2020 - The Silent Majority-Private U.S. Firms and Financial Reporting Choices PDF

DOI: 10.1111/1475-679X.
12306
Journal of Accounting Research
Vol. No. xxxx 2020
Printed in U.S.A.
The Silent Majority: Private U.S.

Firms and Financial Reporting
Choices
P E T R O L I S O W S K Y∗ A N D M I C H A E L M I N N I S†
Received 12 April 2018; accepted 27 March 2020
ABSTRACT
This study uses a comprehensive panel of tax returns to examine the finan-
cial reporting choices of medium-to-large private U.S. firms, a setting that
controls over $9 trillion in capital, vastly outnumbers public U.S. firms across
all industries, yet has no financial reporting mandates. We find that nearly
two-thirds of these firms do not produce audited GAAP financial statements.
∗ Questrom School of Business, Boston University and Norwegian Center for Taxation ;
† University of Chicago Booth School of Business.
Accepted by Mark Lang. We thank Mary Barth, Philip Berger, Michael Donohoe, Merle
Erickson, Christian Leuz, Nemit Shroff, Doug Skinner, Jennifer Tucker, Ross Watts, Joe
Weber, Teri Yohn, and participants at the 2017 SEC-NYU Dialogue on Securities Market
Regulation Conference, University of Amsterdam, Chicago, Illinois at Urbana-Champaign,
Illinois at Chicago, Lancaster, Florida, Michigan, Michigan State, Missouri, Penn State, Vrije
Universiteit, and Yale, the FASB/PCC Summer 2014 meeting, the 2014 University of Min-
nesota Empirical Conference, the 2014 American Accounting Association Annual Meeting,
and the 2014 FARS conference for comments. This paper was previously titled, “Financial
Reporting Choices of U.S. Private Firms: Large-Sample Analysis of GAAP and Audit Use” and
“Which Private Firms Follow GAAP and Why?” Lisowsky was approved access to tax return
information through a contractual agreement with the Internal Revenue Service subject to a
nondisclosure agreement. All statistics are presented in the aggregate in accordance with IRS
disclosure rules. Minnis is a member of the Private Company Council (PCC). Any opinions
are those of the authors and do not necessarily reflect the views of the IRS or the PCC. Minnis
gratefully acknowledges support from the ARAMARK Faculty Research Fund at the University
of Chicago Booth School of Business. An online appendix to this paper can be downloaded at
http://research.chicagobooth.edu/arc/journal-of-accounting-research/online-supplements.
1

C University of Chicago on behalf of the Accounting Research Center, 2020
2 P. LISOWSKY AND M. MINNIS
Guided by an agency theory framework, we find that size, ownership disper-

sion, external debt, and trade credit are positively associated with the choice
to produce audited GAAP financial statements, while asset tangibility, age,
and internal debt are generally negatively related to this choice. Our findings
reveal that (1) equity capital and trade credit exhibit significant explanatory
power, suggesting that the primary focus in the literature on debt is too nar-
row; (2) firm youth, growth, and R&D are positively associated with audited
GAAP reporting, reflecting important monitoring roles of financial report-
ing; and (3) many firms violate standard explanations for financial reporting
choices and substantial unexplained heterogeneity in financial reporting re-
mains. We conclude by identifying opportunities for future research.
JEL codes: M41; M44; M49

