risk. SSRN Electronic Journal 101. .
[Crossref]
225. Johannes Schneider, Christoph Wolf. 2019. Switching Forth and Back: Problem Solving under Time
Pressure. SSRN Electronic Journal 68. . [Crossref]
226. Tong Liu. 2019. Financing Experimentation. SSRN Electronic Journal 25. . [Crossref]
227. Raffaele Conti, Alfonso Gambardella, Elena Novelli. 2019. Specializing in Generality: Firm Strategies
When Intermediate Markets Work. Organization Science 30:1, 126-150. [Crossref]
228. Rustam Abuzov. 2019. The Impact of Venture Capital Screening. SSRN Electronic Journal 49. .
[Crossref]
229. Ana Paula Cusolito, William F. Maloney. Entry and Exit: Creating Experimental Societies 69-113.
[Crossref]
230. Gordon Burtch, Seth Carnahan, Brad N. Greenwood. 2018. Can You Gig It? An Empirical
Examination of the Gig Economy and Entrepreneurial Activity. Management Science 64:12,
5497-5520. [Crossref]
231. Matthew J. Kotchen, Christopher Costello. 2018. Maximizing the impact of climate finance: Funding
projects or pilot projects?. Journal of Environmental Economics and Management 92, 270-281.
[Crossref]
232. Irene Ng. 2018. Mimicking Firms: Future of Work and Theory of the Firm in a Digital Age. Journal
of Creating Value 4:2, 205-210. [Crossref]
233. Michael Fritsch, Sandra Kublina. 2018. Related variety, unrelated variety and regional growth: the
role of absorptive capacity and entrepreneurship. Regional Studies 52:10, 1360-1371. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 285 of 801
234. Gilles Duruflé, Thomas Hellmann, Karen Wilson. 2018. Catalysing entrepreneurship in and around
universities. Oxford Review of Economic Policy 34:4, 615-636. [Crossref]
235. Ufuk Akcigit, William R. Kerr. 2018. Growth through Heterogeneous Innovations. Journal of Political
Economy 126:4, 1374-1443. [Crossref]
236. Michael Ewens, Ramana Nanda, Matthew Rhodes-Kropf. 2018. Cost of experimentation and the
evolution of venture capital. Journal of Financial Economics 128:3, 422-442. [Crossref]
237. Alessandro Nuvolari, Pier Angelo Toninelli, Michelangelo Vasta. 2018. What makes a successful (and
famous) entrepreneur? Historical evidence from Italy (XIX-XX centuries). Industrial and Corporate
Change 27:3, 425-447. [Crossref]
238. Indraneel Chakraborty, Michael Ewens. 2018. Managing Performance Signals Through Delay:
Evidence from Venture Capital. Management Science 64:6, 2875-2900. [Crossref]
239. Philip T. Roundy, Mike Bradshaw, Beverly K. Brockman. 2018. The emergence of entrepreneurial
ecosystems: A complex adaptive systems approach. Journal of Business Research 86, 1-10. [Crossref]
240. J. Daniel Kim. 2018. Is there a startup wage premium? Evidence from MIT graduates. Research Policy
47:3, 637-649. [Crossref]
241. Juanita Gonzalez-Uribe, Michael Leatherbee. 2018. The Effects of Business Accelerators on Venture
Performance: Evidence from Start-Up Chile. The Review of Financial Studies 31:4, 1566-1603.
[Crossref]
242. Paige Clayton, Maryann Feldman, Nichola Lowe. 2018. Behind the Scenes: Intermediary
Organizations that Facilitate Science Commercialization Through Entrepreneurship. Academy of
Management Perspectives 32:1, 104-124. [Crossref]
243. Karen A. Murdock, Claus J. Varnes. 2018. Beyond effectuation. International Journal of Entrepreneurial
Behavior & Research 24:1, 256-272. [Crossref]
244. Renee Barnes, Margarietha Johanna de Villiers Scheepers. 2018. Tackling Uncertainty for Journalism
Graduates. Journalism Practice 12:1, 94-114. [Crossref]
245. Scott Morton Ninomiya, Sarah Burch. Beyond “The Business Case”: The Emerging Role of
Entrepreneurs in the Multilevel Governance of Urban Decarbonization in Canada 325-342. [Crossref]
246. Nancy Richter, Thomas Schildhauer. Startup Clinics: Applied Research and ‘First Aid’ for Early Stage
Startups 29-40. [Crossref]
247. Suting Hong, Konstantinos Serfes, Veikko Thiele. 2018. Competition in the Venture Capital Market
and the Success of Startup Companies: Theory and Evidence. SSRN Electronic Journal . [Crossref]
248. Xuan Tian, Jiajie Xu. 2018. Do Place-Based Programs Affect Local Innovation and Entrepreneurship?.
SSRN Electronic Journal . [Crossref]
249. Jillian Grennan, Roni Michaely. 2018. FinTechs and the Market for Financial Analysis. SSRN
Electronic Journal . [Crossref]
250. Andrea Contigiani, Trevor Young-Hyman. 2018. Evaluation of Early-Stage Ventures: Coherent
Combinations of Experimentation, Planning, and Structure. SSRN Electronic Journal . [Crossref]
251. Johan Hombert, Adrien Matray. 2018. The Long-Term Consequences of the Tech Bubble on Skilled
Workers' Earnings. SSRN Electronic Journal . [Crossref]
252. Gilles Durufll, Thomas F. Hellmann, Karen E. Wilson. 2018. Catalysing Entrepreneurship in and
Around Universities. SSRN Electronic Journal . [Crossref]
253. Sandeep Devanatha Pillai, Brent D. Goldfarb, David Kirsch. 2018. When Does Economic
Experimentation Matter? Finding the Pivot in the Early History of the Automobile Industry. SSRN
Electronic Journal . [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 286 of 801
254. Vladimmr Novvk, Tim Willems. 2018. A Note on Optimal Experimentation Under Risk Aversion.
SSRN Electronic Journal . [Crossref]
255. Wei Chen, Mingfeng Lin, Bryan Zheng Zhang. 2018. Lower Taxes, Smarter Crowd? The Impact of
Tax Incentives on Equity Crowdfunding. SSRN Electronic Journal . [Crossref]
256. Gary Dushnitsky, Diego Zunino. 2018. The Role of Crowdfunding in Entrepreneurial Finance. SSRN
Electronic Journal . [Crossref]
257. Andrea Contigiani. 2018. Experimentation, Learning, and Appropriability in Early-Stage Ventures.
SSRN Electronic Journal . [Crossref]
258. Marco Da Rin, Marina Di Giacomo, Alessandro Sembenelli. 2018. Corporate Income Taxation,
Leverage at Entry, and the Growth of Entrepreneurial Companies. SSRN Electronic Journal .
[Crossref]
259. Trond Døskeland, Per Stromberg. 2018. Evaluating investments in unlisted equity for the Norwegian
Government Pension Fund Global (GPFG). SSRN Electronic Journal 26. . [Crossref]
260. Galina Shirokova, Oleksiy Osiyevskyy, Michael H. Morris, Karina Bogatyreva. 2017. Expertise,
university infrastructure and approaches to new venture creation: assessing students who start
businesses. Entrepreneurship & Regional Development 29:9-10, 912-944. [Crossref]
261. Ramana Nanda, Matthew Rhodes-Kropf. Innovation Policies 37-80. [Crossref]
262. Suna Løwe Nielsen, Poul Rind Christensen, Astrid Heidemann Lassen, Mette Mikkelsen. 2017.
Hunting the Opportunity: The Promising Nexus of Design and Entrepreneurship. The Design Journal
20:5, 617-638. [Crossref]
263. Michael Fritsch. The Effect of New Business Formation on Regional Development 379-400.
[Crossref]
264. Pamela Adams, Victor Ricci. 2017. Trend Pie. The International Journal of Entrepreneurship and
Innovation 18:3, 210-215. [Crossref]
265. Alicia DeSantola, Ranjay Gulati. 2017. Scaling: Organizing and Growth in Entrepreneurial Ventures.
Academy of Management Annals 11:2, 640-668. [Crossref]
266. Sabrina T. Howell. 2017. Financing Innovation: Evidence from R&D Grants. American Economic
Review 107:4, 1136-1164. [Abstract] [View PDF article] [PDF with links]
267. M. Diane Burton, Tom Nicholas. 2017. Prizes, patents and the search for longitude. Explorations in
Economic History 64, 21-36. [Crossref]
268. SHAI BERNSTEIN, ARTHUR KORTEWEG, KEVIN LAWS. 2017. Attracting Early-Stage
Investors: Evidence from a Randomized Field Experiment. The Journal of Finance 72:2, 509-538.
[Crossref]
269. Marvin B. Lieberman, Gwendolyn K. Lee, Timothy B. Folta. 2017. Entry, exit, and the potential for
resource redeployment. Strategic Management Journal 38:3, 526-544. [Crossref]
270. James Caton. 2017. Entrepreneurship, search costs, and ecological rationality in an agent-based
economy. The Review of Austrian Economics 30:1, 107-130. [Crossref]
271. Dennis Lyth Frederiksen, Alexander Brem. 2017. How do entrepreneurs think they create value?
A scientific reflection of Eric Ries’ Lean Startup approach. International Entrepreneurship and
Management Journal 13:1, 169-189. [Crossref]
272. Matthias Dorner, Helmut Fryges, Kathrin Schopen. 2017. Wages in high-tech start-ups – Do
academic spin-offs pay a wage premium?. Research Policy 46:1, 1-18. [Crossref]
273. Swati Bhatt. The Three Trends: Granularity, Behemoths and Cooperation 29-55. [Crossref]
274. Kurt Dopfer, Jason Potts, Andreas Pyka. Upward and Downward Complementarity: The Meso Core
of Evolutionary Growth Theory 69-80. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 287 of 801
275. Marijn A. van Weele, Frank J. van Rijnsoever. Between a Soft Landing and a Hard Place: How Silicon
Valley Software and Life Sciences Business Incubators Facilitate Learning 167-201. [Crossref]
276. Ross Cullen. 2017. Evaluating renewable energy policies. Australian Journal of Agricultural and Resource
Economics 61:1, 1-18. [Crossref]
277. Gilles Chemla, Katrin Tinn. 2017. Learning Through Crowdfunding. SSRN Electronic Journal .
[Crossref]
278. Lin William Cong, Sabrina T. Howell, Ran Zhang. 2017. Does High-Growth Entrepreneurship Need
Public Markets? Evidence from Chinese IPOs. SSRN Electronic Journal 124. . [Crossref]
279. Ramana Nanda, Matthew Rhodes-Kropf. 2017. Coordination Frictions in Venture Capital Syndicates.
SSRN Electronic Journal . [Crossref]
280. Jing Gong, Yiping Song. 2017. Uber Might Buy Me a Mercedes Benz: An Empirical Investigation of
the Sharing Economy and Durable Goods Purchase. SSRN Electronic Journal 36. . [Crossref]
281. Buvaneshwaran Gokul Venugopal. 2017. Homophily, Information Asymmetry and Performance in
the Angels Market. SSRN Electronic Journal . [Crossref]
282. Kevin Boudreau. 2017. Amateurs: Low-Cost Development, Market Participation & Innovation on
Digital Platforms. SSRN Electronic Journal . [Crossref]
283. Ehsan Azarmsa, Lin William Cong. 2017. Insider Investor and Information. SSRN Electronic Journal
. [Crossref]
284. Onesun Steve Yoo, Tingliang Huang, Kenan Arifoglu. 2017. A Theoretical Analysis of the Lean
Startup's Product Development Process. SSRN Electronic Journal . [Crossref]
285. Katherine Lim. 2017. Self-Employment, Workplace Flexibility, and Maternal Labor Supply: A Life-
Cycle Model. SSRN Electronic Journal 98. . [Crossref]
286. Kijpokin Kasemsap. The Importance of Entrepreneurship in Global Business 92-115. [Crossref]
287. Tom Vanacker, Daniel P. Forbes. 2016. Disentangling the Multiple Effects of Affiliate Reputation on
Resource Attraction in New Firms. Organization Science 27:6, 1525-1547. [Crossref]
288. Katerina Nicolopoulou, Nada K. Kakabadse, Kanellos Panagiotis Nikolopoulos, Jose M. Alcaraz,
Konstantina Sakellariou. 2016. Cosmopolitanism and transnational elite entrepreneurial practices.
Society and Business Review 11:3, 257-275. [Crossref]
289. Kurt Dopfer, Jason Potts, Andreas Pyka. 2016. Upward and downward complementarity: the meso
core of evolutionary growth theory. Journal of Evolutionary Economics 26:4, 753-763. [Crossref]
290. Arup Varma, Nenad Jukic, Almir Pestek, Clifford J. Shultz, Svetlozar Nestorov. 2016. Airbnb: Exciting
innovation or passing fad?. Tourism Management Perspectives 20, 228-237. [Crossref]
291. Gerard Roe, Jason Potts. 2016. Detecting new industry emergence using government data: a new
analytic approach to regional innovation policy. Innovation 18:3, 373-388. [Crossref]
292. Gustavo Manso. 2016. Experimentation and the Returns to Entrepreneurship. The Review of Financial
Studies 29:9, 2319-2340. [Crossref]
293. Ethan Mollick, Ramana Nanda. 2016. Wisdom or Madness? Comparing Crowds with Expert
Evaluation in Funding the Arts. Management Science 62:6, 1533-1553. [Crossref]
294. Renato Gomes, Daniel Gottlieb, Lucas Maestri. 2016. Experimentation and project selection:
Screening and learning. Games and Economic Behavior 96, 145-169. [Crossref]
295. David B. Audretsch, Erik E. Lehmann, Stefano Paleari, Silvio Vismara. 2016. Entrepreneurial finance
and technology transfer. The Journal of Technology Transfer 41:1, 1-9. [Crossref]
296. Ramana Nanda, Matthew Rhodes-Kropf. 2016. Financing Entrepreneurial Experimentation.
Innovation Policy and the Economy 16, 1-23. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 288 of 801
297. Yael V. Hochberg. 2016. Accelerating Entrepreneurs and Ecosystems: The Seed Accelerator Model.
Innovation Policy and the Economy 16, 25-51. [Crossref]
298. Nathan Furr, Jackson A. Nickerson, Robert Wuebker. 2016. A Theory of Entrepreneuring. SSRN
Electronic Journal . [Crossref]
299. Ali Mohammadi, Kourosh Shafi. 2016. How Wise Are Crowd? A Comparative Study of Crowd and
Institutions in Peer-to-Business Online Lending Markets. SSRN Electronic Journal . [Crossref]
300. Jason Potts, Gerry Roe. 2016. Detecting New Industry Emergence Using Government Data: A New
Analytic Approach to Regional Innovation Policy. SSRN Electronic Journal . [Crossref]
301. Josh Lerner, James Tighe, Ann Leamon, Susana Garcia Robles. 2016. Impact of Early Stage Equity
Funds in Latin America. SSRN Electronic Journal . [Crossref]
302. Kevin Boudreau. 2016. Notes on Developing a Strategy and Designing a Company. SSRN Electronic
Journal . [Crossref]
303. Jermain Kaminski, Christian Hopp, Tereza Tykvova. 2016. New Technology Assessment in
Entrepreneurial Financing - Can Crowdfunding Predict Venture Capital Investments?. SSRN
Electronic Journal . [Crossref]
304. Joshua S. Gans, Scott Stern, Jane Wu. 2016. Foundations of Entrepreneurial Strategy. SSRN Electronic
Journal . [Crossref]
305. Yaroslav Rosokha, Kenneth A. Younge. 2016. Motivating Innovation: The Effect of Loss Aversion
on the Willingness to Persist. SSRN Electronic Journal 27. . [Crossref]
306. Ian Appel, Joan Farre-Mensa, Elena Simintzi. 2016. Patent Trolls and Small Business Employment.
SSRN Electronic Journal . [Crossref]
307. Yujin Kim, Chirantan Chatterjee. 2016. Improving Investor-Investee Matches with Regulation:
Evidence from the Orphan Drug Act & Global Biotechnology Industry. SSRN Electronic Journal .