Keywords: audit; private firms; accounting choice; financial reporting;
capital allocation
1. Introduction
A significant challenge in accounting research is observing the costs and
benefits of firms’ financial reporting choices. At the most basic level, U.S.
firms are required by the Securities and Exchange Commission (SEC) to
produce audited financial statements in accordance with U.S. GAAP as a
condition for raising publicly traded equity or debt. Moreover, many coun-
tries outside the United States require firms—both public and private—to
prepare audited financial statements in accordance with accounting stan-
dards, such as International Financial Reporting Standards (IFRS). As such,
it is difficult for researchers to study why firms produce audited GAAP fi-
nancial statements because regulation typically masks firms’ endogenous
choices to do so. However, one setting—U.S. firms without publicly traded
securities—has essentially no reporting mandates and thus serves as a fruit-
ful setting to better understand which firms produce audited GAAP finan-
cial statements (Allee and Yohn [2009]). We use a comprehensive panel
data set of confidential business tax returns from the Internal Revenue Ser-
vice (IRS) to understand the propensity of audited GAAP financial state-
ment production of medium-to-large private U.S. firms and to examine the
factors associated with these choices.
Understanding firms’ decisions to produce audited GAAP financial state-
ments is important for several reasons. First, because of the lack of a reg-
ulatory mandate, private firms will only choose to produce audited GAAP
statements if the benefits exceed the costs. This allows us to examine char-
acteristics of firms that presumably derive the most benefits from producing
audited financial statements—and as importantly, can point to mechanisms
that substitute for (and thus reduce the benefit of) such statements. For
example, academics have claimed it should be taken as “given that lenders
[are] the main users of private company financial statements” (Bradshaw
et al. [2014, p. 183]; emphasis added), or that GAAP accounting is less
useful to firms with more intangible, knowledge-based assets (Lev and Gu
PRIVATE U.S. FIRMS AND FINANCIAL REPORTING CHOICES 3
[2016]). We assess these claims by examining the features of firms that en-
dogenously choose to produce audited GAAP statements.
Second, the Financial Accounting Foundation (FAF) recently launched
the Private Company Council (PCC) under the premise that GAAP was not
serving the needs of private firms. In doing so, however, little empirical
analysis has been conducted to understand which firms actually produce
GAAP statements. A baseline understanding of the characteristics of which
firms do so can inform standard setters as GAAP is adjusted to be less costly
and more beneficial to private companies.
Third, beyond understanding the factors related to accounting use, an
important development in the U.S. economy over the last 20 years has been
the steady growth of private firms. Estimates suggest that the number of
private firms increased by more than 10% from 1996 to 2016, whereas the
number of public firms decreased by almost 50% during this same period
(Doidge et al. [2017], Wursthorn and Zuckerman [2018]; U.S. Census Bu-
reau).1 Although there are many causes for this trend, our general under-
standing of this growing part of the economy—including financial report-
ing decisions—is lacking, especially outside of small U.S. businesses. Our
study therefore provides not only a baseline understanding of the use of ac-
counting in medium-to-large private firms in particular, but also enhances
our understanding of the nature of this setting overall.
We use confidential tax return data on medium-to-large private U.S. firms
in our analysis. Since 2008, the IRS has required firms with at least $10
million in assets to report on Schedule M-3 the set of accounting stan-
dards they follow for financial reporting, and whether their financial state-
ments are audited. We find that although the firms in our sample are not
small—each has at least $10 million in assets—they are numerous with
about 70,000 each year. Using conservative assumptions, we estimate that in
2010, nonfinancial private U.S. firms outnumber public firms in Compustat
by eighteen-to-one; this ratio remains significant at three-to-one even after
conditioning on revenues of at least $100 million.2 Moreover, private firms
are more numerous than public firms across all industries, again, even after
conditioning on only the larger firms in our sample. Collectively, the firms
in our sample deploy over $9 trillion in capital. As a result, the data illustrate
1 We define a private firm as one that has neither a publicly traded security nor a require-
ment to file reports with a financial market regulator, such as the SEC (Minnis and Shroff
[2017]). The number of public companies is sourced from CRSP, while the number of private
companies is the number of firms with at least 20 employees (following Doidge et al. [2017])
as reported by the Census Bureau’s Business Dynamics Statistics public data set through 2014.
2 These figures exclude firms from the Financial, Insurance, and Real Estate industries and
other firms with data issues. We describe our sample and data validation tests in section3 and
the online appendix. The ratios of private to public firms are based on 2010 data. The number
of public firms is sourced from Compustat and includes firms with at least $10 million in assets;
this figure was 3,885 in 2010. We have 70,425 firms in the IRS data in 2010. We stress that the
IRS data do not provide a comprehensive view of all private firms, but only those with at least
$10 million of assets.
that the term “private company” is far from synonymous with “small busi-
ness.”3 In short, larger private firms control significant sums of U.S. capital
and far outnumber public firms; because they do not publicly report their
financial statements, we describe them as the “silent majority.”
We begin our analysis of financial reporting practices by tabulating high-
level statistics. We find that 37% of medium-to-large private firms choose to
produce audited GAAP financial statements. More interestingly, even after
conditioning on firms having both ownership dispersion and external debt,
the audited GAAP rate increases only marginally to 40%. That is, we find
many firms with multiple owners and millions of dollars of debt do not
produce audited GAAP statements, while many smaller, single-owner firms
without debt do. These statistics reflect a vast and heterogeneous market
for financial reporting in which firms weigh the costs and benefits of ac-
counting choices in the absence of regulatory mandates. In short, the pro-
duction of audited GAAP financial statements is not a necessary condition
for external financing.
We use agency and information theories as frameworks to better un-
derstand the heterogeneity in financial reporting practices of private U.S.
firms. These frameworks posit that firms maximize value and produce au-
dited GAAP financial statements only when the benefits of doing so exceed
the costs (Watts [1977], Holthausen and Leftwich [1983], Watts and Zim-
merman [1986]). A primary benefit of high quality–verified financial state-
ments is to reduce agency costs and information asymmetry arising from
contracting arrangements between firms and outside parties, such as capi-
tal providers and suppliers (e.g., Armstrong, Guay, and Weber [2010]). As
such, we empirically model private firms’ financial statement production as
a function of proxies for these costs.
The panel structure of the IRS data allows us to examine variation in
financial reporting in both the cross-section and time series. In the cross-
section, we find that firms produce audited GAAP financial statements con-
sistent with theory: larger firms with more ownership dispersion and debt
are significantly more likely to produce audited GAAP financial statements.
But we also uncover other significant dimensions: firms with growth oppor-
tunities, trade credit, and those investing in intangibles (both book and
R&D based) are also more likely to have audited GAAP statements. By
contrast, firms that are older (i.e., potentially reflecting relationships or
reputation-based interaction with capital providers), with greater “inside
3 We note that our descriptive facts offer a contrasting view of private firms relative to prior
research focusing on small businesses. For example, an unfortunately titled paper, “What do
Private Firms Look Like?” (Asker, Farre-Mensa, and Ljungqvist [2011]) has been cited as pro-
viding descriptive statistics of what private firms look like. However, as the authors of that paper
discuss, their sample of firms excludes larger private firms. This caveat is missed in the litera-
ture, which cites that paper as describing the population of private firms (see, e.g., Bradshaw
et al. [2014, p. 180]).
debt” (i.e., debt provided by owners), and with more physical assets (e.g.,
land) are less likely to produce audited GAAP statements.
In the time series, we find that firms raising equity or increasing own-
ership dispersion are more likely to begin producing audited GAAP state-
ments. These results are particularly pronounced in young, high-growth
firms raising outside equity. Interestingly, we find that firms increasing their
level of outside debt, or moving from no debt to some positive level of debt,
do not exhibit any higher or lower likelihood of initiating audited GAAP
statements, regardless of firm age. However, we are cautious about making
strong inferences based on a lack of results related to changes in debt be-
cause our data lack granular detail on specific debt contracts where results
may manifest.
In addition to considering which factors are associated with audited
GAAP financial reporting, the data reveal several notable quantitative facts.
First, although size is the most important determinant, we find that the
rate of audited GAAP production does not exceed 50% until the sample in-
cludes firms with revenues around $100 million. Thus, a substantial portion
of middle market firms does not produce audited GAAP statements. Sec-
ond, although the data reveal that debt is strongly positively related to the
production of audited GAAP statements, we also find that audited GAAP
statements are far from necessary to attract even substantive amounts of
debt. The propensity to produce audited GAAP statements of firms with
zero debt is 38%; conditioning on at least $5 million of debt, this propen-
sity only increases to 49%.4 These findings reinforce that alternative mech-
anisms, such as relationships and collateral, likely play an important role
to accessing debt capital, even in the setting of medium-to-large firms, and
suggest relaxing the assertions in Bradshaw et al. [2014] that lenders are
the main users of financial statements. Finally, the findings highlight the
overlooked importance of equity transactions and trade credit. For exam-
ple, we find that firms adding new equity investors initiate audited GAAP
statements 65% of the time compared to 44% for firms without equity trans-
actions. Given the massive and increasing prevalence of private equity in-
vesting, the role of financial statement choices in equity capital allocation
is an important area for future research to consider.
In sum, our paper sheds light on the prevalence and determinants of fi-
nancial reporting in medium-to-large private U.S. firms. The key findings
are threefold. First, the findings suggest that the debt contracting focus
in the literature is too narrow as other characteristics, such as equity capital
and trade credit, exhibit significant explanatory power. Second, firm youth,
4 We selected $5 million in debt because Slee [2011] suggests this level of debt as a common
threshold for borrowers to provide audited financial statements to banks. Anecdotally, we have
experienced multiple conversations in the course of writing this paper suggesting individuals’
expectations are that firms have to provide audited GAAP statements to their banks if firms
are borrowing more than $5 million in debt. These assumptions are not consistent with the
data.
growth, and R&D are positively associated with audited GAAP reporting,
suggesting important monitoring roles of financial reporting. Third, de-
spite the number of characteristics we examine, many firms violate stan-
dard explanations for financial reporting choices. For example, some firms
with both ownership dispersion and high levels of debt do not produce au-
dited GAAP statements, leaving substantial unexplained heterogeneity to
explore in future research.
Our paper makes several contributions to the accounting literature. First,
we extend prior research investigating the uses and effects of financial
reporting in private firms, generally (e.g., Blackwell, Noland, and Win-
ters [1998], Allee and Yohn [2009], Minnis [2011], Dedman, Kausar, and
Lennox [2014], Kausar, Shroff, and White [2016], Minnis and Sutherland
[2017]), by focusing on larger private U.S. firms, specifically. Our extension
is important because these firms have greater complexity and agency con-
cerns than small U.S. businesses (Cendrowski et al. [2012]) and, as our
results suggest, the findings from the small business literature may not be
generalizable to the larger firms (e.g., Allee and Yohn [2009]). Further-
more, conditional on firms with at least $10 million in assets, our study cov-
ers the population of tax filing firms in the United States, avoiding sample
selection concerns (Minnis [2011], Badertscher et al. [2018]) or interpre-
tation issues in surveys (e.g., Allee and Yohn [2009]). The panel structure
of the IRS data also enables us to examine how firm characteristics vary with
changes in accounting choices. In all, we significantly broaden our scope
and knowledge about the market for private firm financial reporting.
Second, our results suggest that an almost exclusive focus in the litera-
ture on how debt explains audited GAAP use needs to be broadened (e.g.,
see references and discussion in Armstrong, Guay, and Weber [2010]). We
find that equity and trade credit-based explanations are powerful, yet un-
derexplored factors (Costello [2013]). Moreover, our results reveal that the
use of audited GAAP statements is far from a corner solution, as thousands
of firms do not fit standard “expectations.” Our results on equity and age,
as well as trade credit, suggest that the literature should at least reassess
the conventional view taking debt as the primary (or at least only) driver of
financial reporting choices in this setting.
Third, because our setting is similar in economic magnitude to public
firms, both cumulatively and at the firm level, it is interesting to contrast
our findings to those in research on public firms. Specifically, research on
public firms presents evidence that financial statements are less relevant for
knowledge- and intangibles-based, growth opportunity firms (e.g., Lev and
Gu [2016], Govindarajan, Rajgopal, and Srivastava [2018]). However, our
findings suggest that audited GAAP financial statements play a particularly
important role when monitoring and information channels evident in pub-
lic markets do not exist (e.g., Grossman [1976]). In firms with less physi-
cal assets, greater growth opportunities, and more intangible assets (such
as human capital, software, and R&D), our results suggest highly verified
financial reports become more informative mechanisms for monitoring and,
in particular, equity capital allocation. These insights highlight a departure

in the ability to generalize inferences from public firms to private firms, but
in doing so, provide an explanation for why private firms produce audited
GAAP financial statements, even without a regulatory mandate.
Finally, our study contributes to help policy makers and researchers un-
derstand which factors are related to the net benefit of accounting as stan-
dard setters are currently altering the reporting landscape for private firms.
For example, the FAF established a new board in 2012—the PCC—that
proposes alternatives within GAAP for private firms. However, at least two
recent academic reviews highlight the current lack of evidence about pri-
vate firms’ use of accounting and call for more research (see Botosan et al.
[2006], Bradshaw et al. [2014]). We answer this call by providing detailed
analysis on the financial reporting choices of medium-to-large private U.S.
firms, which are more representative of the bulk of U.S. nonpublic eco-
nomic activity.
We note two caveats to our study. First, it is largely descriptive. Our goal
is not to identify causal factors of financial statement characteristics per se,
but rather to uncover new evidence on which firms produce audited GAAP
reports and their associated characteristics, especially as they relate to cap-
ital allocation. Second, although the tax return data set is comprehensive,
it lacks specificity along various dimensions. For example, we do not have
access to specific debt arrangements and do not know if audits are classified
as unqualified or qualified. Such information would be helpful to generate
more specific tests of which mechanisms are substitutes for financial state-
ments and which accounting standards appear to be most onerous, respec-
tively. Although data on private U.S. firms are difficult to obtain, our study
highlights that the setting has sufficiently interesting variation worth inves-
tigating, even with smaller and more targeted data sets with more specific
variables.
2. Setting, Theoretical Framework, and Prior Research

2.1 SETTING
Private U.S. firms are not required by regulatory mandate to publish or
prepare any type of financial report other than a tax return, which itself re-
mains confidential. Instead, they can select the type of financial reporting
along two dimensions: accounting standards and assurance level. GAAP is
the set of standards public U.S. firms are required to follow; however, al-
ternative rules including tax (rules set by the IRS), cash (measuring the
amount of cash paid and collected during the period), IFRS (rules set
by the International Accounting Standards Board), statutory (e.g., rules
specific to particular industries, such as insurance), various hybrid meth-
ods, and others are available to private U.S. firms. The second dimension
is whether to have the financial statements audited by an independent
accountant.5 Technically, these two dimensions are independent choices.

That is, any financial statements prepared on a basis of accounting with a
standard set of rules can be audited, and firms can prepare their financial
reports without an audit. Because private U.S. firms face no public report-
ing or audit requirements, the choices they make along either dimension
are the result of market forces dictating the supply and demand for finan-
cial reports.
2.2 THEORETICAL FRAMEWORK
Our key objective is to examine the extent and conditions under which
private firms prepare audited GAAP financial statements. We use agency
and information theories to frame our analyses (Watts and Zimmerman
[1986]). The foundation of these theories is that, in equilibrium and in the
absence of regulatory fiat, firms will only produce audited GAAP financial
statements when the benefits exceed the costs. The theories are rooted in
the idea that frictions created by information asymmetries between the firm
and outside parties, or among the outside parties, create demand for firms
to provide verifiable financial information using a prescribed set of rules
on which contracts can be written. Audited GAAP financial statements pro-
vide such verified, and therefore contractible, financial information that
enables outside parties to screen and monitor the firm.
Jensen and Meckling [1976, p. 338] motivate a role for financial state-
ments within an agency framework.6 They state, “it would pay [the man-
ager] to agree in advance to incur the cost of providing such reports and to
have their accuracy testified to by an independent outside auditor.” Watts
[1977] and Watts and Zimmerman [1978] apply this framework and derive
multiple hypotheses for the production of financial statements and audit-
ing. Their fundamental prediction is that financial statement production is
an increasing function of agency costs, that is, it increases in outside equity
owners, lenders, and suppliers subject to the costs of production.7
More recent literature also considers inside debtholders—owners or
managers who lend to the firm. In contrast to outside debtholders, in-
side debtholders have substantially less information disadvantage. More-
over, owners and managers holding the firm’s debt align their interests
more with outside debtholders (e.g., Sundaram and Yermack [2007], Ed-
mans and Liu [2011], Anantharaman, Fang, and Gong [2013]). Thus, the
5 An audit is the highest level of assurance that an independent accountant can provide.
Other report types include reviews (which provide negative assurance) and compilations
(which provide no assurance regarding the financial report). The tax returns we analyze do
not report whether the financial statements have been reviewed or compiled.
6 See Benston [1985], Botosan et al. [2006], Kothari, Ramanna, and Skinner [2010], and
references therein for additional discussion.

7 Our list of transacting parties with the firm is far from complete. Examples of other trans-
acting parties with the firm who may demand high quality financial reports are employees,
customers, and governments. Unfortunately, our data do not provide adequate proxies to ex-
amine the interactions with these parties, so we leave such a focus to future research.
presence of inside debt suggests reduced demand for high-quality financial

reporting.
The demand for audited financial reports within an agency framework
rests on informational frictions between the agent and the principal. If all
parties are perfectly informed about the firm, then the demand for high-
quality financial reporting is nonexistent, even in the presence of an agency
relationship. Therefore, theory suggests that the demand for audited GAAP
statements is conditional not only on the number of transacting parties
with the firm, but also on the degree of informational frictions. Based on
prior research, we postulate that these frictions derive from various firm
characteristics. Asset intangibility is potentially one source of information
asymmetry. Although one role of audited GAAP financial statements is to
test the existence and (book) value of assets-in-place, if an outside party
can easily observe and quantify a firm’s assets, then the demand for verified
financial reports could be lower. For example, firms with investments in
real property (e.g., land) are less likely to provide financial reports to their
lenders (Minnis and Sutherland [2017]) because land is easily observable
and difficult for management to misdirect.
Related to, but distinct from, the concept of asset tangibility, firms with
intangible assets in the form of growth opportunities and research and de-
velopment generate two possible outcomes. On the one hand, to the ex-
tent audited GAAP financial statements are designed to verify and quantify
assets-in-place, one expectation is that the presence of growth opportuni-
ties and research projects suggest lower demand for financial reporting.
Indeed, one extreme version of this rationale is suggested by Lev and Gu
[2016], who claim that GAAP reporting is not useful for an innovation-
based economy. On the other hand, audited GAAP statements do more
than simply verify assets-in-place. Auditors test transactions, and GAAP
statements require, for example, reporting material expenditures on R&D
activities. Therefore, an alternative view is that audited GAAP reporting is
a type of outsourced internal control function—that is, as a “stewardship”
or “monitoring” role—which is particularly important precisely when a firm
has limited tangible assets. Moreover, outside capital providers may be par-
ticularly concerned how managers spend their funds when a firm has sub-
stantial growth opportunities and R&D. For example, in the event of a bad
outcome, they may be concerned if the R&D was truly not successful or if
managers wasted the resources on non-R&D expenditures. Audited GAAP
statements are one mechanism of control, which may be particularly useful
when information asymmetry is high in firms with high growth and R&D.
The level of firm performance could also mediate the demand for high-
quality financial reports. Firms that are just slightly profitable or loss-
making have incentives to take at least two dubious actions that prof-
itable firms might not. First, low-profit firms have incentives to avoid
covenant violations by showing slight profits (e.g., Coppens and Peek
[2005]). Audited GAAP statements have lower levels of manipulation and
higher accrual quality to mitigate this concern (Minnis [2011]). Second, if
operations are not profitable, the residual claimants have incentives to

make high variance, “bet-the-firm” types of risky investments, thus steal-
ing value from debtholders. Audited GAAP reporting can ameliorate these
incentives to ensure proper reporting and timely evidence of investment
returns, which is more useful in low-profitability firms.
A final potential source of information asymmetry across firms that
we consider is firms’ reputation and relationships with outside capital
providers. A long line of literature shows that as firms build their reputa-
tions (e.g., as a firm that only invests in positive net present value (NPV)
or low variance projects), both the amount of information asymmetry be-
tween the firm and transacting parties and the need for transacting parties
to monitor the firm decline (Diamond [1991]). For example, Petersen and
Rajan [1994] show that small businesses with fewer, longer lasting lenders
have access to more credit. Minnis and Sutherland [2017] show that small
firms with longer relationships with their lenders are less likely to provide
those lenders with financial reports. Moreover, established, high-reputation
firms are more likely to have better internal controls, reducing the need for
high-quality externally verified reports (Ge and McVay [2005]). Therefore,
we expect older firms that are more likely to have reputations to protect,
relationships with banks, and more sophisticated controls, to be less likely
to produce audited GAAP financial reports.
2.3 PRIOR LITERATURE
Prior research has taken two approaches to understand the factors un-
derlying the demand for accounting absent regulatory mandates. The first
approach examines firms’ accounting choices prior to the development of
securities regulation. For example, Watts and Zimmerman [1983] examine
English merchant guilds and early corporations dating to the 16th century.
Chow [1982] and Barton and Waymire [2004] examine factors associated
with the choice of obtaining an audit in publicly traded U.S. firms prior
to the 1933–1934 U.S. Securities Acts. The second approach uses surveys
or more recently available data sources on private U.S. businesses. For ex-
ample, Abdel-khalik [1993] surveys 134 private companies about their de-
mand for auditing. Minnis [2011] and Badertscher et al. [2018] use data
from Sageworks, Inc., to study a sample of small and medium private U.S.
firms sourced from accounting firms to examine the benefits of audits
and reviews. Lisowsky, Minnis, and Sutherland [2017] use a very large data
set of private company financial statements collected by banks to examine
how credit cycles affect banks’ use of financial reports. Berger, Minnis, and
Sutherland [2017], using the same data set, show how bank expertise me-
diates bank demand for private company financial statements.
Most closely related to our paper is Allee and Yohn [2009] (A&Y, here-
after). A&Y use the 2003 Survey of Small Business Finance (SSBF) to
provide one of the first systematic studies on the accounting choices of
small private businesses in the United States. They find 80% of the small
business survey respondents do not use or produce any type of financial
statement. Of the 20% that do produce financial statements, only 27%