[Crossref]
308. Michael Leatherbee, Riitta Katila. 2016. Stay the Course or Pivot? Antecedents of Cognitive
Refinements of Business Models in Young Firms. SSRN Electronic Journal 59. . [Crossref]
309. Michael Fritsch. Entrepreneurship, Gründungen, Marktdynamik 5-17. [Crossref]
310. William R. Kerr, Ramana Nanda. 2015. Financing Innovation. Annual Review of Financial Economics
7:1, 445-462. [Crossref]
311. Emily Cox Pahnke, Rory McDonald, Dan Wang, Benjamin Hallen. 2015. Exposed: Venture Capital,
Competitor Ties, and Entrepreneurial Innovation. Academy of Management Journal 58:5, 1334-1360.
[Crossref]
312. Rasananda Panda, Surbhi Verma, Bijal Mehta. 2015. Emergence and Acceptance of Sharing Economy
in India. International Journal of Online Marketing 5:3, 1-17. [Crossref]
313. Ramana Nanda, Matthew Rhodes-Kropf. 2015. Financing experiments. Science 348:6240, 1200-1200.
[Crossref]
314. Asif Islam. 2015. Entrepreneurship and the Allocation of Government Spending Under Imperfect
Markets. World Development 70, 108-121. [Crossref]
315. STOYAN TANEV, ERIK STAVNSAGER RASMUSSEN, ERIK ZIJDEMANS, ROY
LEMMINGER, LARS LIMKILDE SVENDSEN. 2015. LEAN AND GLOBAL TECHNOLOGY
START-UPS: LINKING THE TWO RESEARCH STREAMS. International Journal of Innovation
Management 19:03, 1540008. [Crossref]
316. Gerald Carlino, William R. Kerr. Agglomeration and Innovation 349-404. [Crossref]
317. János Kornai. 2015. Milyen is hát a tőke a 21. században?. Innováció Kelet-Közép-Európában 62:9,
909-942. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 289 of 801
318. Charles E. Eesley, Lynn Wu. 2015. Entrepreneurial Adaptation and Social Networks: Evidence from
a Randomized Experiment on a MOOC Platform. SSRN Electronic Journal . [Crossref]
319. Erin L. Scott, Pian Shu, Roman Lubynsky. 2015. Are 'Better' Ideas More Likely to Succeed? An
Empirical Analysis of Startup Evaluation. SSRN Electronic Journal . [Crossref]
320. Sharon Belenzon, Victor Manuel Bennett. 2015. Factor Substitutability and Entry. SSRN Electronic
Journal . [Crossref]
321. Karthik Krishnan, Lucas Lohmer, Pinshuo Wang. 2015. What Do Crowds Really Want?
Crowdfunding and Early Product Access. SSRN Electronic Journal . [Crossref]
322. Sabrina T Howell. 2015. Financing Constraints as Barriers to Innovation: Evidence from R&D Grants
to Energy Startups. SSRN Electronic Journal . [Crossref]
323. Timothy Terrell. 2015. Uncertainty, Asset Specificity, and Entrepreneurial Adaptation. SSRN
Electronic Journal . [Crossref]
324. Tania Babina, Paige Parker Ouimet, Rebecca Zarutskie. 2015. Going Entrepreneurial? IPOs and New
Firm Creation. SSRN Electronic Journal 124. . [Crossref]
325. Per L. Bylund, Robert Wuebker. 2014. Where Do Factor Markets Come from? Toward a Resource-
Based Theory of the Entrepreneurial Firm. SSRN Electronic Journal . [Crossref]
326. Shai Bernstein, Arthur G. Korteweg, Kevin Laws. 2014. Attracting Early Stage Investors: Evidence
from a Randomized Field Experiment. SSRN Electronic Journal 55. . [Crossref]
327. Gerald A. Carlino, William R. Kerr. 2014. Agglomeration and Innovation. SSRN Electronic Journal
3. . [Crossref]
328. William R. Kerr, Ramana Nanda. 2014. Financing Innovation. SSRN Electronic Journal . [Crossref]
329. Diego Comin, Ramana Nanda. 2014. Financial Development and Technology Diffusion. SSRN
Electronic Journal 91. . [Crossref]
330. Gustavo Manso. 2014. Experimentation and the Returns to Entrepreneurship. SSRN Electronic
Journal . [Crossref]
331. Ufuk Akcigit, William R. Kerr. 2010. Growth Through Heterogeneous Innovations. SSRN Electronic
Journal . [Crossref]
332. Kijpokin Kasemsap. The Importance of Entrepreneurship in Global Business 1989-2012. [Crossref]
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 290 of 801
Exhibit I
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 291 of 801
Consumer Finances
FEDS Notes
Twitter Share RSS
September 28, 2020
Disparities in Wealth by Race and Ethnicity in the 2019 Survey of
Consumer Finances
Neil Bhutta, Andrew C. Chang, Lisa J. Dettling, and Joanne W. Hsu with assistance from Julia
Hewitt1
New data from the 2019 Survey of Consumer Finances (SCF) show that long-standing and
substantial wealth disparities between families in different racial and ethnic groups were little
changed since the last survey in 2016; the typical White family has eight times the wealth of the
typical Black family and five times the wealth of the typical Hispanic family.
This FEDS Note explores patterns in wealth holding by race and ethnicity, as well as some key
issues related to the accumulation of wealth, using new data from the 2019 Survey of
Consumer Finances (SCF). We first analyze total wealth among families classified, according to
their self-identification during the interview, as White non-Hispanic, Black or African American
non-Hispanic, Hispanic or Latino, and other or multiple race (we will henceforth refer to these
groups as White, Black, Hispanic, and other, respectively).2 Wealth is defined as the difference
between families' gross assets and their liabilities.3 We will describe patterns at the median (the
typical household within each group) and at the mean (the average among households in each
group).
In the 2019 survey, White families have the highest level of both median and mean family
wealth: $188,200 and $983,400, respectively (Figure 1). Black and Hispanic families have
considerably less wealth than White families. Black families' median and mean wealth is less
than 15 percent that of White families, at $24,100 and $142,500, respectively. Hispanic families'
median and mean wealth is $36,100 and $165,500, respectively. Other families—a diverse
group that includes those identifying as Asian, American Indian, Alaska Native, Native Hawaiian,
Pacific Islander, other race, and all respondents reporting more than one racial identification—
have lower wealth than White families but higher wealth than Black and Hispanic families. The
same patterns of inequality in the distribution of wealth across all families are also evident within
race/ethnicity groups; for each of the four race/ethnicity groups, the mean is substantially higher
than the median, reflecting the concentration of wealth at the top of the wealth distribution for
each group.
Figure 1. White families have more wealth than Black, Hispanic, and other or multiple
race families in the 2019 SCF.
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 1/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 292 of 801
Consumer Finances
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Figures displays median (top panel) and mean (bottom panel) wealth by race and ethnicity, expressed in
thousands of 2019 dollars.
Accessible version
The SCF data provide a snapshot of families' wealth at a point in time. This point-in-time
observation is a result of many complex societal, governmental, and individual factors that play
out over the life cycle and even across generations. Among other factors, inter-generational
transfers, homeownership opportunities, access to tax-sheltered savings plans, and individuals'
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 2/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 293 of 801
Consumer Finances
savings and investment decisions contribute to wealth accumulation and families' financial
security.4 In the remainder of this note, we use the SCF to shed light on how these factors differ
by race and ethnicity and how patterns in wealth-holding have changed since the Great
Recession.5 Before we move on, we note that families were primarily interviewed for the 2019
SCF before the onset of the COVID-19 pandemic and associated changes to the economy.6
Therefore, we urge readers to exercise caution in making any inferences based on the patterns
described in this Note about how US families are faring in 2020.
Recent trends in wealth-holding
Between 2016 and 2019, median wealth rose for all race and ethnicity groups (Figure 2).
Growth rates for the 2016–19 period were faster for Black and Hispanic families, rising 33 and
65 percent, respectively, compared to White families, whose wealth rose 3 percent, and other
families, whose wealth rose 8 percent. That said, the faster growth in wealth for Black and
Hispanic families only resulted in modest changes in the gaps in wealth between these families
and White families. The White-Black gap in median wealth was little changed, from $163,700 in
2016 to $164,100 in 2019, and the White-Hispanic gap fell modestly from $160,000 in 2016 to
$152,100 in 2019.
Figure 2. Black and Hispanic families experienced faster growth in wealth over the last
two surveys after experiencing larger declines in wealth caused by the Great
Recession.
Source: Federal Reserve Board, Survey of Consumer Finances.
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 3/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 294 of 801
Consumer Finances
Notes: Figure displays percent changes in median wealth by race and ethnicity between the 2007 and 2019
Surveys. Survey years are displayed in order from left to right.
Accessible version
The patterns for the 2016-2019 period follow variation across groups in experiences in the
Great Recession (2007 to 2010), the immediate aftermath (2010 to 2013), and the continued
economic expansion (2013-2019). Median wealth fell about 30 percent for all groups during the
Great Recession. However, Black and Hispanic families' wealth continued to fall an additional
20 percent from 2010 to 2013, while White families' wealth was essentially unchanged, and
other families' wealth fell a more modest 10 percent. After 2013, median wealth rose for all
groups, with faster growth for Black, Hispanic, and other families.
Despite growth over the last two surveys, the typical White family and the typical Black family
have yet to recover to their pre-Great Recession levels of wealth. Over the entire 2007-2019
period, wealth fell by 11 percent for the typical White family and by 7 percent for the typical
Black family. Only the typical Hispanic family has seen an increase in wealth relative to before
the Great Recession, rising by about 39 percent, while the typical other family's wealth is about
unchanged since before the Great Recession.
While these cumulative changes in wealth from 2007-2019 may seem striking, there are two
important issues related to the interpretation of changes over time in the SCF from Figure 2,
particularly over the full time period.
First, the SCF interviews a different random sample of US families every three years. The same
family does not appear in consecutive SCFs. Therefore, the appropriate interpretation of the
survey-to-survey changes for a particular group, or cumulative changes over multiple surveys
for a particular group, is about the typical (or average) family within that group. The
inappropriate interpretation is changes for a specific family over time.
Second, the types and number of families that make up each race or ethnicity group change
over time as the underlying population of US families changes. Among other factors, population
aging, changes to immigration flows, and the evolution of self-identification patterns alter the
composition of each race or ethnicity group between surveys. For example, in the 2016 survey,
the other or multiple race group was composed of 50 percent reporting more than one racial
identification and 30 percent reporting Asian, whereas in 2019 these figures changed to 69
percent and 23 percent, respectively. Therefore, appropriately interpreting changes in a group's
wealth, especially over longer time periods, requires acknowledging compositional shifts within
each group. In particular, the robust growth in Hispanic wealth over the last two surveys and the
marked slowdown of growth for other families in 2019 are at least partially attributable to
compositional shifts in the types of families that make up these groups.
Wealth over the life-cycle
Wealth accumulation generally follows a predictable life-cycle arc, wherein families generally
accumulate wealth during their working years, in preparation for retirement. Table 1 displays
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 4/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 295 of 801
Consumer Finances
median wealth by age category based on the age of the reference person, separately, for White,
Black, Hispanic, and other families.7 Following the expected life-cycle savings patterns, within
each race or ethnicity group median wealth is sharply higher for middle-aged families (35 to 54)
compared to young families (under 35) and is highest among older families (55 and over).
Table 1: Wealth rises with age for all families, but substantial wealth gaps between
White and non-White families persist throughout the life-cycle.
White Black Hispanic Other
Under 35 25.4 0.6 11.2 13.5
35-54 185.0 40.1 46.1 154.5
Over 55 315.0 53.8 111.5 213.2
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Table displays median wealth by age group and by race and ethnicity in thousands of 2019 dollars.
Within each age group, the SCF data indicate large differences in wealth across racial and
ethnic groups. Even among young families who have had relatively little time to accumulate
wealth, there are sizeable differences in wealth by race and ethnicity, most starkly between
young Black and young White families. The median young Black family has almost no wealth
($600). In contrast, the median young White family has a wealth of $25,400. Young Hispanic
and other families fall in between, with $11,200 and $13,500 in median wealth, respectively.
Differences in parental resources may contribute to these early life cycle gaps, which we will
discuss in the next section.
In absolute terms, the gaps in median wealth between White and non-White families widen
considerably at older ages. For example, amongst families under 35, White families have
between $11,900 and $24,800 more in median wealth than Black, Hispanic, or other families.
For families over age 55, the gaps widen to between $101,700 and $261,100. In proportional
terms, however, the gaps are relatively stable or diminish with age. With respect to the Black-
White gap at middle and older ages, the median wealth of White families is four to six times
greater than the median wealth of Black families. These within-age ratios are somewhat lower
than the Black-White ratio of nearly eight for all families combined (implied by Figure 1).
Inheritances and other family support
Wealth-holding can differ across groups due to the intergenerational transmission of wealth.
There are numerous ways families can transmit wealth and resources across generations.
Families can directly transfer their wealth to the next generation in the form of a bequest. They
can also provide the next generation with inter vivos transfers (gifts), for example, providing
down payment support to enable a home purchase or a substantial wedding gift. By some
estimates bequests and transfers account for at least half of aggregate wealth (Gale and Scholz
1994), have recently averaged 3 percent of total household disposable personal income
(Feiveson and Sabelhaus 2018), and account for more of the racial wealth gap than any other
demographic or socioeconomic indicator (Hamilton and Darrity 2010).8 In addition to direct
transfers or gifts, families can make investments in their children that indirectly increase their
wealth. For example, families can invest in their children's educational success by paying for
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 5/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 296 of 801
Consumer Finances
college or private schools, which can in turn increase their children's ability to accumulate
wealth. For these reasons, wealth (or a lack thereof) can persist across generations and reflect,
among other factors, a legacy of discrimination or unequal treatment in housing, education, and
labor markets.9
One reason wealth-holding is relatively high among White families is they are considerably
more likely to have received an inheritance or gift. Another reason is White families report other
indicators associated with higher levels of family support (Table 2). For example, nearly 30
percent of White families report having received an inheritance or gift, compared to about 10
percent of Black families, 7 percent of Hispanic families, and 18 percent of other families.
Conditional upon receiving an inheritance or gift, White families also tend to receive larger
inheritances.
Table 2: White families are substantially more likely to receive inheritances, gifts and
other family support than Black and Hispanic families
White Black Hispanic Other
Recieved an Inheritance (Percent) 29.9 10.1 7.2 17.8
Conditional Median Inheritance (Thousands of 2019 dollars) 88.5 85.8 52.2 59.4
Expect an Inheritance (Percent) 17.1 6.0 4.2 14.7
Conditional Median Expected Inheritance (Thousands of 2019 dollars) 195.5 100.0 150.0 100.0
Could get $3,000 from family or friends (Percent) 71.9 40.9 57.8 63.4
Parent(s) have a College Degree (Percent) 34.4 24.8 15.2 40.0
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Table displays inheritances and gifts received, expected inheritances, and other indicators of family support, by
race and ethnicity, expressed in either Percent or Thousands of 2019 dollars. Parent(s) with a college degree refers to
the parents of the reference person.
Some families may not yet have received an inheritance (for example, if their parents are still
alive), but expect to receive one in the future. White families are both more likely to have
received an inheritance and are also more likely to expect to receive an inheritance: about 17
percent of White families expect an inheritance, compared to 6 percent of Black families, 4
percent of Hispanic families, and 15 percent of other families. Similarly, conditional upon
expecting to receive an inheritance in the future, White families expect to receive relatively
larger inheritances.