have them audited, while 49% follow accrual accounting, which could
encompass any number of noncash basis accounting approaches, such as
GAAP, tax, IFRS, or other hybrid methods.8 The firms included in their
study are small by design, given the objective and title of the survey, with
the average firm reporting assets of approximately $350,000. Moreover, be-
cause A&Y are restricted to a small sample in a single year, they are unable
to perform across-industry or panel analyses and have little power to gen-
erate statistically significant inferences (e.g., basic theoretical predictions
on size and debt are not significantly related to financial statement pro-
duction in their table 6). Despite these empirical challenges, A&Y provide
an important first examination of accounting practices for very small U.S.
private firms.
Although these studies generally find that financial reporting sophistica-
tion, such as obtaining an audit, is increasing in agency costs, the inferences
are not always clear and likely not generalizable to larger firms or more re-
cent periods (Leftwich [2004]). Importantly, a major difference between
our study and previous studies is the setting. Although larger private U.S.
firms are similar to small U.S. businesses in that they lack an accounting
mandate, they are in many ways more similar to public firms in terms of
size, economics, and complexity. U.S. government data show that the ma-
jority of private U.S. firm economic activity is not in small businesses: firms
with less than $1 million in receipts or 20 employees—which make up 90%
of the SSBF data set—collect less than 5% of total receipts in the economy
and pay only about 14% of overall wages (U.S. Census Bureau, Statistics of
U.S. Businesses, Bureau of Labor Statistics). In addition, 45% of SSBF firms
are sole proprietorships (i.e., not separate legal entities from their sole own-
ers) and 90% have less than $1 million in assets (in fact, nearly one-third of
firms have less than $25,000 in assets). As such, it is unclear to what extent
agency-related issues, such as separation of ownership and control, gener-
alize to more complex firms with more substantive legal formation issues,
broader dispersion of ownership, and multifaceted capital considerations.9
8 Thus, that study suggests that only about 5% (i.e., 27% of 20%) of small firms have audits.
Unfortunately, the SSBF survey question does not ask what type of financial statements the
firms actually produce or whether they are audited; instead it asks what records the survey re-
spondent used to answer the survey questions. Specifically, the survey asks the respondent, “Do
you have records available to help you answer questions about the firm’s income, expenses,
and balance sheet, such as tax records, statements, worksheets, or any other records?” The
follow-up question is “What records are you using?” It is unclear whether the question refers
to the records being used to answer the survey, or the records being used in the business itself.
A&Y report that of the 4,004 firms responding, 790 responded as using financial statements,
but 1,682 responded as having used memory. Therefore, it is uncertain whether the findings
in A&Y refer to the actual production of financial statements, or simply the use of financial
statements in answering the SSBF survey.
9 Although studying small, entrepreneurial firms is certainly important for a variety of rea-
sons, our point is simply that “private firms” include not only small businesses, but also large,
Furthermore, despite this research, firms’ financial reporting practices

and underlying reasons for those practices are still the subject of intense
debate in the literature (Bradshaw et al. [2014]). Our study aims to analyze
the setting of larger U.S. private firms and critically investigate their finan-
cial reporting choices to reveal new inferences beyond those established in
the small business or public company settings.
3. Data
We use panel data from the population of private U.S. firms with at least
$10 million in assets. These data are sourced from Schedule M-3 of the
U.S. federal income tax returns of Subchapter C corporations (Form 1120),
Subchapter S corporations (Form 1120S), and partnerships (Form 1065).
Since 2004, Schedule M-3 requires companies with assets of $10 million
or more to disclose to the IRS whether the financial statements have been
audited by an independent accountant, and since 2008, the IRS further
requires firms to report on Schedule M-3 which set of accounting standards
they use for financial reporting. We use these Schedule M-3 disclosures to
identify whether a firm undergoes a financial statement audit and what set
of accounting standards it follows for financial reporting.10
For the purposes of this study, the IRS generously provided access to one
of the authors all Schedules M-3 for entities filing Forms 1120, 1120S, and
1065 for fiscal years 2008–2010 at the consolidated U.S. parent level. By def-
inition, because filers of Schedule M-3 report assets of $10 million or more,
our sample begins with medium-to-large firms in the population; there are
no IRS disclosure requirements on financial statement audits or GAAP
usage for firms with assets less than $10 million. Despite this truncation,
table 1 reports that the initial data set provides 630,076 nonpublicly
traded firm-year observations across all three entity types.11 We remove
complex organizations with significant agency concerns and external capital demands that
are typically omitted from previously used data sets. Prior research also investigates two other
settings. First, non-U.S. settings used to investigate the accounting practices of private firms
include Dedman, Kausar, and Lennox [2014], Hope, Thomas, and Vyas [2011], Lennox and
Pittman [2011], and Kausar, Shroff, and White [2016]. However, they are limited to firms with
differing mandates regarding accounting and disclosure (for both public and private firms)
relative to the United States. Thus, the cost–benefit trade-off of financial reporting decisions
is likely different between the U.S. and non-U.S. settings, so the results may not be compara-
ble. Second, in a recent working paper, Gaver, Mason, and Utke [2019] examine the financial
reporting of private investment funds and find at most 53% of those funds prepare audited
GAAP statements. That study adds to research on financial reporting outside of public firms
and the debt contracting perspective by examining a specific equity fund setting.
10 We take several steps to ensure the internal and external validity of the data; see online
appendix A. In particular, we find that the summary statistics of audited GAAP production of
the firms in our tax return sample are quite similar to two other independently collected data
sets, reassuring our approach (see the appendix). Also see the online appendix for additional
details about Schedule M-3, including the 2010 version of the schedule.
11 Specifically, we omit any firm answering “Yes” to having filed with the SEC or having a
publicly traded stock.

TABLE 1
Sample Selection
IRS Form Source:
1120 1120S 1065 Total

All parent-level nonpublicly traded consolidated
Tax returns fiscal years 2008—2010 118,723 115,445 395,908 630,076
With total assets ࣙ $10Ma
Less: Financial and real estate firmsb (45,340) (30,246) (317,792) (393,378)
Less: Foreign-owned firmsc (19,324) – (476) (19,800)
Population estimate test sample 54,059 85,199 77,640 216,898
Less: Tax return paper filers (28,152) (42,359) (54,977) (125,488)
Firm-level test sample (e-filers only) 25,907 42,840 22,663 91,410
This table reports the sample selection process. We begin with all tax returns (all Forms 1120 and 1065)
that also file Schedule M-3 provided by the IRS confidentially to one of the authors.
a
$10M or more in assets is the filing requirement for Schedule M-3, which is the only IRS form that
contains both financial statement standards and financial statement audit rate data required for this study.
b
We remove firms in NAICS Codes 52, 53, and 55.
c
We remove firms with ࣙ25% foreign ownership. Sources: Form 1120: Schedule K Line 7. Form 1065:
Schedule B Line 1e. 1120S firms cannot be foreign owned.
(1) financial firms (NAICS code 52) because their accounting and audit
choices are affected by regulation and (2) real estate and management
firms (NAICS codes 53 and 55) because there are many instances of firms
with significant assets, but no revenues or expenses, suggesting these are
nonoperating holding companies. We also drop firms with foreign owner-
ship greater than 25% because these entities may face alternative regulatory
regimes or may be subsidiaries of publicly traded foreign firms. This step
leaves a sample of 216,898 observations.12
The sample of 216,898 observations includes firms that file their tax re-
turns electronically (e-filers; n = 91,410) or on paper (paper filers; n =
125,488). One constraint of the IRS data is that the Schedule M-3 disclo-
sures on audit and GAAP use are available to us only for e-filers and not for
paper filers. We therefore use the data in two different ways. For any statis-
tic that considers the population, we use both paper and e-filers because
we have data on firm-level characteristics for both. However, for all anal-
yses that examine the relation between firm characteristics and account-
ing choices, we only use the 91,410 e-filing observations for which actual
12 By only using consolidated parent tax returns, and by eliminating financial, real estate,
and management firms, as well as all firms with substantial foreign ownership, we substan-
tially reduce the risk of double-counting entities and assets, for example, a private equity firm
(which would be considered a financial firm) owning a fund that invests in a private company.
Moreover, the author with confidential access did two hand-checks of the data: (1) he identi-
fied a sample of private companies from other data and confirmed they exist in the tax return
sample and (2) he examined a random sample of firms in the tax return data set to ensure
they are operating private companies. In addition, the list of company names was perused
to identify any systematic inclusion of shell companies (i.e., with unusual names) or other
oddities with the data. He found no evidence of inappropriate firms remaining in our sample.
reporting choices are disclosed. We describe our approach in detail in the

next section.
Before proceeding to the analysis, we note a few points about our ap-
proach and the data. Throughout the study, we classify the joint decision
to follow GAAP and receive an audit (i.e., “audited GAAP” financial state-
ments) as the key dependent variable. We do this for two reasons. First,
as suggested by Kothari, Ramanna, and Skinner [2010], we can only be
assured that a firm follows GAAP when it is audited, thus it is the joint
decision that results in “following GAAP.” Also, audited GAAP statements
are required of public firms and our setting provides a useful comparison
to these firms. Second, we do this for parsimony as it reduces the dimen-
sionality of the analyses reported. As a practical matter, inferences are very
similar if we consider both dimensions of the accounting decision, which
we report in the online appendix. Finally, with respect to the unit of obser-
vation in our study, we refer to a “firm” according to the parent-level tax
return provided to us by the IRS. However, we acknowledge there may be
cross or common ownership among firms, making the ultimate “boundary”
of the firm not conclusively defined in our data. Although this issue can also
affect publicly traded firms (e.g., Azar, Schmalz, and Tecu [2018]), we have
confirmed the nature of the data with the IRS and have conducted signifi-
cant checks to limit issues surrounding, for example, the double-counting
of firms (see the online appendix for further details).
4. Analysis
4.1 DESCRIPTIVE ANALYSIS OF PRIVATE FIRMS
As the comprehensive data on medium-to-large private U.S. firms we use
in our study are rare, we begin our analysis by providing descriptive statistics
of the firms in table 2. All variables are defined in the appendix. The first six
columns report statistics for the entire sample (i.e., including paper filers),
and the last two columns report means for the paper and e-filing samples,
respectively. We winsorize continuous variables at the top and bottom 1%.13
Firms in our sample are substantially larger than prior studies of private
firms and control vastly more resources. The mean (median) firm has $53.7
($24.3) million in revenue and $57.7 ($19.4) million in total assets.14 For
2010, we find that the (unwinsorized) cumulative total assets controlled
by the private firms in our sample exceed $9 trillion (untabulated). The
statistics emphasize that these firms are much different from the private
firms examined in prior studies, for example, average total assets are less
than $0.4 million in Allee and Yohn [2009] and $6.5 million in Minnis
[2011].
13 Per IRS disclosure rules, we are not permitted to report the descriptive statistics at any
finer partitions to prevent revealing any firm-specific information.

14 We define Revenue as total income before all deductions and cost of goods sold—this data
item is most similar to Compustat’s “sale” variable, plus interest, rents, royalties, and gains.
TABLE 2
Descriptive Statistics
Population Paper Only E-File Only
n = 216,898 n = 125,488 n = 91,410
n Mean P25 Median P75 SD Mean Mean

Financial reporting choices
GAAP AUDIT 216,898 37% 0% 0% 100% 48% 32% 44%∗∗∗
GAAP 216,898 79% 100% 100% 100% 41% 75% 84%∗∗∗
AUDIT 216,898 38% 0% 0% 100% 49% 33% 45%∗∗∗
Firm characteristics
LOG REVENUE 216,898 2.92 1.80 3.23 4.06 1.65 2.58 3.39∗∗∗
REVENUE ($M) 216,898 53.67 5.06 24.25 56.90 94.03 39.95 72.50∗∗∗
LOG NUM OWNER 216,898 1.83 1.10 1.39 2.20 1.14 1.77 1.91∗∗∗
NUMBER OF OWNERS 216,898 14.49 2 3 8 29.27 12.97 16.58∗∗∗
NUMBER OF OWNERS (if < 101) 200,409 6.91 2 3 6 12.05 6.33 7.73∗∗∗
OWNER EQ 1 216,898 16% 0% 0% 0% 37% 15% 18%∗∗∗
OWNER EQ 2 216,898 24% 0% 0% 0% 43% 27% 20%∗∗∗
OWNER EQ 3to5 216,898 27% 0% 0% 100% 44% 28% 25%∗∗∗
OWNER EQ 6to10 216,898 12% 0% 0% 0% 32% 11% 12%∗∗∗
OWNER EQ 11to100 216,898 13% 0% 0% 0% 34% 12% 15%∗∗∗
OWNER EQ 101orMore 216,898 8% 0% 0% 0% 27% 7% 9%∗∗∗
LOG DEBT 216,898 1.38 0.00 1.21 2.42 1.40 1.22 1.59∗∗∗
(Continued)
PRIVATE U.S. FIRMS AND FINANCIAL REPORTING CHOICES
15
16
T A B L E 2—Continued
Population Paper Only E-File Only
n = 216,898 n = 125,488 n = 91,410
n Mean P25 Median P75 SD Mean Mean

DEBT ($M) 216,898 11.48 0.00 2.37 10.22 29.87 9.67 13.96∗∗∗
LOG DEBT FROM SH 139,258 0.23 0.00 0.00 0.00 0.57 0.24 0.22∗∗∗
DEBT FROM SH ($M) 139,258 1.40 0.00 0.00 0.00 51.66 1.37 1.42
LOG ACCT PAY 216,898 0.91 0.06 0.61 1.44 0.98 0.77 1.10∗∗∗
ACCOUNTS PAYABLE ($M) 216,898 3.88 0.06 0.85 3.24 9.93 3.07 4.79∗∗∗
PROFIT MGN 216,898 −0.01 −0.02 0.01 0.07 0.35 −0.01 −0.01
P. LISOWSKY AND M. MINNIS
LOSS 216,898 0.33 0 0 1 0.47 0.33 0.33

PE TO ASSETS 216,898 0.20 0.01 0.09 0.31 0.25 0.19 0.21∗∗∗
PE TO WAGES 216,898 4.60 0.00 0.34 2.18 15.40 4.43 4.83∗∗∗
LAND TO ASSETS 216,898 0.04 0.00 0.00 0.01 0.12 0.04 0.03∗∗∗
INTAN 216,898 0.46 0 0 1 0.50 0.41 0.53∗∗∗
GROWTH 122,448 0.01 −0.15 0.00 0.13 0.57 0.03 −0.01∗∗∗
LOG AGE 128,701 3.05 2.56 3.22 3.71 0.91 3.07 3.03∗∗∗
AGE (years) 128,701 28.13 12 24 40 20.92 28.27 28.00∗∗
LOG ASSETS 216,898 3.31 2.66 3.02 3.66 0.92 3.26 3.39∗∗∗
TOTAL ASSETS ($M) 216,898 57.72 13.23 19.42 38.02 165.42 54.76 61.78∗∗∗
C CORP 216,898 25% 0% 0% 0% 43% 22% 28%∗∗∗
This table reports the descriptive statistics for the variables used in all analyses. The paper filer GAAP and AUDIT statistics are estimates based on a propensity score match with
the e-filing firms as described in the online appendix. The n for NUMBER OF OWNERS(if < 101) for e-filers = 83,498 and for paper filers = 116,911. The n for LOG DEBT FROM SH
and DEBT FROM SH ($M) for e-filers = 70,511 and for paper filers = 68,747. The n for GROWTH = 46,388 for e-filers and 76,060 for paper filers. The n for LOG AGE = 68,580 for
e-filers and 60,121 for paper filers. See the appendix for variable definitions. ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.
FIG. 1.—Number of private and public firms and percentage of private firms with audited
GAAP financial statements by Compustat sales size quintile for the year 2010. This chart re-
ports the distribution of private firms (paper and e-file) partitioned by Compustat sales rev-
enue quintiles for the year 2010. The bar charts report the number of firms in each size
quintile and the line chart reports the percentage of private firms in that quintile that pro-
duce audited GAAP financial statements. The quintiles are based on the Compustat variable
“sale” for all nonfinancial U.S. firms in Compustat. The sizes of the private firms are based on
the variable Revenue as defined in the appendix. The financial reporting data (i.e., % Audited
GAAP) for the e-filing firms are as reported on Schedule M-3; estimates are used for the pa-
per filing firms as described in online appendix B. In comparing tax return data to Compustat
data, we note two important caveats. First, “Revenues” per the tax return are not an exact com-
parison to the “sale” variable in Compustat. Various differences in definitions, consolidation
rules, foreign versus domestic revenues all need to be considered. Second, Compustat does
not cover all firms that have SEC registration requirements—that is, it is not a comprehensive
sample of publicly traded firms. This chart is meant to demonstrate broad relative magnitudes
of the two settings rather than precise estimates.
Also in contrast to prior studies of private firms, the firms in our sam-
ple are much more similar to public U.S. firms. To illustrate this compari-
son, figure 1 plots the number of nonfinancial firms in both the IRS data
(paper and e-filers) and Compustat for the year 2010 by Compustat size
quintile (based on the variable “sale”). Although Compustat data are not
directly comparable to tax return data because of consolidation and other
accounting differences, this chart is simply intended to give a sense of the
relative magnitude of the two settings. As the chart indicates, there are sub-
stantially more private firms in the bottom three quintiles (total of 69,563)
compared to Compustat (total of 2,331), and the number of private firms
is similar in the fourth quintile (777 public firms versus 682 private firms).
These four quintiles represent firms with revenues up to $2.3 billion each. It
is not until the largest quintile (i.e., firms each with revenues greater than
$2.3 billion) that Compustat firms substantially outnumber private firms