White and other families are more likely to report other indicators associated with higher levels
of family support. For example, White and other families are considerably more likely to report
being able to obtain $3,000 from a family member or friend in a financial emergency than Black
or Hispanic families. They are also more likely to have a parent with a college degree. Since
higher levels of education are associated with higher levels of wealth (see, for example, the
Bulletin article), this association suggests White and other families are likely to have wealthier
parents than Black or Hispanic families.
Homeownership
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 6/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 297 of 801
Consumer Finances
For many families, housing is the biggest component of wealth. The relationship between
housing and family wealth is complex. On the one hand, the ability to purchase a home is a
reflection of wealth a family already has (or their parents' wealth, as noted earlier), as significant
funds are generally required for a down payment and closing costs. On the other hand,
homeownership has also been found to yield strong financial returns on average and to be a
key channel through which families build wealth (Goodman and Mayer 2018).10
Life-cycle patterns of homeownership by age and by race and ethnicity are similar to the
patterns of wealth (Figure 3). Homeownership rises sharply from young to middle-age
regardless of race or ethnicity. At the same time, within each age group there are significant
gaps in homeownership between White and non-White families, with the biggest gaps between
White and Black families. Among young families, about 46 percent of White families own their
home, compared to just 17 percent of Black families. This gap may partially reflect differences in
parental wealth, as previous research has found that Black families are far less likely to receive
down payment assistance from their parents, delaying transitions into homeownership (Charles
and Hurst 2002).11
Figure 3. Homeownership rises with age regardless of race or ethnicity, though there
are significant differences in homeownership between White and non-White families
throughout the life-cycle.
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Figure displays homeownership rates by age group and by race and ethnicity. Race categories are
displayed in order from left to right.
Accessible version
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 7/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 298 of 801
Consumer Finances
This Black-White gap of nearly 30 percentage points narrows somewhat among middle-aged
and older families. For example, 73 percent of middle-aged White families own their home
compared to about 51 percent of middle-aged Black families. Meanwhile, the gap in the
homeownership rate between young White families and young Hispanic families is about 18
percentage points. In contrast to the White-Black gap, the White-Hispanic homeownership gap
expands somewhat among middle-aged and older families.
In addition to gaps in homeownership, there are also significant gaps in home values among
homeowners by race and ethnicity. For homeowners, the typical White families' home value is
$230,000 and the typical other families' home value is $310,000. The typical Black and Hispanic
families' home values are lower, at $150,000 and $200,000, respectively. Gaps in home values
are caused both by gaps in purchase prices and housing appreciation, which are a reflection of
a combination of factors including resource gaps (e.g., income and down payments), residential
segregation, and age of entry into homeownership.12
Retirement accounts and plan participation
Participation in retirement accounts and retirement plans is another important channel through
which families build wealth, and they provide financial security in retirement. These assets
include individual retirement accounts (IRAs), which typically are not dependent on a family's
employer, and two types of employer-sponsored plans: defined contribution plans (DC), which
are account-type job pensions such as 401(k)s, and traditional pensions (defined benefit plans,
DB). Assets held in IRA and DC account plans are subject to preferential tax treatment, and DB
plans guarantee a stream of income in retirement.
Ownership of IRA and DC retirement accounts rises among middle-aged families and then falls
among older families (Figure 4). In all age groups, Black and Hispanic families are far less likely
to have such retirement accounts. For example, among middle-aged families –who have the
highest rates of account ownership— 65 percent of White families have at least one retirement
account, compared to 44 percent of Black families, and just 28 percent of Hispanic families.
Figure 4. Retirement account ownership peaks at middle age, though ownership is less
likely for Black and Hispanic families at all ages.
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 8/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 299 of 801
Consumer Finances
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Figure displays the percent of families that own a retirement account (IRA or DC plan) with a net positive
balance by age group and by race and ethnicity. Race categories are displayed in order from left to right.
Accessible version
One reason for gaps in participation in retirement accounts is that not all families are eligible to
participate in an employer-sponsored retirement plan. Families may not be eligible for an
employer plan because their employer does not offer plans at all or they are offered but the
employee is not eligible (for example, because the employee works part time or the employee
has insufficient tenure at the company). Families who lack access to employer-sponsored plans
miss out on a common added benefit: many employers contribute to these plans, either by
matching some or all of the employee's contributions to the plan in the case of DC plans or by
providing employees a guaranteed income stream in retirement for DB plans.
Among working-age families (those under age 55), White families and – to a lesser extent other
families – have more widespread access to employer-sponsored retirement plans than Black or
Hispanic families (Figure 5, blue bars).13 The disproportionate access for White families relative
to Hispanic families is the most stark – for every three White families that can access an
employer-sponsored retirement plan only two Hispanic families have access.
Figure 5. White and other or multiple race families have broader access and
participate more in employer-sponsored retirement plans.
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200… 9/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 300 of 801
Consumer Finances
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Figures displays the percent of families with access to employer-sponsored retirement plans (DC or DB
plans, blue bars) and the percent of families that participate in an employer-sponsored retirement plan (orange
bars) among families under 55 years old, by race or ethnicity. Key identifies bar chart in order left to right.
Accessible version
Even for those families with access to an employer-sponsored plan, not all families opt to
participate in a plan. For working-age families, participation rates are lower than access rates
(Figure 5, orange bars). About 60 percent of White and 54 percent of other families participate
in a retirement plan, compared to 45 percent of Black families and 34 percent of Hispanic
families. The share of each group that participates among those with access –also known as
the "take-up" rate—also varies across groups. In addition to having higher rates of access, take-
up rates are higher amongst White and other families than Black or Hispanic families. While
about 90 percent of White and other families with access to a plan participate, about 80 percent
of Black families and about 75 percent of Hispanic families with access to a plan participate
(implied by Figure 5). These differences in participation may be caused by a variety of factors,
including whether or not a family has sufficient income to enable saving in this manner, the
types of funds offered by employer-sponsored plans, whether participation is by default or not,
and financial literacy.
IRA and DC plans can only provide financial security in retirement if families accrue sufficient
balances in the plans through investment contributions and subsequent returns on those
contributions. Among working-age families, balances in retirement accounts also vary by race
and ethnicity, and balances are larger amongst groups with higher rates of access and
participation. For working-age families that have balances in such accounts, the typical White
family has about $50,000 saved, which is two and a half times the amount saved as the typical
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-2020… 10/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 301 of 801
Consumer Finances
Black or Hispanic family, who have about $20,000 saved in retirement accounts. The median
balances for other families falls in between White and Black or Hispanic families, at about
$34,000. Difference in balances likely reflect a combination of factors including differences in
returns from the funds that contributions are invested in, differences families' lifetime
contributions to retirement accounts, and differences in employer matching to DC plans. For
example, the differences in access to employer-sponsored plans (Figure 5) imply that fewer
Black or Hispanic families are eligible for a plan with an employer match.
Overall, these gaps in retirement plan access, participation, and account balances suggest non-
White families will be less financially secure in retirement than White families. But our
discussion ignores Social Security benefits and the net present value of DB plans, which are
key components of many families' retirement planning. These benefits are hard to account for
because they involve assumptions about families' future earnings and years of work. Still,
research suggests that accounting for these benefits can reduce overall inequality in retirement
resources (Bricker, Goodman, Moore, and Volz, 2020; Sabelhaus and Volz, forthcoming).14
Emergency Savings
Families often aim to build up easily-accessible savings to help deal with unexpected expenses
and disruptions to their income. Although interest rates on highly-liquid transaction accounts are
generally quite low and, as a result, these highly-liquid accounts are less important for long-term
wealth-building than higher-return assets like housing or retirement accounts, savings can help
families avoid costly borrowing or missed payments when unexpected events arise.
Nearly all families have some type of highly-liquid asset, such as a checking account, savings
account, or pre-paid card (Table 3). Conditional upon having a liquid asset, however, the typical
White family has considerably more liquid savings than the typical Black, Hispanic, or other
family. While the typical Black or Hispanic family has $2,000 or less in liquid savings, the typical
White family has more than four times that amount. Other families fall somewhere in the middle,
with the typical family holding $5,000 in liquid savings.
Table 3: White families have more emergency savings than Black or Hispanic
families.
White Black Hispanic Other
Has Liquid Assets (Percent) 98.8 96.8 95.5 98.8
Conditional Mean Liquid Assets (Thousands of 2019 dollars) 8.1 1.5 2.0 5.0
Has Direct or Indirect Equity (Percent) 60.8 33.5 24.2 53.8
Conditional Median of Equities (Thousands of 2019 dollars) 50.6 14.4 14.9 28.8
Source: Federal Reserve Board, 2019 Survey of Consumer Finances.
Notes: Liquid assets and equities by race and ethnicity, expressed in either Percent or Thousands of 2019 dollars.
Families may also choose to tap into higher-return assets in a financial emergency, such as
money stored in directly or indirectly-held equities. These assets include directly-held stocks
and mutual funds, which can be sold, and equities held in quasi-liquid accounts (such as
retirement accounts), which can be liquidated or borrowed against if a financial emergency were
to occur.
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-2020… 11/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 302 of 801
Consumer Finances
There are substantial disparities in ownership of equities across families grouped by their race
and ethnicity. While more than half of White and other families have equities, just over 24
percent of Hispanic families and just under 34 percent of Black families have any equities.
Conditional on having equities, there are also substantial gaps in amounts held. For example,
the typical White family has $50,600 in equities they could tap into in an emergency, compared
to just $14,400 for the typical Black family and $14,900 for the typical Hispanic family.
These gaps in emergency savings are particularly relevant in light of the COVID-19 pandemic
and associated job losses. Because the 2019 SCF data were collected just before the onset of
the pandemic, these gaps in savings suggest large disparities in families' ability to weather the
pandemic. Indeed, Bhutta, Blair, Dettling, and Moore (2020) find that without the substantial
cash assistance included in the Coronavirus Aid, Relief, and Economic Security (CARES) Act,
there would be large disparities by race and ethnicity in the share of families who could cover
their normal, recurring expenses if they were to lose their job for six months or more.15 They
find that just 10 percent of Hispanic families and 14 percent of Black families have enough
savings to cover six months of expenses, compared to 36 percent of White families and 27
percent of other families. But with the cash assistance in the CARES Act (i.e., unemployment
insurance and direct stimulus payments), over 90 percent of all family-groups could cover their
expenses for six months. Still, much of the CARES Act has expired or will expire in the coming
months. If job losses persist, then unequal levels of savings could lead to disparities by race or
ethnicity in financial distress during the pandemic. Furthermore, growing evidence suggests
Black, Hispanic, and Native American communities are being disproportionally affected by the
health and economic effects of COVID-19.16
1. Division of Research and Statistics, Board of Governors of the Federal Reserve System. The analysis and
conclusions set forth are those of the authors and do not reflect the views of the Board of Governors or the Federal
Reserve staff. Bhutta: neil.bhutta@frb.gov, Chang: andrew.c.chang@frb.gov, ORCID 0000-0002-9769-789X, Dettling:
lisa.j.dettling@frb.gov, Hsu: joanne.w.hsu@frb.gov, ORCID 0000-0002-0715-6230. Return to text
2. For more on the race and ethnicity classifications used in this FEDS Note, see the appendix to Neil Bhutta, Jesse
Bricker, Andrew C. Chang, Lisa J. Dettling, Sarena Goodman, Joanne W. Hsu, Kevin B. Moore, Sarah Reber, Alice
Henriques Volz, and Richard A. Windle. 2020. "Changes in U.S. Family Finances from 2016 to 2019: Evidence from
the Survey of Consumer Finances (PDF)," Federal Reserve Bulletin 106(5) (henceforth, the Bulletin article). The other
or multiple race group consists of a very racially/ethnically diverse set of families, including those identifying as Asian,
American Indian, Alaska Native, Native Hawaiian, Pacific Islander, other race, and all respondents reporting more than
one racial identification. Because of small sample sizes, we do not have statistical power to further disaggregate this
group of families. In 2019, families reporting more than one racial identification were the largest subgroup of the other
or multiple race group (about 69 percent of families), followed by Asian families (about 23 percent of families).
Because of the varied composition of the other group and changes in its composition over time, readers should
exercise caution when making inferences. Return to text
3. Our concept of wealth is equivalent to the net worth concept defined in the Bulletin article. See the appendix to the
Bulletin article for more details on components of wealth or net worth. Except for received inheritances, dollar values
are adjusted to 2019 dollars using the "current methods" version of the consumer price index for all urban consumers
(CPI-U-RS), which is available since 1977. Received inheritances are adjusted to 2019 dollars using the consumer
price index for all urban consumers (CPI-U) to account for inheritances received prior to 1977. Return to text
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-2020… 12/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 303 of 801
Consumer Finances
4. For a recent discussion of the Black-White wealth gap, see Kriston McIntosh, Emily Moss, Ryan Nunn, and Jay
Shambaugh. 2020. "Examining the Black-white wealth gap. " Brookings Institution, February 27, 2020. Return to
text
5. For more information on how other components of wealth and socio-demographic characteristics vary by race and
ethnicity –which were, in general, little changed since the 2016 survey—see Lisa J. Dettling, Joanne W. Hsu, Lindsay
Jacobs, Kevin B. Moore, and Jeffrey P. Thompson. 2017. "Recent Trends in Wealth-Holding by Race and Ethnicity:
Evidence from the Survey of Consumer Finances ," FEDS Notes. Washington: Board of Governors of the Federal
Reserve System, September 27, 2017. Return to text
6. See the appendix to the Bulletin article for more information on the timing of interviews. Return to text
7. See the appendix to the Bulletin article for the definition of the reference person. Return to text
8. William Gale and John Karl Scholz. 1994. "Intergenerational Transfers and the Accumulation of Wealth. " Journal
of Economic Perspectives, 8(4): 145-160. Laura Feiveson and John Sabelhaus. 2018. "How Does Intergenerational
Wealth Transmission Affect Wealth Concentration? " FEDS Notes. Washington: Board of Governors of the Federal
Reserve System, June 1, 2018. Darrick Hamilton and William Darity. 2010. "Can 'Baby Bonds' Eliminate the Racial
Wealth Gap in Putative Post-Racial America? " The Review of Black Political Economy 37(3–4): 207–16. Return to
text
9. For a more detailed discussion of discrimination and institutional barriers in housing markets, see, for example, The
Color of Law (2017) by Richard Rothstein or Race for Profit (2019) by Keeanga-Yamahtta Taylor. For more on
inequities in educational opportunities, see, for example, Ivy Morgan and Ary Amerikaner (2018) "Funding Gaps: An
Analysis of School Funding Equity Across the U.S. and within Each State " or Children of the Dream (2019) by
Rucker C. Johnson. For more on labor market outcomes, see, for example, Devah Pager (2007) "The Use of Field
Experiments for Studies of Employment Discrimination: Contributions, Critiques, and Directions for the Future "
Annals of the American Academy of Political and Social Sciences 609 (January): 104-133, or Raj Chetty, Nathaniel
Hendren, Maggie R. Jones, and Sonya R. Porter (2020) "Race and Economic Opportunity in the United States: an
Intergenerational Perspective " The Quarterly Journal of Economics 135(2); 711–783. Return to text
10. Laurie S. Goodman and Christopher Mayer. 2018. "Homeownership and the American Dream. " Journal of
Economic Perspectives 32(1): 31-58. Return to text
11. Kerwin Kofi Charles and Erik Hurst. 2002. "The Transition to Home Ownership and the Black-White Wealth Gap.
" The Review of Economics and Statistics 84(2): 281–297. Return to text
12. A previous FEDS Note on this topic has additional information on income and other resource gaps by race and
ethnicity (Dettling, Hsu, Jacobs, Moore, and Thompson 2017). The Color of Law (2017) by Richard Rothstein provides
a detailed discussion on the lasting effects of residential segregation. Return to text
13. We count families as having access to an employer-sponsored plan if either the reference person or the
spouse/partner of the reference person is eligible to participate in a DC or a DB plan provided by a current employer
plus all families that participate in DC or a DB plan, which are those families that have a retirement account with a
balance from a previous employer, have the rights to a future pension from a previous employer, or are self-employed
at a business that offers a pension, retirement, or tax-deferred savings plan. Return to text
14. Jesse Bricker, Sarena Goodman, Kevin Moore, and Alice Henriques Volz. 2020. "Wealth and Income
Concentration in the SCF: 1989-2019," FEDS Notes. John Sabelhaus and Alice Henriques Volz (forthcoming). "Social
Security Wealth, Inequality, and Lifecycle Saving," in Measuring and Understanding the Distribution and Intra/Inter-
Generational Mobility of Income and Wealth. NBER Book Series Studies in Income and Wealth. The University of
Chicago Press. Return to text
15. Neil Bhutta, Jacqueline Blair, Lisa Dettling, and Kevin Moore. 2020. "COVID-19, the CARES Act, and Families'
Financial Security " National Tax Journal 73(3): 645-672. Return to text
16. For health effects, see, for example, "The COVID Racial Data Tracker ". For economic effects, see, for
example, Steven Brown (2020) "How COVID-19 is Affecting Black and Latino Families' Employment and Financial
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-2020… 13/14
8/29/23, 5:13 PM Case 1:23-cv-03424-TWT Document
The Fed - Disparities in 59-4and Filed
Wealth by Race 08/31/23
Ethnicity in the Page
2019 Survey of 304 of 801
Consumer Finances
Well-Being " Urban Institute, May 6, 2020. Return to text
Please cite this note as:
Bhutta, Neil, Andrew C. Chang, Lisa J. Dettling, and Joanne W. Hsu (2020). "Disparities in
Wealth by Race and Ethnicity in the 2019 Survey of Consumer Finances," FEDS Notes.