(777 public firms versus 180 private firms). Although strong inferences
cannot be made comparing these two data sets, an important descriptive
takeaway is that, broadly speaking, many of the private firms in our study
are very comparable in magnitude to most public firms in Compustat, and
there are many more private than public firms.15 In all, we find that pri-
vate firms outnumber public firms across much of the public company size
distribution, except for the very largest firms.
The table 2 descriptive statistics also reveal that many firms are unprof-
itable and have negative growth, likely indicative of the economic condi-
tions during the sample period. Average taxable income scaled by total
income (PROFIT MGN) is –1.1%, but this distribution is skewed left as
the median firm has PROFIT MGN of 1.0% and 33% of firms report net
losses.16 The average firm in the sample that exists at least two years is
growing revenues by 1.0% per year (GROWTH), but there is wide varia-
tion; growth ranges from –14.7% at the 25th percentile to +12.8% at the
75th percentile.
We extract balance sheet variables from Schedule L. We first measure two
types of debt: external (provided by nonrelated parties) and internal (pro-
vided by owners). The mean (median) firm has almost $11.5 ($2.4) million
in external debt and $1.4 ($0) million of internal debt. Also, most firms
in the sample use trade credit. We use accounts payable to proxy for the
extent of financial reporting demand from trade creditors. The mean (me-
dian) accounts payable is $3.9 ($0.9) million.17 We measure capital-to-labor
intensity using the ratio of net plant and equipment (PE) per Schedule L
to total wages reported on page 1 of the firm’s tax return (PE TO WAGES).
The mean (median) firm has PE that is 4.6 (0.3) times the amount of wages.
We measure asset tangibility using (1) net PE as the share of total assets
(PE TO ASSETS), which equals 20% (9%) for the mean (median) firm; (2)
the ratio of land to total assets (LAND TO ASSETS), which equals 4% (0%)
for the mean (median) firm; and (3) INTAN, equal to 1 if the firm reports a
nonzero gross intangible assets balance, and 0 otherwise; we find that 46%
of firms report nonzero intangible assets.
15 If we condition that the private firm must be a C corporation, which is the most common
form of publicly traded firm, our distribution is as follows: 10,732 in Q1, 4,825 in Q2, 985
in Q3, 243 in Q4, and 66 in Q5, for a total of 16,851 private C corporations. Note that this
number far surpasses total public firms in Compustat, as well as includes some very large
private firms with revenues over $2.3 billion. Interestingly, more than half of the private firms
in all quintiles—even the largest two quintiles—are “pass through” entities (i.e., partnerships
and S corporations), not C corporations, suggesting that viewing pass through entities as only
“small businesses” is inaccurate.
16 “Taxable income” is referred to as “Ordinary Business Income” for pass-through entities.
We use the term “taxable income” for both types of entities for simplicity.
17 We use logged transformations in our regression models when measuring revenues
(LOG SIZE), external debt (LOG DEBT), internal debt (LOG DEBT FROM SH), and trade
credit (LOG ACCT PAY).
We caution that, unlike the income statement variables, which follow tax
rules for all firms regardless of the financial accounting basis they choose,
the balance sheet variables are reported using the book basis of account-
ing and could be mechanically related to firms’ accounting choices. For
example, if a firm uses the tax basis of accounting, assets such as equip-
ment (inventory) may be understated (overstated) relative to a firm that
uses GAAP. This is a primary reason why we use revenues rather than total
assets as our primary measure of size—revenues (as well as profit margin
and growth) are measured on the tax basis for all firms. As we discuss in the
next section, we also address this issue by using different specifications that
exclude balance sheet variables or substitute income statement variables for
balance sheet variables.
The tax returns also provide measures for ownership, age, and organi-
zational form. The IRS data contain the number of firm owners, but firms
are only required to report this field if the number of owners is 100 or
less. Thus, the IRS reporting requirement limits our ability to identify the
number of owners if they exceed 100 as only a few firms voluntarily dis-
close this number; otherwise the disclosure is left blank.18 Therefore, we
handle the NUMBER OF OWNERS variable three ways: (1) we set missing
values to equal 101 owners; (2) we exclude firms reporting more than
100 owners or missing values; and (3) we use indicator variables for vari-
ous ownership levels, including if it exceeds 100 or is missing. The own-
ership variable is right-skewed, as the mean NUMBER OF OWNERS is 14.5
including all available observations (after setting the missing values to 101)
and 6.9 after excluding firms with more than 100 owners or missing val-
ues. The variables reveal that 40% of our sample has just one or two
owners (OWNER EQ 1 and OWNER EQ 2), 27% have three to five owners
(OWNER EQ 3to5), and 8% have more than 100 owners or report missing
values (OWNER EQ 101orMore).19
18 The term “Owners” collectively indicates shareholders (for corporations), partners (for
partnerships), or members (for limited liability companies). Note that the data set has even
more right tail skewness than reported in the descriptive statistics because, as with all our
continuous variables, we winsorize the number of owners—when data are available—at the
99th percentile. In fact, the data contain 233 observations that (voluntarily) report more than
500 owners (the 99th percentile is 176 owners). These firms are professional services part-
nerships, which are exempt from SEC rules that require public financial reporting for firms
with 500 or more owners. Overall, when we use continuous owner data, we log the data to
create LOG NUM OWNER. We tabulate robustness tests in the online appendix and find that
the specification of the number of owners does not affect our inferences.
19 Unfortunately, the IRS data do not allow us to identify owners, so we cannot study owner-
ship variation in greater detail. In particular, given the recent increase in private equity (PE)
financing, one might be interested in the role of PE firms as owners. PitchBook, a relatively
comprehensive data set for North American PE activity (see Brown et al. [2015.]) reports that
about 5,900 companies were PE-backed in 2017 (PitchBook [2017]). It further estimates that
the number of PE-backed firms grew by 5% per year since 2009, suggesting that about 4,000
companies were PE-sponsored in 2009—and this number does not filter real estate or financial
firms as we have done in our study. PitchBook also reports that much of the PE deal activity
In terms of age, only corporations (Forms 1120 and 1120S) report the
year of incorporation. The mean (median) corporation is 28 (24) years
old (AGE). We also report the share of our sample firms that are C corpora-
tions, the dominant form of organization for public firms. We find that only
25% of our sample firms are C corporations. The remaining firms are split
between S corporations (Form 1120S) or partnerships and limited liability
companies (Form 1065).
In table 2, we also report the population estimate of audited GAAP finan-
cial statement production. However, recall from section 3 that we have all
variables for the paper filing firms except their financial reporting choices.
Therefore, we estimate the population rate of audited GAAP production
using e-filing firms as counterfactual estimates for paper filing firms (see
the online appendix for details). As table 2 reports, we estimate that 37%
of the population of medium-to-large private U.S. firms produces audited
GAAP financial statements (GAAP AUDIT). This joint decision breaks down
to 79% of firms using GAAP (GAAP) and 38% having their financial state-
ments audited (AUDIT).20 Given the differences in size, ownership disper-
sion, and levels of debt between paper filers and e-filers, it is not surprising
that the estimated rate of audited GAAP production is lower for paper filers
(32%) compared to the actual rate for e-filers (44%). Figure 1 plots the per-
centage of private firms that produce audited GAAP financial statements in
each Compustat size quintile. Although this shows a strong positive relation
between firm size and GAAP audit use, it also highlights that even many
large firms do not produce audited GAAP financial statements.
As a final descriptive analysis, we present the distribution of (e-filing)
firm-years and propensity to produce audited GAAP statements by industry.
Table 3 sorts industries from highest to lowest rates of audited GAAP pro-
duction. Columns 1 and 2 report the number of firm-years and percentage
of sample, respectively.21 Column 3 reports the percentage of firm-years
producing audited GAAP statements. Information has the highest rate at
62%, followed closely by Utilities, Manufacturing, Services, and Healthcare.
These results suggest that as the U.S. economy continues its evolution to-
ward industries with substantial human and intangible capital (i.e., infor-
mation and services industries), audited GAAP reporting could be an im-
portant tool to verify, value, and monitor these firms. Consistent with this
occurs in the sub-$100 million market, suggesting that the activity is not concentrated solely
among the largest firms in our sample. Therefore, a likely upper bound on the share of PE-
owned firms in our population does not exceed 5.7% (or 4,000/70,000 private firms in 2009,
per our table 3).
20 We tabulate all accounting choices separately for e-filing firms in the online appendix. Of
the 16% of e-filing firms not claiming GAAP as their set of accounting standards, 12.5% use
tax basis, while 0.2% use IFRS, 0.6% use statutory, and 2.9% use an alternative set of standards
(e.g., cash basis). Note that the audit rate is slightly higher than the joint GAAP-and-audit rate
because a very small portion of firms have their non-GAAP financial statements audited.
21 The online appendix compares the distribution of firm-years across industries to other
commonly used data sets.

TABLE 3
Industry Analysis of Audited GAAP Financial Statement Production
(1) (2) (3)
NAICS Industry Name Total Number Total Percentage GAAP Audit
51 Information 3,607 4% 62%
22 Utilities 862 1% 57%
31–33 Manufacturing 19,907 22% 56%
54, 56 Prof., Scientific, Tech., 10,180 11% 50%
Admin, Waste Mgt.
Services
61–62 Education and Healthcare 3,770 4% 48%
42 Wholesale Trade 11,606 13% 47%
48–49 Transportation and 2,931 3% 46%
Warehousing
81 Other and unclassified 1,265 1% 44%
23 Construction 12,745 14% 44%
21 Mining 3,040 3% 32%
71–72 Arts, Entertainment, 6,311 7% 30%
Recreation,
Accommodation, and
Food Service
44–45 Retail Trade 11,907 13% 29%
11 Agriculture, Forestry, 3,279 4% 25%
Fishing and Hunting
Total 91,410 100% 44%
This table presents a summary of the use of GAAP and auditing by industry for our sample of e-filer firm-
years for all years 2008–2010. Column 1 reports the number of firm-years; Column 2 reports the percentage
of firm-years across the industries; Column 3 reports the percentage of firms within the industry producing
audited GAAP financial statements.
view, table 3 also reveals that industries with significant tangible assets (e.g.,
Mining and Agriculture) and cash-based businesses (e.g., Food Service and
Retail Trade) have the lowest rates of audited GAAP statements. We note
that within Agriculture—the industry with the lowest audited GAAP rate of
25%—almost half of the firms have less than $5 million in revenues, yet
have at least $10 million in assets, reinforcing the notion that agriculture
not only includes many small operations but that this is also an asset inten-
sive industry.
The main takeaways from the descriptive analysis of the economics of
large private firms are as follows: They are far more numerous than public
companies, operate across all major industries, cumulatively deploy trillions
of dollars of capital, and have significant financial reporting heterogeneity.
We now examine more rigorously how this financial reporting heterogene-
ity relates to firm characteristics and capital structure.
4.2 DETERMINANTS OF PRODUCING AUDITED GAAP FINANCIAL STATEMENTS

4.2.1. Cross-Sectional Analyses. We examine the factors related to private
firms’ audited GAAP financial reporting using across-firm, across-industry,
and within-firm changes tests. As a reminder, from this point forward we
only consider e-filing firms for which we can link actual (rather than esti-
mated) financial reporting choices to firm-level data.
We begin by using two-way sorts to examine three constructs most com-
monly viewed as associated with financial statement production—firm size
(revenues), separation of ownership and control (number of owners), and
external debt. Table 4, panel A, partitions firm-years into size and own-
ership dispersion cells, and panel B partitions firm-years into size and
debt cells. Each cell reports the number of firm-years and proportion
of those firm-years with audited GAAP financial statements. We find that
the rate of audited GAAP financial statement production is clearly in-
creasing in all three variables, both unconditionally (examining the out-
side rows and columns) and conditionally on the other variable (exam-
ining the interior rows and columns). As expected, the upper left cells
have the lowest audited GAAP rates while the bottom right cells have the
highest.
Two additional insights emerge. First, although the relation between au-
dited GAAP rates and each of the variables is generally monotonic, it is
not linear. The relation between audited GAAP rates and ownership dis-
persion is the most apparent in firms with between 1 and 10 owners, and
the relation flattens out considerably after firms reach about 20 owners.
Increases in ownership dispersion on the intensive margin after this point
seem to have little relation with audited GAAP rates. Debt has a similarly
monotonic yet concave relation, with an inflection point near $25 million.
Data in table 4 and figure 1 (which reports the proportion of firms pro-
ducing audited GAAP statements by Compustat size quintile) show that
size has a concave relation that begins to flatten near $200 million in
revenues.
Second, the results from table 4 do not support corner solutions or text-
book “rules of thumb” about thresholds regarding the production of au-
dited GAAP financial statements.22 Although size, ownership dispersion,
and debt have strongly positive relations with audited GAAP use, many
firms do not meet standard expectations. For instance, focusing on firm-
years with both more than one owner and nonzero external debt results in a
conditional propensity to produce audited GAAP statements of only 40%,
that is, less than half the firms with ownership dispersion and outside debt.
Focusing specifically on debt, the propensity to produce audited GAAP
statements of firms with zero debt is 38% (as seen at the bottom of the
first column); after conditioning on firms with at least $5 million of debt,
this propensity increases to only 49% (i.e., again less than half the sample).
This difference suggests that many firms with relatively high levels of exter-
nal debt do not produce audited GAAP statements while many firms with
no debt do produce audited GAAP statements. The former result is consis-
tent with the idea that lenders to private companies—even large ones—rely
22 For example, in the textbook Private Capital Markets, Slee [2011] suggests $5 million in
debt as a common threshold for borrowers to provide audited financial statements to banks.
TABLE 4
Two-Way Sorts of Audited GAAP Financial Statement Rates by Size, Ownership Dispersion, and Debt
Panel A: Size and ownership dispersion partitions
Number of Owners
1 2–5 6–10 11–15 16–20 21–25 26–30 31–35 36–40 41–45 46–50 51–100 >100 Total
Total revenue ($ millions) <20 4,538 14,675 3,452 1,265 669 439 362 296 211 183 177 725 2,724 29,716
25% 19% 27% 34% 36% 40% 46% 38% 45% 46% 43% 51% 50% 27%
20–50 5,469 12,742 3,370 1,191 589 385 272 209 135 124 128 470 1,697 26,781
42% 37% 48% 53% 57% 56% 55% 59% 59% 57% 61% 69% 64% 44%
50–100 3,405 8,275 2,218 792 409 296 235 179 126 72 78 419 1,205 17,709
46% 46% 59% 67% 68% 73% 66% 73% 67% 54% 64% 60% 67% 52%
100–500 2,466 5,711 1,972 842 559 278 217 186 167 128 110 463 1,666 14,765
61% 61% 71% 81% 77% 82% 85% 83% 84% 80% 87% 77% 66% 67%
>500 337 584 263 124 124 59 61 46 38 33 31 119 620 2,439
69% 70% 84% 84% 85% 93% 85% 96% 89% 91% 81% 86% 60% 73%
Total 16,215 41,987 11,275 4,214 2,350 1,457 1,147 916 677 540 524 2,196 7,912 91,410
42% 36% 49% 56% 59% 61% 62% 62% 64% 61% 62% 64% 60% 44%
(Continued)
PRIVATE U.S. FIRMS AND FINANCIAL REPORTING CHOICES
23
24
Panel B: Size and debt partitions
Total Debt ($ millions)
0 0–5 5–10 10–15 15–20 20–25 25–30 30–35 35–40 40–45 45–50 50–100 >100 Total
Total revenue ($ millions) <20 10,493 7,206 4,849 3,014 1,503 786 486 322 214 155 128 361 199 29,716
25% 32% 27% 21% 23% 25% 28% 25% 29% 34% 38% 34% 39% 27%
20–50 6,047 10,094 5,591 2,253 907 480 316 206 166 120 76 356 169 26,781
41% 43% 39% 46% 59% 58% 68% 69% 66% 68% 74% 71% 49% 44%
50–100 3,678 4,959 3,290 2,258 1,112 611 407 242 170 143 114 451 274 17,709
49% 53% 47% 44% 52% 63% 64% 71% 71% 75% 82% 78% 67% 52%
100–500 3,122 2,714 1,513 1,192 1,038 815 630 450 385 290 262 1,168 1,186 14,765
58% 63% 64% 67% 66% 67% 71% 70% 74% 77% 79% 80% 79% 67%
>500 505 204 105 75 78 66 50 53 40 46 39 281 897 2,439
55% 77% 79% 79% 67% 83% 80% 68% 83% 67% 72% 81% 79% 73%
Total 23,845 25,177 15,348 8,792 4,638 2,758 1,889 1,273 975 754 619 2,617 2,725 91,410
38% 44% 40% 40% 48% 53% 58% 59% 63% 66% 70% 72% 73% 44%
This table partitions the full sample of e-filing firms (2008–2010) by the number of owners and sale revenues (panel A) and amount of debt and sales revenue (panel B). Each
cell reports the number of firms in each portfolio and the percentage of those firms that produce audited GAAP financial statements.
on alternative mechanisms to safeguard their loans, such as relationships

or appraisal reports (e.g., Cassar, Ittner, and Cavaluzzo [2015], Minnis and
Sutherland [2017]). The latter result suggests that private companies may
hire external auditors to serve in a control capacity over management be-
havior (Watts and Zimmerman [1983]) or demand high-quality financial
reports for managing the firm (e.g., Barrios, Lisowsky, and Minnis [2019]).
Although our data are too coarse to fully examine these possibilities, the
results demonstrate that while the expected relations between financial re-
porting and external capital emerge, the extent of the relation may not
be as large as expected, and the relation is nonlinear.23 Moreover, au-
dited GAAP statements are not a necessary condition for attracting external
capital.
To more critically examine the relation between firm characteristics and
audited GAAP production, we use the following linear probability model
for firm i in industry j at year t:

m
GAAP AUDIT i, j,t = βk FIRM CHARACTERISTIC i, j,t,k
k=0
j
+γi,t + λit + εi, j,t , (1)
where GAAP AUDIT is set to 1 if the firm produces audited GAAP financial
statements during the year (0 otherwise), k denotes the firm characteris-
j
tic, and γi,t and λit represent industry and year fixed effects, respectively.
Table 5 presents the results. We begin in column 1 with a baseline specifi-
cation that includes the variables for size, ownership dispersion, and exter-
nal debt, as well as proxies for trade credit (LOG ACCT PAY), performance
(PROFIT MGN and an indicator for negative profits, LOSS), and a variety of
asset tangibility variables (INTAN, PE TO ASSETS, and LAND TO ASSETS).
We also include indicators for whether the firm is a C corporation, and in-
dustry and year effects. The specification in column 1 is the most detailed
specification we can estimate without sample attrition related to the avail-
ability of additional variables.
In table 5, column 1, we find a significantly positive relation between au-
dited GAAP use and size, ownership dispersion, and external debt. These
results are consistent with inferences from table 4. Moreover, we find a sig-
nificantly positive relation between trade credit and audited GAAP produc-
tion, consistent with trade creditors serving as important outside transact-
ing parties demanding verified information (Costello [2013]). In addition,
23 Our data do not allow us to measure two prominent forms of equity and debt financing:
employee stock option plans and operating leases, respectively. The former is important be-
cause the Private Company Council has found that it is not uncommon for private companies
to issue stock-based compensation and yet struggle with the accounting for this form of com-
pensation (see https://www.fasb.org/pcc for documents related to this topic). The latter is
important because lessors typically have stronger repossession rights and therefore potentially
rely less on financial reporting mechanisms (see Sutherland [2018] for a discussion).
TABLE 5
Firm Characteristics of Audited GAAP Financial Statement Production: Firm Level
Dependent Variable: GAAP AUDIT
(1) (2) (3) (4) (5)