Washington: Board of Governors of the Federal Reserve System, September 28, 2020,
https://doi.org/10.17016/2380-7172.2797.
Disclaimer: FEDS Notes are articles in which Board staff offer their own views and present analysis on a range of
topics in economics and finance. These articles are shorter and less technically oriented than FEDS Working Papers
and IFDP papers.
Last Update: September 28, 2020
https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-2020… 14/14
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 305 of 801
Exhibit J
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 306 of 801
Income in the United States: 2021
Current Population Reports
by Jessica Semega and Melissa Kollar
Issued September 2022
P60-276
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 307 of 801
Acknowledgments The Social, Economic, and Housing Statistics Division of the
U.S. Census Bureau recognizes Trudi J. Renwick for her 14 years
of service with the Census Bureau. Dr. Renwick retired in 2022.
From her dissertation that developed a methodology for basic
family budgets to her research on the development of the
Supplemental Poverty Measure, she spent her career dedicated
to improving the measurement of poverty and the well-being of
low-income families. While at the Census Bureau, Dr. Renwick pro-
vided leadership to the income, poverty, and program participation
branches. Her dedication to her staff, subject matter expertise, and
overall guidance were appreciated and will be greatly missed.
Jessica Semega and Melissa Kollar prepared this report, Daniel C. Lin
prepared the tax model and provided content for Appendix C, and
Matthew Unrath prepared Appendix D, all of the Income Statistics
Branch. Jonathan Spader, Assistant Division Chief for Housing
Characteristics in the Social, Economic, and Housing Statistics Division,
reviewed and provided overall guidance.
Mallory Bane and Susan S. Gajewski, under the supervision of David
Watt, all of the Demographic Systems Division, and Lisa Cheok of the
Demographic Programs Directorate—Survey Operations, processed the
Current Population Survey, 2022 Annual Social and Economic Supplement
(CPS ASEC) file.
Adam W. Reilly, with the assistance of Kirk E. Davis, Raymond E.
Dowdy, Lan N. Huynh, and Chandararith R. Phe, programmed and pro-
duced the historical, detailed, and publication tables under the direc-
tion of Hung X. Pham, Chief of the Tabulation and Applications Branch,
Demographic Surveys Division.
Emily Hood, under the supervision of David V. Hornick, of the
Demographic Statistical Methods Division, conducted statistical review.
Lisa Cheok, assisted by Roberto Cases, of the Demographic Programs
Directorate—Survey Operations, provided overall direction for the survey
implementation. Charlie Carter, Agatha Jung, and Johanna Rupp, of
the Application Development and Services Division, prepared and pro-
grammed the computer-assisted interviewing instrument used to conduct
the CPS ASEC.
Additional people within the U.S. Census Bureau also made significant
contributions to the preparation of the report. Liana E. Fox, Charles
Hokayem, Matthew Marlay, Lindsay Monte, Rachel Shattuck, and Bruce
Webster Jr. reviewed the contents.
Stacey Barber, Faye E. Brock, and Steven Brown provided publication
management, editorial review, and 508 compliancy for print and elec-
tronic media. Paula Lancaster and Stephen Gibson provided graphic
design and composition, all under the direction of Corey Beasley, Chief
of the Graphic and Editorial Services Branch, Public Information Office.
The authors would like to also thank the Census Bureau field represen-
tatives and telephone interviewers who conducted the interviews that
provide the data in this report. Without their dedication, the preparation
of this report or any report from the Current Population Survey would be
impossible.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 308 of 801
Income in the United States: 2021 Issued September 2022
P60-276
U.S. Census Bureau
Robert L. Santos,
Director
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 309 of 801
Suggested Citation
Jessica Semega and Melissa Kollar,
U.S. Census Bureau,
Current Population Reports,
P60-276,
Income in the United States: 2021,
U.S. Government Publishing Office,
Washington, DC,
September 2022.
U.S. CENSUS BUREAU
Robert L. Santos,
Director
Ron S. Jarmin,
Deputy Director and Chief Operating Officer
Victoria A. Velkoff,
Associate Director for Demographic Programs
David G. Waddington,
Chief, Social, Economic, and Housing Statistics Division
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 310 of 801
Contents TEXT
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
HOUSEHOLD INCOME BY SELECTED CHARACTERISTICS. . . . . . . . . . . 2
All Households. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Type of Household . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Race and Hispanic Origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Age of Householder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Nativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Residence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Educational Attainment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
INCOME INEQUALITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Money Income Inequality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Equivalence-Adjusted Income Inequality. . . . . . . . . . . . . . . . . . . . . . . . . . . 7
EARNINGS AND WORK STATUS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Total and Full-Time, Year-Round Workers. . . . . . . . . . . . . . . . . . . . . . . . . . 7
Workers by Sex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
SUMMARY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
FIGURES
Figure 1. Median Household Income and Percent Change by
Selected Characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Figure 2. Real Median Household Income by Race and Hispanic
Origin: 1967 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Figure 3. Income Distribution Measures and Percent Change
Using Money Income and Equivalence-Adjusted
Income. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Figure 4. Median Earnings and Percent Change by Work Status
and Sex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Figure 5. Total and Full-Time, Year-Round Workers 15 Years and
Older With Earnings by Sex: 1967 to 2021. . . . . . . . . . . . 9
Figure 6. Female-to-Male Earnings Ratio and Median Earnings of
Full-Time, Year-Round Workers 15 Years and Older
by Sex: 1960 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Figure D-1. Historical Median Income Using Alternative Price
Indexes: 1967 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Figure D-2. Real Year-Over-Year Income Growth Using Alternative
Price Indexes: 2016 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . 51
U.S. Census Bureau Income in the United States: 2021 iii
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 311 of 801
APPENDIXES
Appendix A. Estimates of Income. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
How Income Is Measured. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Business Cycles—Recessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Cost-of-Living Adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Appendix B. Effects of 2020 Census-Based Population Controls on 2020 Income Estimates. . . . . . . . . . . . . . . 39
Effects on Income and Earnings Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Appendix C. Post-Tax Household Income. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Appendix D. Historical Income Alternative Inflation Series and Request for Comments. . . . . . . . . . . . . . . . . . . 49
Alternative Price Indexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Implications for Income Estimates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Request for Comments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Appendix E. Additional Information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Source and Accuracy of the Estimates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Accessing Income Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Other Sources of Income Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Questions and Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
APPENDIX TABLES
Table A-1. Income Summary Measures by Selected Characteristics: 2020 and 2021 . . . . . . . . . . . . . . . . . . . . . 15
Table A-2. Households by Total Money Income, Race, and Hispanic Origin of Householder: 1967 to 2021. . 16
Table A-3. Income Distribution Measures Using Money Income and Equivalence-Adjusted Income:
2020 and 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Table A-4a. Selected Measures of Household Income Dispersion: 1967 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . 30
Table A-4b. Selected Measures of Household Income Dispersion: 1967 to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . 31
Table A-5. Selected Measures of Equivalence-Adjusted Income Dispersion: 1967 to 2021. . . . . . . . . . . . . . . . 33
Table A-6. Earnings Summary Measures by Selected Characteristics: 2020 and 2021. . . . . . . . . . . . . . . . . . . . 36
Table A-7. Number and Real Median Earnings of Total Workers and Full-Time, Year-Round Workers
With Earnings by Sex and Female-to-Male Earnings Ratio: 1960 to 2021. . . . . . . . . . . . . . . . . . . 37
Table B-1. Income Summary Measures by Selected Characteristics: 2020 Estimates Using 2010
Census-Based Population Controls and 2020 Census-Based Population Controls . . . . . . . . . . . 40
Table B-2. Earnings Summary Measures by Selected Characteristics: 2020 Estimates Using 2010
Census-Based Population Controls and 2020 Census-Based Population Controls . . . . . . . . . . . 41
Table C-1. Post-Tax Household Income Summary Measures by Selected Characteristics: 2020 and 2021. . . 45
Table C-2. Summary Measures by Selected Characteristics Using Money Income and Post-Tax
Income: 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Table C-3. Distribution Measures Using Post-Tax Income and Equivalence-Adjusted Post-Tax
Income: 2020 and 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Table C-4. Distribution Measures Using Money Income, Post-Tax Income, Equivalence-Adjusted Income,
and Equivalence-Adjusted Post-Tax Income: 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Table D-1. Historical Median Income Using Alternative Price Indexes: 1967 to 2021 . . . . . . . . . . . . . . . . . . . . . 53
iv Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 312 of 801
Income in the United States: 2021
INTRODUCTION and does not account for changes by select characteristics of the
in income or inflation that have householder, followed by a section
The U.S. Census Bureau collects
occurred more recently in 2022. 3 on income inequality and one on
data and publishes estimates on
median earnings and work status.4
income, earnings, and inequality In 2021, Congress passed the
in order to evaluate national eco- American Rescue Plan Act (ARPA) This report is released along-
nomic trends and to understand in response to the COVID-19 pan- side two other reports focused
their effect on the well-being of demic. ARPA provided additional on poverty estimates and health
households and individuals. income to families through a third insurance coverage in the United
round of stimulus payments and States. For poverty and health
This report presents estimates
expansions to the Child Tax Credit, insurance estimates, refer to
on income in the United States
Earned Income Tax Credit, and the “Poverty in the United States:
for calendar year 2021, based
Child and Dependent Care Credit. 2021” and “Health Insurance
on information collected in
The income estimates in the main Coverage in the United States:
the 2022 and earlier Current
sections of this report are based 2021.”5
Population Survey Annual Social
on the concept of money income,
and Economic Supplements Highlights
as measured by the CPS ASEC. It
(CPS ASEC) conducted by the
includes all income received by • Real median household income
Census Bureau.* Estimates for
each person in the household who was $70,784 in 2021, not statis-
2020 in this report will not match
is aged 15 and older, excluding tically different from the 2020
those published last year due to
certain receipts such as capital estimate of $71,186 (Figure 1
the implementation of the 2020
gains. Money income is pretax and Table A-1).
Census-based population con-
and does not include stimulus
trols. Appendix B provides details. • Based on the money income
payments and tax credits such
To adjust for changes in the cost Gini index, income inequal-
as those from ARPA. Appendix
of living over time, historical ity increased by 1.2 percent
A provides a detailed explana-
income estimates in this report are between 2020 and 2021; this
tion of how income is measured
expressed in real or 2021 dollars.1 represents the first time the
using the CPS ASEC. For post-tax
This inflation adjustment is based Gini index has shown an annual
household income estimates that
on the Consumer Price Index for increase since 2011 (Figure 3
include stimulus payments and tax
all Urban Consumers Retroactive and Table A-3).
credits, refer to Appendix C.
Series (R-CPI-U-RS) for 2021 and • Between 2020 and 2021, the
earlier years, which measured a The continued response to the change in the number of total
4.7 percent increase in consumer COVID-19 pandemic, rising infla- workers was not statistically
prices between 2020 and 2021. 2 tion, shifts in worker composition, significant; however, there was
This is the largest annual increase and other macroeconomic condi- an increase of about 11.1 million
in the cost-of-living adjustment tions also shaped the experiences full-time, year-round workers
since 1990. It is important to note of households in 2021. The pur- (from approximately 106.3 mil-
that this report covers income pose of this report is to present lion to 117.4 million), suggesting
estimates for 2021 and prior years estimates of median household a shift from working part-time
income, income inequality, worker or part-year in 2020 to full-
* The Census Bureau reviewed this data earnings, and related measures for time, year-round work in 2021
product for unauthorized disclosure of 2021 based on data from the CPS (Table A-6).
confidential information and approved the
disclosure avoidance practices applied to ASEC. • The real median earnings of
this release CBDRB-FY22-357. All compara-
tive statements have undergone statisti- This report begins with a section all workers (including part-
cal testing and are statistically significant time and full-time workers)
at the 90 percent confidence level unless on median household income
otherwise noted. increased 4.6 percent between
U.S. Census Bureau Income in the United States: 2021 1
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 313 of 801
2020 and 2021, while median All Households from 2020 (Figure 2 and Table
earnings of those who worked A-1).12 Among the race groups,
Real median household income
full-time, year-round decreased Asian households had the high-
was $70,784 in 2021. This estimate
4.1 percent (Figure 4). est median income ($101,418) in
is not statistically different from
More information on these topics 2021, followed by non-Hispanic
the 2020 estimate of $71,186 and
can be found in the relevant sec- Whites ($77,999) and Hispanics
2.8 percent lower than the 2019
tions of this report. ($57,981).13 Black households
median, the year before the most
had the lowest median income
recent recession (Figure 1 and
HOUSEHOLD INCOME BY Table A-1).7 Household income in
($48,297).
SELECTED CHARACTERISTICS 2019 was the highest since 1967, The real median incomes of dif-
This section focuses on real even after adjusting for the effect ferent groups can be compared
median household income by of the CPS ASEC survey redesign, by calculating the ratio of the
selected characteristics of the subsequent processing changes, median income of a specific
householder such as race and and nonresponse bias (Figure 2 group to the median income of
Hispanic origin, nativity, region, and Table A-2). 8 non-Hispanic White households.
and education. The householder is For 2021, the ratio of Asian to
the person (or one of the people) Type of Household9
non-Hispanic White household
in whose name the home is owned The 2021 real median income of income was 1.30. In other words,
or rented and the person to whom family households was not statisti- the median Asian household had
the relationship of other house- cally different from the 2020 esti- a household income 1.30 times
hold members is recorded. Each mate, while nonfamily households greater than that of the median
household has one householder, experienced a 1.9 percent decline non-Hispanic White household.
and those in group quarters are over the same period (Figure The ratio of Black to non-Hispanic
excluded from the household 1 and Table A-1). Among fam- White household income was
population.6 ily households, married couples 0.62, while the ratio of Hispanic
had the highest median income to non-Hispanic White household
For most demographic character-
in 2021 ($106,921), followed by income was 0.74. None of these
istics of the householder shown
those maintained by men with no ratios were statistically different
in Figure 1, the 2021 real median
spouse present ($70,525). Family from 2020.
household income estimates were
households maintained by women
not statistically different from the
with no spouse present had the Age of Householder
2020 estimates. Between 2020
lowest median income ($51,168). Real median income in 2021 for
and 2021, declines in median
household income were experi- householders under the age of
Looking at nonfamily households,
enced by nonfamily households, 65 ($80,734) was not statistically
real median household income for
those with a householder aged 65 different from the 2020 median.
female householders decreased
and older, those maintained by However, median income for
4.7 percent between 2020 and
a native-born person, and those householders aged 65 and over
2021, while the change for male
with a householder with some declined 2.6 percent between
householders was not statistically
college. The only demographic 2020 and 2021 (Figure 1). Table
significant.10
group to experience an increase A-1 provides estimates for a more
in median household income Race and Hispanic Origin11 detailed set of age categories.
between 2020 and 2021 was Among the age categories, house-
Real median household incomes
householders with at least a bach- holders aged 15 to 24 and those
in 2021 for non-Hispanic Whites,
elor’s degree. More details are 45 to 54 experienced increases
Blacks, Asians, and Hispanics
available in the sections below. of 5.2 percent and 2.6 percent,
were not statistically different
respectively, in their median
household incomes.14
2 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 314 of 801
Figure 1.
Median Household Income and Percent Change by Selected Characteristics
(Households as of March of the following year)
2021 Median Income Change: 2020 to 2021
ALL HOUSEHOLDS $70,784 -0.6
Type of Household
Family households $91,162 0.5
Nonfamily households
$41,797 -1.9
Race and Hispanic
Origin of Householder
White $74,262 -1.0
White, not Hispanic $77,999 -1.2
Black 0.3
$48,297
Asian $101,418 1.8
Hispanic (any race)
$57,981 -0.1
Age of Householder
Under 65 years $80,734 0.3
65 years and older $47,620 -2.6
Nativity of Householder
Native-born $71,522 -1.4
Foreign-born $66,043 1.5
Region
Northeast $77,472 -2.0
Midwest $71,129 0.9
South $63,368 -1.5
West $79,430 0.9
Metropolitan Statistical
Area (MSA) Status
Inside MSA $73,823 -1.1
Inside principal cities $64,839 -1.2
Outside principal cities $79,599 -0.5
Outside MSA $53,750 -1.0
Educational Attainment
of Householder1
No high school diploma $30,378 -2.4
High school, no college $50,401 0.9
Some college $64,378 -4.0
Bachelor’s degree or higher $115,456 2.7
Denotes a statistically
significant change
1
Householders aged 25 and older. In 2021, the median household income for this group was $72,046.
Note: Statistically significant indicates the change is statistically different from zero at the 90 percent confidence level. Margins
of error and other related estimates and notes are available in Table A-1. Information on confidentiality protection, sampling error,
nonsampling error, and definitions is available at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 2021 and 2022 Annual Social and Economic Supplements (CPS ASEC).
Householders aged 45 to 54 25 to 34 ($74,862), and 15 to 24 Nativity16
($97,089) had the highest median ($51,645).15 Householders aged 65
Between 2020 and 2021, real
income in 2021, followed by and over ($47,620) had the lowest
median income of households
householders aged 35 to 44 median incomes.
maintained by a native-born
($90,312), 55 to 64 ($75,842),
person declined 1.4 percent, while
U.S. Census Bureau Income in the United States: 2021 3
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 315 of 801
Figure 2.
Real Median Household Income by Race and Hispanic Origin: 1967 to 2021
(Households as of March of the following year)
2021 R-CPI-U-RS adjusted dollars Recession
110,000
100,000 $101,418
90,000
80,000
Asian $77,999
70,000 $70,784
60,000 White, not Hispanic
$57,981
All races
50,000 $48,297
Hispanic (any race)
40,000
Black
30,000
20,000
10,000
0
1959 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2021
Note: The data for 2017 and beyond reflect the implementation of an updated processing system. The data for
2013 and beyond reflect the implementation of the redesigned income questions. Refer to Table A-2 for historical
race footnotes. The data points are placed at the midpoints of the respective years. Median household income data
are not available prior to 1967. More information on the R-CPI-U-RS dollar adjustment and recessions is available in
Appendix A. Information on confidentiality protection, sampling error, nonsampling error, and definitions is available
at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 1968 to 2022 Annual Social and Economic Supplements
(CPS ASEC).
the median income of households Region18 income ($53,750). The 2021 real
maintained by a foreign-born median incomes of households for
Median incomes were highest
person was not statistically differ- all categories of MSAs available in
in the West ($79,430) and the
ent from 2020 (Figure 1 and Table Table A-1 were not statistically dif-
Northeast ($77,472), followed
A-1). The foreign-born can be clas- ferent from their respective 2020
by the Midwest ($71,129) and the
sified into two categories: incomes. 22
South ($63,368).19 None of the
those who are naturalized
regions experienced a statisti-
U.S. citizens and those who are Educational Attainment23
cally significant change in median
not U.S. citizens. Neither group This section pertains to house-
household income between 2020
experienced a statistically sig- holders aged 25 and over. From
and 2021 (Figure 1 and Table
nificant change in their median 2020 to 2021, real median house-
A-1). 20
household income between 2020 hold income increased 2.7 percent
and 2021.17 Residence21 for householders with at least a
bachelor’s degree and declined
Households maintained by natu- In 2021, households inside met-
4.0 percent for those with some
ralized citizens had the highest ropolitan statistical areas (MSAs)
college. The 2021 real median
median household income in 2021 but outside principal cities had
incomes of householders with no
($74,150), followed by native- the highest median income
high school diploma and those
born householders ($71,522). ($79,599), followed by households
with a high school diploma but no
Households maintained by non- inside principal cities ($64,839).
college were not statistically dif-
citizens had the lowest median Households outside metropoli-
ferent from their respective 2020
household income ($57,132). tan areas had the lowest median
4 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 316 of 801
median incomes (Figure 1 and is distributed across the popula- unequal, but it does not offer
Table A-1). 24 tion; higher income inequality insight into whether the top of
represents less equal income the distribution is increasing dis-
Householders with more educa-
distribution or growth. The proportionally compared to the
tion had higher income. In 2021,
Census Bureau reports various middle of the distribution (“upper-
households maintained by some-
measures of income inequality: tail” inequality) or if the middle
one with at least a bachelor’s
(1) the Gini index, (2) the ratio of the distribution is gaining
degree had the highest median
of income percentiles, (3) the more compared to the lower end
income ($115,456), followed by
shares of aggregate household (“lower-tail” inequality). Percentile
those with some college ($64,378)
income by quintiles, (4) the Theil income ratios, particularly of the
and those with a high school
index, (5) the mean logarithmic 90th, 50th, and 10th percentiles
diploma ($50,401). Householders
deviation of income (MLD), and of the overall income distribution,
aged 25 and over with no high
(6) the Atkinson measures. This are widely used to provide addi-
school diploma had the lowest
section focuses on the first three tional information on observed
median income ($30,378).
measures pertaining to money changes in income inequality. 26
The median household income of income and equivalence-adjusted The ratio of the 90th to 10th
different education groups can be income, which are defined below percentile increased from 12.90
compared by calculating the ratio and shown in Table A-3 and Figure in 2020 to 13.53 in 2021, meaning
of the median income of a specific 3. Historical estimates for all six income at the 90th percentile was
group to the median income of summary measures of money 13.53 times higher than income at
householders with no high school income inequality are available the 10th percentile, an increase of
diploma. For 2021, the ratio for in Table A-4a and Table A-4b, 4.9 percent. The ratio of the 50th
householders with a bachelor’s and corresponding estimates for to 10th percentile (“lower-tail”
degree or higher was 3.8, meaning equivalence-adjusted income are inequality) increased 4.0 percent,
that householders with a bache- available in Table A-5. Post-tax from 4.34 in 2020 to 4.52 in 2021,
lor’s degree or higher had median income inequality estimates are while the ratio of the 90th to 50th
incomes 3.8 times greater than available in Tables C-3 and C-4. percentile (“upper-tail” inequality)
householders with no high school was not significantly different over
diploma. The ratio for household- Money Income Inequality this period. 27 Specifically, house-
ers with some college was 2.1, The Gini index is a statistical mea- hold income decreased 4.4 per-
while the ratio for householders sure of income inequality ranging cent at the 10th percentile limit,
with a high school diploma but from 0.0 to 1.0. It measures the while the change in income at the
no college was 1.7. The 2021 ratio amount that any two incomes dif- 90th percentile limit was not sta-
for householders with at least a fer, on average, relative to mean tistically significant between 2020
bachelor’s degree was higher than income. It is a natural indicator and 2021. 28 This indicates that
their 2020 ratio of 3.6, while the of how far apart or “spread out” declines in income at the bottom
differences in the ratios from 2020 incomes are from one another. of the income distribution may be
were not statistically significant A value of 0.0 represents per- contributing to the increase in the
for those with some college and fect equality, and a value of 1.0 Gini index.
those with a high school diploma. indicates total inequality. Based
The quintile shares of aggregate
on the money income Gini index,
household income provide addi-
INCOME INEQUALITY income inequality increased by 1.2
tional information about how
While the median represents percent between 2020 and 2021
income is distributed across the
the mid-point of the household (from 0.488 to 0.494); this repre-
population. A quintile is one of five
income distribution, other points sents the first time the Gini index
equal groups ranked by income
along the distribution provide has shown an annual increase
from lowest to highest, so that 20
additional information on how since 2011. 25
percent of all households are in
income is changing for those each group. In 2021, households
An increase in the Gini index
above and below the median. in the lowest quintile received
indicates that the distribution
Income inequality refers to how 2.9 percent of aggregate house-
of income is becoming more
evenly income or income growth hold income, while households
U.S. Census Bureau Income in the United States: 2021 5
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 317 of 801
Figure 3.
Income Distribution Measures and Percent Change Using Money Income
and Equivalence-Adjusted Income
Percent Change: 2020 to 2021
MONEY INCOME
Shares of Aggregate 20201 2021
Income by Percentile
Lowest quintile 3.0 2.9 -3.6
Second quintile 8.2 8.0 -1.7
Third quintile 14.0 13.9 -1.0
Fourth quintile 22.6 22.6 -0.2
Highest quintile 52.2 52.7 0.8
Top 5 percent 23.0 23.5 2.2
Summary Measures
Gini index of
income inequality 0.488 0.494 1.2
Income percentile ratios:
90th/10th 12.90 13.53 4.9
90th/50th 2.97 2.99 0.8
50th/10th 4.34 4.52 4.0
EQUIVALENCE-ADJUSTED INCOME
Shares of Aggregate
Income by Percentile
Lowest quintile 3.4 3.3 -2.0
Second quintile 8.9 8.8 -0.7
Third quintile 14.5 14.4 -0.6
Fourth quintile 22.4 22.3 -0.8
Highest quintile 50.8 51.2 0.8
Top 5 percent 22.5 23.0 2.0
Summary Measures
Gini index of
income inequality 0.469 0.474 0.9
Income percentile ratios:
90th/10th 10.73 10.89 1.5
90th/50th 2.80 2.81 0.1
50th/10th 3.83 3.88 1.3
Denotes a statistically
significant change
1 Implementation of 2020 Census-based population controls.
Note: Percent change estimate may be different due to rounded components. Statistically significant indicates the change is
statistically different from zero at the 90 percent confidence level. Margins of error and other related estimates are available in
Table A-3. Information on confidentiality protection, sampling error, nonsampling error, and definitions is available at
<https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 2021 and 2022 Annual Social and Economic Supplements (CPS ASEC).
6 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 318 of 801
in the highest quintile received the distribution based on money Based on equivalence-adjusted
52.7 percent of aggregate house- income treats a household income income, changes in inequality
hold income. Within the high- of $30,000 the same, regardless between 2020 and 2021 were not
est quintile, the top 5 percent of of whether one person or four statistically significant as mea-
households received 23.5 percent people live in the household. In sured by the shares of aggregate
of aggregate household income. contrast, the equivalence-adjusted income and the ratios of income
The share of aggregate household income would be the same for a percentiles (Table A-3). Table
income decreased in the low- single-person household with an A-5 shows equivalence-adjusted
est quintile (from 3.0 percent to income of $30,000 and a house- measures of the income distribu-
2.9 percent) and in the second hold with two married adults and tion, as well as the Gini index,
quintile (from 8.2 percent to 8.0 two children and an income of MLD, Theil index, and Atkinson
percent) between 2020 and 2021. nearly $65,000. The equivalence measures, for income years 1967
The changes in the other quintiles adjustment used here is based on to 2021.
were not statistically significant. the equivalence scale used in the
Supplemental Poverty Measure EARNINGS AND WORK
In 2021, households in the lowest STATUS
(SPM). 29 This section presents
quintile had incomes of $28,007
the same inequality measures as This section presents median
or less. Households in the sec-
the prior section but using equiv- earnings and work status for
ond quintile had incomes from
alence-adjusted income. These individuals aged 15 and older with
$28,008 to $55,000, those in
equivalence-adjusted income earnings. Earnings are the sum of
the third quintile had incomes
inequality measures are located in wage and salary income and non-
from $55,001 to $89,744, and
Table A-3 and Figure 3. farm and farm self-employment
those in the fourth quintile had
income (gross receipts minus
incomes from $89,745 to $149,131. For both 2020 and 2021, the Gini
expenses), comprising 78 per-
Households in the highest quintile index was lower when based on
cent of aggregate total income.
had incomes of $149,132 or more. an equivalence-adjusted income
Unemployment insurance pay-
The top 5 percent of households estimate (0.469 in 2020 and 0.474
ments are not included in earn-
in the income distribution had in 2021) than on the traditional
ings. Total workers (also referred
incomes of $286,305 or more. money-income estimate (0.488 in
to as “all workers”) include both
Table A-4a provides the income 2020 and 0.494 in 2021), suggest-
part-time and full-time workers.
limits for each decile and house- ing a more equal income distribu-
A full-time, year-round worker is
hold income ratios at selected tion when household composition
a person who worked at least 35
percentiles for income years 1967 is taken into account. Generally,
hours per week (full-time) and at
to 2021. Table A-4b provides the income shares in the lowest,
least 50 weeks per year (year-
quintile measures, as well as the second, and third quintiles are
round). 30 As with median house-
Gini index, MLD, Theil index, and higher with equivalence-adjusted
hold income, earnings estimates
Atkinson measures, for income income than money income,
are expressed in real or con-
years 1967 to 2021. while the reverse is true for the
stant dollar terms, meaning that
fourth and highest quintiles. This
Equivalence-Adjusted Income median earnings estimates for
redistribution reflects the higher
Inequality 2020 are inflation-adjusted by 4.7
concentration of single-person
percent to 2021 dollars. Year-to-
Another way to measure income households and smaller house-
year percent changes reflect this
inequality is to replace money hold sizes at the lower end of
adjustment. 31
income with an equivalence- the income distribution. While
adjusted income estimate that the money income Gini index
Total and Full-Time, Year-Round
takes into consideration the increased between 2020 and Workers
number of people living in the 2021, the change in the equiva-
Between 2020 and 2021, the
household and how those people lence-adjusted Gini index was not
change in the number of total
share resources and benefit from statistically significant.
workers was not statistically
economies of scale. For example,
U.S. Census Bureau Income in the United States: 2021 7
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 319 of 801
Figure 4.
Median Earnings and Percent Change by Work Status and Sex
(People 15 years and older with earnings as of March of the following year)
2021 Median Earnings Change: 2020 to 2021
Total Workers $45,470 4.6
Men $50,983 -0.9
Women $39,201 4.5
Full-Time, Year-Round
Workers $56,473 -4.1
Men $61,180 -4.7
Women $51,226 -4.0
Denotes a statistically
significant change
Note: Statistically significant indicates the change is statistically different from zero at the 90 percent confidence level. Margins of error
and other related estimates and notes are available in Table A-6. Information on confidentiality protection, sampling error, nonsampling
error, and definitions is available at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 2021 and 2022 Annual Social and Economic Supplements (CPS ASEC).
significant; however, there was an full-time, year-round decreased 4.1 median earnings among all work-
increase of about 11.1 million full- percent from 2020. This decline in ers increased 4.6 percent from
time, year-round workers (from median earnings may reflect both 2020, but the increase was not
approximately 106.3 million to the effects of inflation surpassing experienced equally by men and
117.4 million), suggesting a shift nominal gains in earnings as well women. The 2021 median earnings
from working part-time or part- as the addition of full-time jobs of working women increased 4.5
year in 2020 to full-time, year- in the lower half of the earnings percent from their 2020 median,
round work in 2021. 32 The increase distribution. while the change for their male
in the number of full-time, year- counterparts was not statistically
round workers corresponds with Workers by Sex significant (Figure 4 and Table
an increase of 4.6 percent in the Looking at the details of median A-6). 33 Between 2020 and 2021,
real median earnings of all work- earnings and worker composition the changes in the number of male
ers between 2020 and 2021. by sex can add more context to the and female workers were not sta-
Conversely, the 2021 real median annual changes experienced by the tistically significant.
earnings of those who worked total working population. The 2021
8 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 320 of 801
Figure 5.
Total and Full-Time, Year-Round Workers 15 Years and Older With Earnings by Sex:
1967 to 2021
(People with earnings as of March of the following year)
Numbers in millions Recession
100
90 88.9 million
80 79.1 million
70
66.4 million
60
Male workers
51.0 million
50
Male full-time,
year-round workers
40
30 Female workers
20
Female full-time,
10
year-round workers
0
1959 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2021
Note: Refer to Table A-7 for historical footnotes. The data points are placed at the midpoints of the respective years.
Data on earnings of full-time, year-round workers are not readily available before 1960. Data are for people aged 14
and older for years prior to 1980. More information on recessions is available in Appendix A. Information on
confidentiality protection, sampling error, nonsampling error, and definitions is available at
<https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 1968 to 2022 Annual Social and Economic Supplements
(CPS ASEC).
Consistent with the findings for The number of male full-time, full-time, year-round increased 9.7
all full-time, year-round workers, year-round workers increased by percent from the 2020 estimate of
median earnings of men ($61,180) about 6.1 million between 2020 68.0 percent to 74.6 percent. The
and women ($51,226) who worked and 2021, while the increase in share of women working full-time,
full-time, year-round decreased the number of their female coun- year-round increased 11.4 percent
by 4.7 percent and 4.0 percent, terparts was approximately 5.0 from 57.9 percent in 2020 to 64.5
respectively, between 2020 and million (Figure 5 and Table A-6). percent in 2021.
2021 (Figure 4 and Table A-6). 34 In 2021, the share of men working
U.S. Census Bureau Income in the United States: 2021 9
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 321 of 801
Figure 6.
Female-to-Male Earnings Ratio and Median Earnings of Full-Time, Year-Round Workers
15 Years and Older by Sex: 1960 to 2021
(People as of March of the following year)
Ratio in percent Recession
100
90
Female-to-male earnings ratio 83.7 percent
80
70
60
50
0
Earnings in thousands (2021 R-CPI-U-RS adjusted dollars)
80
Earnings of men $61,180
60
$51,226
40
Earnings of women
20
0
1959 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2021
Notes: Refer to Table A-7 for historical footnotes. The data points are placed at the midpoints of the respective years.
Data on earnings of full-time, year-round workers are not readily available before 1960. Data are for people aged 14 and
older for years prior to 1980. More information on the R-CPI-U-RS dollar adjustment and recessions is available in
Appendix A. Information on confidentiality protection, sampling error, nonsampling error, and definitions is available at
<https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>.
Source: U.S. Census Bureau, Current Population Survey, 1961 to 2022 Annual Social and Economic Supplements
(CPS ASEC).
The female-to-male earnings ratio SUMMARY was accompanied by an increase
compares the median earnings of in real median earnings among
This report provides estimates
women working full-time, year- all workers and, conversely, a
of household income, income
round to the median earnings of decrease in median earnings for
inequality, and worker earnings in
men working full-time, year-round. full-time, year-round workers.
the United States for 2021. Overall,
The 2021 female-to-male earn- The decline in real median earn-
real median household income in
ings ratio was 0.837, not statisti- ings among full-time, year-round
2021 was not statistically differ-
cally different from the 2020 ratio workers between 2020 and 2021
ent from 2020, although income
(0.831). The last time the female- may reflect the addition of full-
inequality increased due to
to-male earnings ratio experi- time jobs in the lower half of the
declines in income at the bot-
enced a statistically significant earnings distribution, the effects
tom of the income distribution as
annual change was in 2016 (Figure of inflation surpassing nominal
measured by the Gini index and
6 and Table A-7). For historical gains in earnings, or other factors.
percentile ratios. In 2021, there
statistics from 1960 to 2021 on Further analysis of the data would
was a shift from individuals work-
median earnings and number of be necessary to know how these
ing part-time or part-year to full-
workers by sex, refer to Table A-7. and other factors affected the
time, year-round work. This shift
earnings estimates.
10 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 322 of 801
ENDNOTES 9
A family household is a household 13
The small sample size of the Asian
maintained by a householder who is related population and the fact that the CPS ASEC
1
“Real” refers to income after adjusting to at least one other person in the house- does not use separate population controls
for inflation. hold by birth, marriage, or adoption and for weighting the Asian sample to national
2
The R-CPI-U-RS is provided by the includes any unrelated individuals who totals contribute to the large variances
U.S. Bureau of Labor Statistics (BLS). In reside there. Married-couple households surrounding estimates for this group. The
2021, BLS renamed the Consumer Price include both opposite-sex and same-sex American Community Survey, based on a
Index Research Series (CPI-U-RS), the couples. A nonfamily household is a house- much larger sample of the population, is a
Retroactive Series (R-CPI-U-RS). More holder living alone (a one-person house- better source for estimating and identify-
information on the R-CPI-U-RS and the hold) or sharing the home exclusively with ing changes for small subgroups of the
index values for 1947 to 2021 are available nonrelatives. population.
in Appendix A. For an in-depth discussion 10
The difference between the 2020– 14
The following differences between
of the effects of using different inflation 2021 percent change in median income of the 2020–2021 percent changes in median
indexes on household income estimates nonfamily households and those main- household income were not statistically sig-
and information on a proposed change tained by male householders was not statis- nificant: householders under the age of 65
to the index used in this report, refer to tically significant. and each age category of households aged
Appendix D. 11
Federal surveys give respondents the 25 to 54; householders aged 15 to 24 and
3
For more information on the 2021 option of reporting more than one race. each age category of householders aged
inflation-adjustment, refer to “How Therefore, two basic ways of defining a race 25 to 54; householders aged 25 to 34 and
Inflation Affects the Census Bureau’s group are possible. A group, such as Asian, each age category 35 and older; house-
Income and Earnings Estimates,” at <www. may be defined as those who reported holders aged 35 to 44 and every other age
census.gov/newsroom/blogs/random- Asian and no other race (the race-alone category; and householders aged 55 to 64
samplings/2022/09/inflation-income-and- or single-race concept) or as those who and those over the age of 65.
earnings-estimates.html>. reported Asian regardless of whether they 15
The difference between the 2021
4
Median income is the amount that also reported another race (the race-alone- median household income for householders
divides the income distribution into two or-in-combination concept). The body aged 25 to 34 and those aged 55 to 64 was
equal groups, one-half having incomes of this report (text and figures) provides not statistically significant.
above the median, one-half having incomes data using the first approach (race alone). 16
Native-born households are those
below the median. Calculated differences The appendix tables provide data using in which the householder was born in the
throughout this report may differ due to both approaches. Use of the single-race United States, Puerto Rico, the U.S. Island
rounding. population does not imply that it is the Areas of Guam, the Commonwealth of the
5
Creamer, John, Emily A. Shrider, Kalee preferred method of presenting or analyz- Northern Mariana Islands, American Samoa,
Burns, and Frances Chen, “Poverty in the ing data. The Census Bureau uses a variety the Virgin Islands of the United States, or a
United States: 2021,” Current Population of approaches. In this report, the terms foreign country but had at least one parent
Reports, P60-277, U.S. Census Bureau, “White, not Hispanic” and “non-Hispanic who was a U.S. citizen. All other households
Washington, DC, September 2022, <www. White” are used interchangeably and refer are considered foreign-born regardless
census.gov/library/publications/2022/ to people who are not Hispanic and who of the date of entry into the United States
demo/p60-277.html> and Keisler-Starkey, reported White and no other race. This or citizenship status. The CPS does not
Katherine and Lisa N. Bunch, “Health report uses non-Hispanic Whites as the interview households in Puerto Rico. Of all
Insurance in the United States: 2021,” comparison group for other race groups householders, 84.4 percent were native-
Current Population Reports, P60-278, and Hispanics. born; 8.6 percent were foreign-born,
U.S. Census Bureau, Washington, DC, Since Hispanics may be any race, data in naturalized citizens; and 6.9 percent were
September 2022, <https://www.census. this report for Hispanics overlap with data not U.S. citizens.
gov/library/publications/2022/demo/p60- for race groups. Of those who reported 17
The differences among the 2020–2021
278.html>. only one race, being Hispanic was reported percent changes in median household
6
If a married couple owns the home by 16.6 percent of White householders, income by nativity of the householder were
jointly, either spouse may be listed as the 5.6 percent of Black householders, and 2.9 not statistically significant.
householder. percent of Asian householders. 18
The Northeast region includes
7
Refer to Appendix A for informa- Data users should exercise caution when Connecticut, Maine, Massachusetts,
tion on business cycles and recessions interpreting aggregate results for the New Hampshire, New Jersey, New York,
as defined by the National Bureau of Hispanic population or for race groups Pennsylvania, Rhode Island, and Vermont.
Economic Research. For more information because these populations consist of many The Midwest region includes Illinois,
on changes in household income during distinct groups that differ in socioeconomic Indiana, Iowa, Kansas, Michigan, Minnesota,
previous recessions, refer to DeNavas- characteristics, culture, and nativity. Data Missouri, Nebraska, North Dakota, Ohio,
Walt, Carmen, Bernadette D. Proctor, and were first collected for Hispanics in 1972 South Dakota, and Wisconsin. The South
Jessica C. Smith, “Income, Poverty, and and for Asians and Pacific Islanders in 1987. region includes Alabama, Arkansas,
Health Insurance Coverage in the United More information is available at <www. Delaware, Florida, Georgia, Kentucky,
States: 2008,” Current Population Reports, census.gov/programs-surveys/cps.html>. Louisiana, Maryland, Mississippi, North
P60-236, U.S. Census Bureau, Washington, 12
The differences among the 2020–2021 Carolina, Oklahoma, South Carolina,
DC, September 2009, <www.census.gov/ percent changes in household median Tennessee, Texas, Virginia, West Virginia,
prod/2009pubs/p60-236.pdf>. income for the race groups were not statis- and the District of Columbia. The West
8
For more information on historical tically significant. region includes Alaska, Arizona, California,
income comparisons across the recent Colorado, Hawaii, Idaho, Montana, Nevada,
survey redesigns, refer to “Was Household New Mexico, Oregon, Utah, Washington,
Income the Highest Ever in 2019?” at and Wyoming.
<www.census.gov/library/stories/2020/09/
was-household-income-the-highest-ever-
in-2019.html>.
U.S. Census Bureau Income in the United States: 2021 11
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 323 of 801
19
The difference in 2021 median house- householders with a high school diploma 30
For school personnel, summer vaca-
hold incomes for the Northeast and the but no college and those with at least a tion is counted as weeks worked if they
West was not statistically significant. bachelor’s degree was not statistically are scheduled to return to their job in the
20
The differences among the 2020–2021 significant. fall. For more detailed information on work
percent changes in median household 25
Money income is the baseline measure experience, refer to Table PINC-05, “Work
income for the regions were not statistically of income in this report. Money income is Experience in 2021—People 15 Years Old
significant. calculated pretax; refer to Appendix A for and Over by Total Money Earnings in 2021,
21
The definitions of metropolitan statis- a detailed list of all income components. Age, Race, Hispanic Origin, and Sex” at
tical areas and principal cities are available For inequality estimates based on post-tax <www.census.gov/data/tables/time-series/
at <www.census.gov/programs-surveys/ income, refer to Appendix C. demo/income-poverty/cps-pinc/pinc-05.
metro-micro/about.html>. 26
Wimer, Christopher, Zachary html>.
22
The differences among the 2020–2021 Parolin, Amy Fenton, Liana Fox, and 31
The index used in this report to
percent changes in median household Christopher Jencks, “The Direct Effect inflation-adjust is Consumer Price Index
incomes for all categories of metropoli- of Taxes and Transfers on Changes in for all Urban Consumers Retroactive Series
tan statistical areas were not statistically the U.S. Income Distribution, 1967–2015,” (R-CPI-U-RS); more information and histori-
significant. Demography, August 24, 2020, <https:// cal index values are available in Appendix A.
23
Information on educational read.dukeupress.edu/demography/ 32
Estimates for counts of workers in
attainment in the CPS ASEC is available article/57/5/1833/168378/The-Direct- 2020 and 2021 are based on 2020 Census
at <www.census.gov/programs-surveys/ Effect-of-Taxes-and-Transfers-on>, population controls and not directly com-
cps/technical-documentation/subject- accessed on June 21, 2022. parable to estimates for 2019, which were
definitions.html#educationalattainment>. 27
The difference between the 2020– 2010 Census-based. Refer to Appendix
Those with an associate degree are 2021 percent changes in the ratios of the B for more information on the change in
included in the “some college” category. 90th- to 10th-percentile and the 50th- population controls.
24
The 2020–2021 percent change in to 10th-percentile was not statistically 33
The differences between the 2020–
median household income for all house- significant. 2021 percent changes in median earnings
holders aged 25 and over declined 1.3 28
The differences among the 2020–2021 for all workers and working females with
percent, not statistically different from the percent changes in household income at earnings were not statistically significant.
2020–2021 percent changes for household- each percentile limit were not statistically 34
The following differences between
ers with no high school diploma and those significant. the 2020–2021 percent changes in median
with a high school diploma but no college. 29
For more details on the three- earnings were not statistically significant:
The differences among the 2020–2021 parameter equivalence scale, refer to all full-time, year-round workers and female
percent changes in median household Creamer, John, Emily A. Shrider, Kalee full-time, year-round workers; and male
income for householders with no high Burns, and Frances Chen, “Poverty in the full-time, year-round workers and female
school diploma were not statistically signifi- United States: 2021,” Current Population full-time, year-round workers.
cant from either those with a high school Reports, P60-277, U.S. Census Bureau,
diploma or those with some college. The Washington, DC, September 2022, <www.
difference between the 2020–2021 percent census.gov/library/publications/2022/
change in median household income for demo/p60-277.html>.
12 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 324 of 801
APPENDIX A. ESTIMATES OF INCOME
How Income Is Measured
For each person 15 years and Business Cycles—Recessions
older in the sample, the Current Peak month Year Trough month Year
Population Survey Annual Social November 1948 October 1949
and Economic Supplement (CPS July 1953 May 1954
ASEC) asks questions on the
August 1957 April 1958
amount of money income received
in the preceding calendar year April 1960 February 1961
from each of the following sources. December 1969 November 1970
1. Earnings November 1973 March 1975
January 1980 July 1980
2. Unemployment compensation
July 1981 November 1982
3. Workers’ compensation
July 1990 March 1991
4. Social Security
March 2001 November 2001
5. Supplemental Security Income
December 2007 June 2009
6. Public assistance
February 2020 April 2020
7. Veterans’ payments
Source: National Bureau of Economic Research, <www.nber.org/research/
8. Survivor benefits data/us-business-cycle-expansions-and-contractions>.
9. Disability benefits
10. Pension or retirement income
such as Supplemental Nutrition household composition, are as of
11. Interest Assistance/food stamps, health the survey date.
12. Dividends benefits, and subsidized housing.
Data users should consider these
13. Rents, royalties, and estates and In addition, money income does
elements when comparing income
trusts not reflect the fact that noncash
levels. Moreover, readers should
benefits often take the form of the
14. Educational assistance be aware that for many differ-
use of business transportation and
15. Alimony ent reasons there is a tendency in
facilities, full or partial payments
household surveys for respondents
16. Child support by business for retirement pro-
to underreport their income. Based
17. Financial assistance from out- grams, or medical and educational
on an analysis of independently
side of the household expenses.
derived income estimates, the
18. Other income The income of the household does Census Bureau determined that
Data on income collected in the not include amounts received by respondents report income earned
CPS ASEC by the U.S. Census people who were members during from wages or salaries more
Bureau cover money income all or part of the previous year if accurately than other sources of
received (exclusive of certain these people no longer resided in income, and that the reported
money receipts such as capital the household at the time of the wage and salary income is nearly
gains) before payments for per- interview. However, the CPS ASEC equal to independent estimates of
sonal income taxes, Social Security, includes income data for people aggregate income.
union dues, Medicare deductions, who are current residents but did
etc. Money income also excludes not reside in the household dur- Business Cycles—Recessions
tax credits such as the Earned ing the previous year. It should be
Business cycle peaks and troughs
Income Tax Credit, the Child Tax noted that although the income
used to delineate the begin-
Credit, and special COVID-19- statistics refer to receipts during
ning and end of recessions, as
related stimulus payments. Money the preceding calendar year, the
shown in the text box “Business
income does not reflect that some demographic characteristics, such
Cycles—Recessions,” are deter-
families receive noncash benefits as age, labor force status, and
mined by the National Bureau
U.S. Census Bureau Income in the United States: 2021 13
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 325 of 801
of Economic Research (NBER), Annual Average and Annual Percent Change of the Consumer Price
a private research organization. Index Retroactive Series (R-CPI-U-RS) Using Current Methods All
The data points in the time series Items: 1947 to 2021
figures in this report use July as a R-CPI-U-RS1 R-CPI-U-RS1
reference. According to the NBER Income index Percentage Income index Percentage
year (December change from year (December change from
chronology, the most recent peak
1977 = 100) year prior 1977 = 100) year prior
occurred in February 2020. The
1947. . . . . 37.5 X 1985. . . . . 165.7 3.4
most recent trough occurred in 1948. . . . . 40.5 8.0 1986. . . . . 168.6 1.8
April 2020. More information on 1949. . . . . 40.0 –1.2 1987. . . . . 174.4 3.4
business cycle dating is avail- 1950. . . . . 40.5 1.3 1988. . . . . 180.7 3.6
1951. . . . . 43.7 7.9 1989. . . . . 188.6 4.4
able at <www.nber.org/research/ 1952. . . . . 44.5 1.8 1990. . . . . 197.9 4.9
business-cycle-dating>. 1953. . . . . 44.8 0.7 1991. . . . . 205.1 3.6
1954. . . . . 45.2 0.9 1992. . . . . 210.2 2.5
1955. . . . . 45.0 –0.4 1993. . . . . 215.5 2.5
Cost-of-Living Adjustment 1956. . . . . 45.7 1.6 1994. . . . . 220.0 2.1
To accurately assess changes in 1957. . . . . 47.2 3.3 1995. . . . . 225.3 2.4
1958. . . . . 48.5 2.8 1996. . . . . 231.3 2.7
income and earnings over time, an 1959. . . . . 48.9 0.8 1997. . . . . 236.3 2.2
adjustment for changes in the cost 1960. . . . . 49.7 1.6 1998. . . . . 239.5 1.4
of living is required. This report 1961. . . . . 50.2 1.0 1999. . . . . 244.6 2.1
1962. . . . . 50.7 1.0 2000. . . . . 252.9 3.4
and other data products, such as 1963. . . . . 51.4 1.4 2001. . . . . 260.1 2.8
tables and figures based on the 1964. . . . . 52.1 1.4 2002. . . . . 264.2 1.6
CPS ASEC, use the Consumer Price 1965. . . . . 52.9 1.5 2003. . . . . 270.2 2.3
1966. . . . . 54.4 2.8 2004. . . . . 277.5 2.7
Index Retroactive Series for all
1967. . . . . 56.1 3.1 2005. . . . . 286.9 3.4
Urban Consumers All Items (R-CPI- 1968. . . . . 58.3 3.9 2006. . . . . 296.2 3.2
U-RS), provided by the U.S. Bureau 1969. . . . . 60.9 4.5 2007. . . . . 304.6 2.8
of Labor Statistics (BLS) for 1978 1970. . . . . 63.9 4.9 2008. . . . . 316.3 3.8
1971. . . . . 66.7 4.4 2009. . . . . 315.2 –0.3
through 2021, to adjust for changes 1972. . . . . 68.7 3.0 2010. . . . . 320.4 1.6
in the cost of living. The R-CPI- 1973. . . . . 73.0 6.3 2011. . . . . 330.5 3.2
U-RS was formerly known as the 1974. . . . . 80.3 10.0 2012. . . . . 337.5 2.1
1975. . . . . 86.9 8.2 2013. . . . . 342.5 1.5
Consumer Price Index Research
1976. . . . . 91.9 5.8 2014. . . . . 348.3 1.7
Series (CPI-U-RS). For years prior 1977. . . . . 97.7 6.3 2015. . . . . 348.9 0.2
to 1978, the Census Bureau used 1978. . . . . 104.4 6.9 2016. . . . . 353.4 1.3
estimates provided by BLS from the 1979. . . . . 114.3 9.5 2017. . . . . 361.0 2.2
1980. . . . . 127.1 11.2 2018. . . . . 369.8 2.4
CPI-U-X1 series. The CPI-U-X1 is an 1981. . . . . 139.1 9.4 2019. . . . . 376.5 1.8
experimental series that preceded 1982. . . . . 147.5 6.0 2020. . . . . 381.2 1.2
the R-CPI-U-RS and estimates the 1983. . . . . 153.8 4.3 2021. . . . . 399.0 4.7
1984. . . . . 160.2 4.2
inflation rate in the Consumer Price
X Not applicable.
Index for all Urban Consumers 1
The U.S. Census Bureau uses the Bureau of Labor Statistics’ (BLS) Consumer Price Index
(CPI-U) when applying the current for all Urban Consumers Retroactive Series (R-CPI-U-RS) for 1978 through 2021. In 2021, BLS
renamed the Research Series (CPI-U-RS) the Retroactive Series. For 1967 to 1977, the Census
rental equivalence method of mea- Bureau uses estimates provided by BLS from the CPI-U-X1 series. The CPI-U-X1 is an experi-
suring the cost of homeownership mental series that preceded the CPI-U-RS and estimates the inflation rate in the CPI-U when
for years prior to 1983. The index applying the current rental equivalence method of measuring the cost of homeownership
for years prior to 1983. The Census Bureau derived the R-CPI-U-RS for years before 1967 by
used to make the constant dollar applying the 1967 R-CPI-U-RS-to-CPI-U ratio to the 1947 to 1966 CPI-U.
conversions in the main body of Note: Data users can compute the percentage changes in prices between earlier years’
data and 2021 data by dividing the annual average R-CPI-U-RS for 2021 by the annual aver-
this report is shown in the text box age for the earlier year(s). More information on the R-CPI-U-RS is available at <www.bls.gov/
“Annual Average Consumer Price cpi/research-series/r-cpi-u-rs-home.htm>.
Index Retroactive Series (R-CPI-
U-RS) Using Current Methods All
Items: 1947 to 2021.” Appendix D
discusses alternative price indexes
and how they would affect esti-
mates of income over time includ-
ing a proposal to change the index
used in this report to make histori-
cal income adjustments.
14 Income in the United States: 2021 U.S. Census Bureau
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 326 of 801
Table A-1.
Income Summary Measures by Selected Characteristics: 2020 and 2021
(Income in 2021 dollars, adjusted using the R-CPI-U-RS. Households as of March of the following year. Information on
confidentiality protection, sampling error, nonsampling error, and definitions is available at <https://www2.census.gov/
programs-surveys/cps/techdocs/cpsmar22.pdf>)
20201 2021 Percent change in
Median income Median income real median income
Characteristic Number (dollars) Number (dollars) (2021 less 2020)*
(thou- Margin of (thou- Margin of Margin of
sands) Estimate error2 (±) sands) Estimate error2 (±) Estimate error2 (±)
HOUSEHOLDS
All households. . . . . . . . . . . . . . . . . . . . . . . 129,244 71,186 921 131,202 70,784 605 –0.6 1.31
Type of Household
Family households. . . . . . . . . . . . . . . . . . . . . . . 83,711 90,722 894 84,265 91,162 787 0.5 1.15
Married-couple. . . . . . . . . . . . . . . . . . . . . . . . 61,288 106,582 891 61,435 106,921 937 0.3 1.09
Female householder, no spouse present. . 15,461 51,554 1,515 15,618 51,168 925 –0.7 3.10
Male householder, no spouse present. . . . 6,963 70,478 2,458 7,212 70,525 1,904 0.1 4.09
Nonfamily households . . . . . . . . . . . . . . . . . . . 45,533 42,607 676 46,937 41,797 590 *–1.9 1.75
Female householder. . . . . . . . . . . . . . . . . . . 23,859 37,516 712 24,221 35,737 811 *–4.7 2.49
Male householder. . . . . . . . . . . . . . . . . . . . . . 21,674 49,625 1,329 22,716 49,466 1,467 –0.3 3.42
Race3 and Hispanic Origin of Householder
White. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100,931 74,978 771 102,057 74,262 912 –1.0 1.30
White, not Hispanic. . . . . . . . . . . . . . . . . . . . 84,712 78,912 889 85,078 77,999 1,080 –1.2 1.43
Black. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17,319 48,175 1,327 17,698 48,297 1,679 0.3 4.08
Asian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7,002 99,622 3,983 7,276 101,418 2,868 1.8 4.52
Hispanic (any race). . . . . . . . . . . . . . . . . . . . . . 18,340 58,015 1,213 19,230 57,981 1,585 –0.1 3.09
Age of Householder
Under 65 years. . . . . . . . . . . . . . . . . . . . . . . . . . 94,593 80,456 771 95,370 80,734 613 0.3 1.09
15 to 24 years. . . . . . . . . . . . . . . . . . . . . . . . . 5,498 49,094 1,612 6,061 51,645 1,575 *5.2 4.65
25 to 34 years. . . . . . . . . . . . . . . . . . . . . . . . . 20,570 74,958 1,213 20,990 74,862 1,932 –0.1 2.90
35 to 44 years. . . . . . . . . . . . . . . . . . . . . . . . . 22,304 89,711 1,788 22,601 90,312 1,561 0.7 2.52
45 to 54 years. . . . . . . . . . . . . . . . . . . . . . . . . 21,803 94,633 2,024 21,647 97,089 1,598 *2.6 2.45
55 to 64 years. . . . . . . . . . . . . . . . . . . . . . . . . 24,417 77,872 2,176 24,070 75,842 1,443 –2.6 3.01
65 years and older. . . . . . . . . . . . . . . . . . . . . . . 34,651 48,866 976 35,832 47,620 1,037 *–2.6 2.46
Nativity of Householder
Native-born. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109,633 72,552 1,022 110,800 71,522 692 *–1.4 1.41
Foreign-born . . . . . . . . . . . . . . . . . . . . . . . . . . . 19,611 65,061 1,052 20,402 66,043 1,494 1.5 2.57
Naturalized citizen. . . . . . . . . . . . . . . . . . . . . 11,202 72,467 2,140 11,332 74,150 2,458 2.3 3.86
Not a citizen. . . . . . . . . . . . . . . . . . . . . . . . . . 8,409 57,804 1,813 9,070 57,132 2,152 –1.2 4.85
Region
Northeast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22,471 79,032 1,576 22,640 77,472 2,705 –2.0 3.54
Midwest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27,811 70,528 1,881 28,050 71,129 1,284 0.9 2.63
South. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49,759 64,355 859 50,612 63,368 1,218 –1.5 2.03
West. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29,203 78,755 1,225 29,900 79,430 1,482 0.9 2.20
Residence4
Inside metropolitan statistical areas. . . . . . . 111,460 74,622 694 113,267 73,823 941 –1.1 1.33
Inside principal cities. . . . . . . . . . . . . . . . . . . 43,273 65,609 1,385 43,625 64,839 1,503 –1.2 2.53
Outside principal cities. . . . . . . . . . . . . . . . . 68,188 80,017 913 69,642 79,599 1,109 –0.5 1.61
Outside metropolitan statistical areas . . . . . 17,784 54,300 1,222 17,935 53,750 2,026 –1.0 3.27
Educational Attainment of Householder
Total, aged 25 and older. . . . . . . . . . . . . . . . 123,746 73,013 912 125,141 72,046 627 *–1.3 1.28
No high school diploma. . . . . . . . . . . . . . . . . . 9,961 31,130 1,098 10,012 30,378 774 –2.4 3.90
High school, no college . . . . . . . . . . . . . . . . . . 31,401 49,965 1,103 32,214 50,401 795 0.9 2.30
Some college. . . . . . . . . . . . . . . . . . . . . . . . . . . 33,434 67,075 1,426 33,791 64,378 1,483 *–4.0 2.59
Bachelor’s degree or higher . . . . . . . . . . . . . . 48,950 112,393 1,692 49,125 115,456 1,771 *2.7 1.94
* An asterisk preceding an estimate indicates change is statistically different from zero at the 90 percent confidence level.
1
Implementation of 2020 Census-based population controls.
2
A margin of error (MOE) is a measure of an estimate’s variability. The larger the MOE in relation to the size of the estimate, the less reliable
the estimate. This number, when added to and subtracted from the estimate, forms the 90 percent confidence interval. MOEs shown in this table
are based on standard errors calculated using replicate weights.
3
Federal surveys give respondents the option of reporting more than one race. Therefore, two basic ways of defining a race group are pos-
sible. A group, such as Asian, may be defined as those who reported Asian and no other race (the race-alone or single-race concept) or as those
who reported Asian regardless of whether they also reported another race (the race-alone-or-in-combination concept). This table shows data
using the first approach (race alone). The use of the single-race population does not imply that it is the preferred method of presenting or analyz-
ing data. The Census Bureau uses a variety of approaches. Data for American Indians and Alaska Natives, Native Hawaiians and Other Pacific
Islanders, and those reporting Two or More Races are not shown separately.
4
Information on metropolitan statistical areas and principal cities is available at <www.census.gov/programs-surveys/metro-micro/about/
glossary.html>.
Note: Inflation-adjusted estimates may differ slightly from other published data due to rounding.
Source: U.S. Census Bureau, Current Population Survey, 2021 and 2022 Annual Social and Economic Supplements (CPS ASEC).
U.S. Census Bureau Income in the United States: 2021 15
Table A-2.
Households by Total Money Income, Race, and Hispanic Origin of Householder: 1967 to 2021—Con.
(Income in 2021 dollars, adjusted using the R-CPI-U-RS. Households as of March of the following year. Information on confidentiality protection, sampling error,
nonsampling error, and definitions is available at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>)
Median income Mean income
Percent distribution
Race and Hispanic (dollars) (dollars)
origin of householder Number $15,000 $25,000 $35,000 $50,000 $75,000 $100,000 $150,000
and year (thou- Under to to to to to to to $200,000 Margin of Margin of
sands) Total $15,000 $24,999 $34,999 $49,999 $74,999 $99,999 $149,999 $199,999 and over Estimate error1 (±) Estimate error1 (±)
ALL RACES
2021. . . . . . . . . . . . . . . . 131,202 100 9.3 8.1 7.8 10.9 16.2 11.9 15.9 8.3 11.6 70,784 605 102,316 1,029
20202 . . . . . . . . . . . . . . . 129,244 100 8.8 8.2 8.1 11.0 16.2 12.3 15.8 8.4 11.3 71,186 921 102,020 1,098
2019. . . . . . . . . . . . . . . . 128,451 100 8.5 7.5 8.0 11.3 16.1 12.1 16.3 8.7 11.5 72,808 959 103,949 1,104
2018. . . . . . . . . . . . . . . . 128,579 100 9.4 8.3 8.4 11.5 16.4 12.8 15.6 7.9 9.8 68,168 746 97,129 969
16 Income in the United States: 2021
20173 . . . . . . . . . . . . . . . 127,669 100 9.4 8.6 8.6 11.8 15.8 12.6 15.2 7.7 10.1 67,571 585 96,868 1,037
2017. . . . . . . . . . . . . . . . 127,586 100 9.5 8.8 8.5 11.8 15.6 12.8 15.3 7.9 9.7 67,832 609 95,296 945
2016. . . . . . . . . . . . . . . . 126,224 100 9.8 8.5 8.6 11.8 16.5 12.2 15.6 7.8 9.2 66,657 810 93,871 871
2015. . . . . . . . . . . . . . . . 125,819 100 9.8 9.3 9.3 11.7 16.2 12.1 15.7 7.6 8.4 64,631 604 90,645 758
2014. . . . . . . . . . . . . . . . 124,587 100 10.7 9.7 9.5 12.0 16.6 11.9 14.7 7.1 7.8 61,468 739 86,763 840
20134 . . . . . . . . . . . . . . . 123,931 100 10.5 9.9 9.2 11.8 16.5 12.4 14.5 7.1 8.0 62,425 1,253 87,599 1,272
20135 . . . . . . . . . . . . . . . 122,952 100 10.4 10.1 9.5 12.1 16.9 12.9 14.2 6.9 6.9 60,507 529 84,624 956
2012. . . . . . . . . . . . . . . . 122,459 100 10.4 10.3 9.2 12.9 16.6 12.5 14.5 6.8 6.9 60,313 406 84,262 819
2011. . . . . . . . . . . . . . . . 121,084 100 10.6 9.8 9.2 13.3 17.0 12.2 14.4 6.8 6.8 60,428 498 84,118 731
20106 . . . . . . . . . . . . . . . 119,927 100 10.6 10.1 9.1 12.8 16.6 12.3 14.9 6.9 6.8 61,364 666 83,924 737
20097 . . . . . . . . . . . . . . . 117,538 100 9.3 9.4 9.6 12.5 17.1 12.8 15.3 7.0 7.1 63,011 444 86,048 506
2008. . . . . . . . . . . . . . . . 117,181 100 9.3 9.3 9.5 12.4 16.8 12.9 15.6 7.2 7.0 63,455 284 86,314 502
2007. . . . . . . . . . . . . . . . 116,783 100 8.9 9.4 8.9 12.0 17.1 12.8 15.9 7.5 7.4 65,801 302 88,562 509
2006. . . . . . . . . . . . . . . . 116,011 100 9.0 8.9 8.8 12.8 17.2 12.6 15.8 7.2 7.7 64,930 459 89,674 569
2005. . . . . . . . . . . . . . . . 114,384 100 9.3 9.0 9.4 12.1 17.1 13.0 15.8 7.0 7.4 64,427 355 88,094 547
20048 . . . . . . . . . . . . . . . 113,343 100 9.5 9.1 9.1 12.3 17.4 12.7 15.8 7.0 7.1 63,745 464 86,940 539
2003. . . . . . . . . . . . . . . . 112,000 100 9.4 9.3 8.7 12.4 17.2 12.5 16.1 7.2 7.1 63,967 457 87,223 525
2002. . . . . . . . . . . . . . . . 111,278 100 9.0 9.5 8.7 12.6 16.7 13.4 15.9 7.4 6.8 64,047 345 87,369 539
2001. . . . . . . . . . . . . . . . 109,297 100 8.7 9.3 8.5 12.6 16.9 13.4 16.1 7.2 7.3 64,779 326 89,293 585
20009 . . . . . . . . . . . . . . . 108,209 100 8.4 8.9 8.6 12.5 17.1 13.4 16.3 7.4 7.3 66,248 343 90,142 584
199910. . . . . . . . . . . . . . . 106,434 100 8.3 9.0 8.7 12.6 16.9 13.5 16.3 7.3 7.3 66,385 510 89,289 762
1998. . . . . . . . . . . . . . . . 103,874 100 9.1 9.3 8.7 12.4 17.2 13.7 16.2 7.1 6.4 64,781 630 86,389 767
1997. . . . . . . . . . . . . . . . 102,528 100 9.4 9.5 9.3 12.2 17.9 13.3 15.9 6.5 6.0 62,484 475 83,907 772
1996. . . . . . . . . . . . . . . . 101,018 100 9.7 9.9 9.4 12.6 17.8 13.4 15.8 6.0 5.5 61,225 508 81,289 749
199511. . . . . . . . . . . . . . . 99,627 100 9.7 10.0 9.4 13.0 18.4 13.5 15.1 5.9 5.0 60,348 574 79,584 717
199412. . . . . . . . . . . . . . . 98,990 100 10.5 10.1 9.6 13.1 18.1 13.0 15.0 5.6 4.9 58,515 439 78,228 692
199313. . . . . . . . . . . . . . . 97,107 100 10.9 10.1 9.4 13.5 18.3 13.0 14.5 5.7 4.5 57,843 445 76,704 682
199214. . . . . . . . . . . . . . . 96,426 100 10.8 10.4 9.4 13.2 18.4 13.7 14.7 5.5 4.0 58,153 453 73,726 509
1991. . . . . . . . . . . . . . . . 95,669 100 10.6 9.9 9.1 13.6 18.7 13.9 14.7 5.6 4.0 58,607 464 73,773 499
1990. . . . . . . . . . . . . . . . 94,312 100 10.2 9.6 9.0 13.0 19.5 13.7 15.0 5.5 4.4 60,370 507 75,411 524
Footnotes provided at end of table.
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 327 of 801
U.S. Census Bureau
Table A-2.
Households by Total Money Income, Race, and Hispanic Origin of Householder: 1967 to 2021—Con.
(Income in 2021 dollars, adjusted using the R-CPI-U-RS. Households as of March of the following year. Information on confidentiality protection, sampling error,
nonsampling error, and definitions is available at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>)
Median income Mean income
Percent distribution
U.S. Census Bureau
Race and Hispanic (dollars) (dollars)
origin of householder Number $15,000 $25,000 $35,000 $50,000 $75,000 $100,000 $150,000
and year (thou- Under to to to to to to to $200,000 Margin of Margin of
sands) Total $15,000 $24,999 $34,999 $49,999 $74,999 $99,999 $149,999 $199,999 and over Estimate error1 (±) Estimate error1 (±)
1989. . . . . . . . . . . . . . . . 93,347 100 9.8 9.4 9.5 12.7 18.9 14.0 15.5 5.8 4.5 61,153 553 77,261 553
1988. . . . . . . . . . . . . . . . 92,830 100 10.5 9.4 9.2 12.9 18.5 14.6 15.3 5.4 4.3 60,115 483 75,112 552
198715. . . . . . . . . . . . . . . 91,124 100 10.8 9.5 9.3 12.9 18.7 14.3 15.2 5.3 3.9 59,624 463 74,149 501
1986. . . . . . . . . . . . . . . . 89,479 100 11.1 9.6 9.1 13.5 18.7 14.4 14.8 5.1 3.7 58,920 502 72,793 487
198516. . . . . . . . . . . . . . . 88,458 100 11.1 10.0 9.6 13.7 19.5 14.0 14.3 4.6 3.2 56,871 507 69,990 456
198417. . . . . . . . . . . . . . . 86,789 100 11.1 10.5 9.8 13.9 19.6 13.7 14.0 4.5 3.0 55,828 418 68,403 414
1983. . . . . . . . . . . . . . . . 85,407 100 11.4 10.4 10.3 14.1 19.7 14.0 13.3 4.0 2.6 54,182 405 65,897 405
1982. . . . . . . . . . . . . . . . 83,918 100 11.7 10.5 10.1 14.0 20.4 13.9 13.1 3.8 2.6 54,564 405 65,758 400
1981. . . . . . . . . . . . . . . . 83,527 100 11.5 10.3 10.7 13.8 20.0 14.3 13.4 3.8 2.2 54,713 472 65,363 392
1980. . . . . . . . . . . . . . . . 82,368 100 11.0 10.4 10.2 13.7 20.2 15.0 13.4 4.0 2.2 55,596 470 66,122 398
197918. . . . . . . . . . . . . . . 80,776 100 10.8 9.9 9.6 13.7 20.1 15.3 13.9 4.1 2.6 57,462 448 68,259 425
1978. . . . . . . . . . . . . . . . 77,330 100 10.5 10.3 9.7 13.7 20.1 15.5 13.9 3.9 2.4 57,572 384 67,761 428
1977. . . . . . . . . . . . . . . . 76,030 100 10.8 11.0 9.7 13.9 20.6 14.9 13.5 3.3 2.2 55,427 343 65,751 329
197619. . . . . . . . . . . . . . . 74,142 100 11.0 10.6 10.1 13.9 21.2 15.3 12.7 3.2 2.0 55,078 336 64,786 329
197520. . . . . . . . . . . . . . . 72,867 100 11.2 10.9 10.0 14.2 21.7 14.8 12.4 2.9 1.8 54,180 363 63,266 325
197420, 21. . . . . . . . . . . . . 71,163 100 10.8 10.2 9.7 14.4 21.7 15.3 12.5 3.3 2.0 55,636 351 65,062 335
1973. . . . . . . . . . . . . . . . 69,859 100 10.7 10.4 9.1 13.4 21.5 15.7 13.4 3.6 2.3 57,456 360 66,447 333
197222. . . . . . . . . . . . . . . 68,251 100 11.4 10.0 9.6 13.7 21.8 15.6 12.5 3.3 2.2 56,319 353 65,548 334
197123. . . . . . . . . . . . . . . 66,676 100 12.4 9.7 9.8 14.7 23.2 14.6 11.3 2.8 1.6 54,006 344 62,111 325
1970. . . . . . . . . . . . . . . . 64,778 100 12.4 9.6 9.4 14.5 23.5 14.9 11.3 2.8 1.7 54,536 329 62,448 329
1969. . . . . . . . . . . . . . . . 63,401 100 12.3 9.3 9.2 14.6 23.8 15.5 11.1 2.7 1.6 54,962 334 62,530 323
1968. . . . . . . . . . . . . . . . 62,214 100 12.5 9.7 9.6 15.5 24.3 14.9 9.9 2.2 1.3 52,992 315 59,953 315
196724. . . . . . . . . . . . . . . 60,813 100 13.8 10.0 9.0 17.1 24.6 13.3 8.7 2.1 1.4 50,803 304 56,820 304
WHITE ALONE25
2021. . . . . . . . . . . . . . . . 102,057 100 8.2 7.6 7.6 10.8 16.2 12.3 16.6 8.6 12.1 74,262 912 105,804 1,183
20202 . . . . . . . . . . . . . . . 100,931 100 7.5 7.7 7.9 10.9 16.1 12.7 16.5 8.8 11.8 74,978 771 105,209 1,250
2019. . . . . . . . . . . . . . . . 100,568 100 7.2 7.1 7.5 11.1 16.3 12.5 17.0 9.1 12.2 76,519 848 107,812 1,264
2018. . . . . . . . . . . . . . . . 100,528 100 7.8 7.8 8.1 11.3 16.6 13.3 16.4 8.3 10.4 72,229 697 101,366 1,115
20173 . . . . . . . . . . . . . . . 100,113 100 8.0 8.1 8.3 11.6 16.0 13.0 16.0 8.2 10.8 71,658 931 101,153 1,167
2017. . . . . . . . . . . . . . . . 100,065 100 8.0 8.3 8.3 11.5 15.8 13.1 16.2 8.5 10.3 72,144 756 99,067 1,095
2016. . . . . . . . . . . . . . . . 99,400 100 8.3 8.0 8.4 11.7 16.8 12.5 16.5 8.2 9.7 69,840 620 97,485 992
2015. . . . . . . . . . . . . . . . 99,313 100 8.1 8.8 9.2 11.7 16.2 12.5 16.6 7.9 8.9 68,740 717 94,033 884
2014. . . . . . . . . . . . . . . . 98,679 100 9.2 9.3 9.2 11.9 16.8 12.4 15.5 7.5 8.4 65,144 669 90,374 986
20134 . . . . . . . . . . . . . . . 98,807 100 9.0 9.6 9.0 11.6 16.6 13.0 15.2 7.5 8.5 66,106 991 90,663 1,453
20135 . . . . . . . . . . . . . . . 97,774 100 8.8 9.6 9.2 12.0 17.1 13.5 14.9 7.4 7.4 64,372 814 88,350 1,043
2012. . . . . . . . . . . . . . . . 97,705 100 8.7 9.9 9.1 12.8 16.7 13.0 15.3 7.2 7.4 63,492 747 87,976 902
Case 1:23-cv-03424-TWT Document 59-4 Filed 08/31/23 Page 328 of 801
2011. . . . . . . . . . . . . . . . 96,964 100 8.9 9.3 9.0 13.2 17.3 12.7 15.1 7.2 7.4 63,036 447 87,903 838
20106 . . . . . . . . . . . . . . . 96,306 100 8.8 9.8 8.8 12.8 16.9 12.6 15.8 7.3 7.3 64,394 518 87,685 830
Footnotes provided at end of table.
Income in the United States: 2021 17
Table A-2.
Households by Total Money Income, Race, and Hispanic Origin of Householder: 1967 to 2021—Con.
(Income in 2021 dollars, adjusted using the R-CPI-U-RS. Households as of March of the following year. Information on confidentiality protection, sampling error,
nonsampling error, and definitions is available at <https://www2.census.gov/programs-surveys/cps/techdocs/cpsmar22.pdf>)
Median income Mean income
Percent distribution
Race and Hispanic (dollars) (dollars)
origin of householder Number $15,000 $25,000 $35,000 $50,000 $75,000 $100,000 $150,000
and year (thou- Under to to to to to to to $200,000 Margin of Margin of
sands) Total $15,000 $24,999 $34,999 $49,999 $74,999 $99,999 $149,999 $199,999 and over Estimate error1 (±) Estimate error1 (±)
20097 . . . . . . . . . . . . . . . 95,489 100 7.8 9.0 9.4 12.3 17.3 13.2 16.0 7.5 7.5 65,649 321 89,299 566
2008. . . . . . . . . . . . . . . . 95,297 100 7.9 9.0 9.2 12.2 16.8 13.4 16.4 7.5 7.5 65,990 315 89,805 569
2007. . . . . . . . . . . . . . . . 95,112 100 7.5 9.0 8.7 11.8 17.2 13.1 16.6 8.0 8.0 68,266 332 92,128 577
2006. . . . . . . . . . . . . . . . 94,705 100 7.6 8.5 8.6 12.6 17.4 13.0 16.6 7.5 8.2 68,260 326 93,091 638
2005. . . . . . . . . . . . . . . . 93,588 100 7.8 8.5 9.2 12.1 17.2 13.3 16.6 7.4 7.9 67,525 485 91,735 625
18 Income in the United States: 2021
Full text truncated for size; the complete document is in the PDF above.File and source
- File
- American_Alliance_for_Equal_Rights_v._Fearless_Fund_Management_LLC_FILING_RESTRI__gand_1-23-cv-03424__doc59_att4__gov.uscourts.gand.318833.59.4.pdf
- Size
- 25,442,389 bytes
- SHA-256
- 467c08c2e1d2cb860962afac42638a1c043410bb888b0f2990927471810d8ab5
- Original
- storage.courtlistener.com