Asset Size Subsamples
ࣘ$25M >$25–$100M ࣙ$100M

LOG REVENUE 0.078∗∗∗ 0.090∗∗∗ 0.083∗∗∗ 0.095∗∗∗ 0.041∗∗∗
(7.18) (10.09) (8.12) (8.72) (3.21)
LOG NUM OWNER 0.031∗∗∗ 0.040∗∗∗ 0.047∗∗∗ 0.039∗∗∗ 0.025∗∗∗
(3.52) (12.35) (8.81) (9.26) (4.05)
LOG DEBT 0.029∗∗∗ 0.031∗∗∗ 0.025 0.041∗∗∗ 0.017∗∗∗
(3.68) (3.83) (2.23) (4.15) (3.65)
LOG ACCT PAY 0.051∗∗∗ 0.045∗∗∗ 0.053∗∗∗ 0.043∗∗∗ 0.026∗∗∗
(3.59) (4.03) (2.84) (3.60) (3.22)
−0.061 −0.069 −0.066∗∗∗ −0.069∗∗∗
∗∗ ∗∗
PROFIT MGN 0.053
(−2.37) (−2.77) (−2.09) (−2.28) (0.95)
−0.007 −0.002 −0.027∗
∗∗
LOSS 0.010 0.027
(−0.81) (1.17) (2.32) (−0.18) (−1.73)
INTAN 0.017 0.003 0.009 −0.002 0.000
(1.35) (0.30) (0.75) (−0.14) (0.01)
∗∗
PE TO ASSETS 0.024 0.056 0.005 0.098 0.245∗∗∗
(0.68) (1.58) (0.12) (2.18) (6.51)
PE TO WAGES −0.0003 −0.0003 −0.0001 −0.0006∗ −0.0006
(−1.57) (−1.08) (−0.39) (−1.69) (−1.12)
−0.148 −0.135 −0.197∗∗∗ −0.093 −0.174
∗∗∗ ∗∗∗
LAND TO ASSETS
(−3.99) (−3.27) (−3.30) (−1.48) (−1.03)
GROWTH 0.048∗∗∗ 0.029∗∗∗ 0.043∗∗∗ 0.066∗∗∗
(8.12) (2.87) (5.75) (5.88)
−0.046 −0.062∗∗∗ −0.052∗∗∗ −0.021∗
∗∗∗
LOG DEBT FROM SH
(−9.19) (−7.19) (−6.79) (−1.89)
LOG AGE −0.012∗∗∗ −0.015∗ −0.017∗∗∗ 0.004
(−2.39) (−1.73) (−2.84) (0.58)
C CORP 0.092∗∗∗ 0.068∗∗∗ 0.074∗∗∗ 0.070∗∗∗ 0.038∗
(3.52) (6.21) (5.69) (4.84) (1.90)
Industry FE? Yes Yes Yes Yes Yes
Year FE? Yes Yes Yes Yes Yes
Adjusted R2 19.9% 21.0% 14.4% 16.8% 12.3%
Observations 91,410 36,058 19,203 12,554 4,301
Where GAAP AUDIT = 1 40,623 18,437 7,188 7,866 3,383
This table reports firm-level linear probability model coefficient estimates in which the dependent vari-
able is GAAP AUDIT. All models use robust standard errors clustered by three-digit NAICS industry code.
Continuous variables are winsorized at the 1st and 99th percentile levels. Please see the appendix for all
variable definitions. ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.
performance (PROFIT MGN) and asset tangibility (LAND TO ASSETS)

have a significantly negative relation to audited GAAP production.24
24 In untabulated tests, we further investigate the interaction between land and debt by
interacting LOG DEBT with LAND TO ASSETS to the specification of table5, column 1. We
find a significantly negative coefficient on the interaction confirming the results in Minnis
In column 2 of table 5, we add variables for growth, inside debt, and

firm age. To do so, we must restrict our sample to firms with at least two
years of data (to calculate growth) and C and S corporations only (who
report inside debt and firm age). Restricting the sample to corporations
further ensures our results are robust to excluding partnerships, and thus
are potentially more comparable to public firms. First, GROWTH has a
significantly positive coefficient, consistent with high-growth firms finding
audited GAAP statements beneficial potentially because of future capital
demands or because they are more useful for managing growth. Second,
inside debt is significantly and negatively related to audited GAAP pro-
duction, consistent with inside capital providers having lower information
asymmetry and more closely aligned incentives with outside debtholders.
Third, the significantly negative sign on AGE is consistent with mature firms
having stronger reputations and/or relationships that reduce the need for
audited GAAP statements. In untabulated nonparametric tests, we find that
the relation between AGE and GAAP AUDIT to be monotonically negative,
though flattens after about four years of firm age.
In columns 3–5 of table 5, we partition the sample conditional on firm
size to examine how our results might compare to prior literature that uses
smaller businesses or large, public firms. We note that the overall tenor of
the results is similar across the partitions, with some degree of coefficient
magnitude differences and interesting exceptions. First, the negative rela-
tion with performance and age are only significant in the smaller two size
partitions. Second, the coefficient on PE is positive in the larger two size
partitions. Although theory does not shape our priors about differences
across size partitions, at a minimum the results within the smallest firms
column 3 suggest that theoretical expectations manifest despite prior lit-
erature not finding empirical evidence for these standard theories in small
firms (e.g., Allee and Yohn [2009]). The findings also provide comfort that
the results are not driven by the largest firms in the sample.25
and Sutherland [2017] that the presence of real estate mitigates the usefulness of financial
reporting in debt contracting. We also interact LOG DEBT with PE TO ASSETS and find a
negative, but statistically insignificant, coefficient.
25 At this point, it is worth contrasting in more detail our findings with prior research, in
particular Allee and Yohn’s [2009] (A&Y) table 6, which is the most similar analysis to our
table 5. A&Y examine the sophistication of accounting with two different dependent variables
in their table 6: a count variable ranging from 0 to 3, representing company prepared, com-
piled, reviewed, and audited financial statements, respectively, and an indicator variable for
whether or not the firm reports using an accrual basis of accounting. The first item to note
from A&Y is the lack of significant results for most variables. For example, firm size (total
assets)—the most significant variable in our study and others (e.g., Minnis [2011])—is not sig-
nificantly related to whether the firm has an audit or follows accrual accounting. In addition,
variables indicating new equity, the extent of leverage, the amount of sales growth are also in-
significant in A&Y. Ownership dispersion and trade credit is only significant in their accruals
regression. In contrast to the results in our table 5, firm age in A&Y is positively associated with
accounting sophistication. Ultimately, the lack of significance and mixed results of the A&Y
We conduct a variety of robustness tests, which we tabulate in the on-

line appendix. First, we reconsider the specification of our dichotomous
dependent variable, GAAP AUDIT. Recognizing that there is a continuum
of accounting choices, we use an ordered logit model in which the depen-
dent variable is coded 0 for unaudited, non-GAAP firms; 1 for unaudited
GAAP firms; or 2 for audited GAAP firms. Our inferences are unchanged.26
We also (1) control for size using assets (instead of revenues), scale debt by
assets (thus creating a leverage variable), and scale trade payables by total
assets; (2) eliminate firms with less than $5 million in revenues; (3) use in-
terest deductions instead of debt; (4) exclude firms with missing ownership
data; and (5) use semiparametric indicator (rather than logged) variables
for the number of owners. Again, all inferences are similar. Finally, we re-
estimate the regression from table 5, column 1, by year to examine the
stability of the coefficient estimates over time and find similar results each
year, with the exception of the loss indicator, which is only significant in
2008. Collectively, the cross-sectional tests are robust to a variety of specifi-
cation checks and stable across industry and time.
A substantial portion of a firm’s characteristics is inherent to its indus-
try. Recall from table 3 that the propensity for a firm to produce audited
GAAP statements varies widely across industries. We further note that in-
cluding industry fixed effects increases the adjusted R2 in column 1 of ta-
ble 5 from 15.7% to 19.9%, or a 27% increase (untabulated). We therefore
examine cross-sectional variation in audited GAAP production by regress-
ing the estimated industry fixed effect coefficients from column 1 of table 5
on a variety of industry-level variables that we derive from public compa-
nies using Compustat. This industry-level approach using public firm data
has two advantages. First, we can examine characteristics that we cannot
with the IRS data. For example, our tax return data do not report certain
measures we would like to have, such as research and development and
growth opportunities (e.g., market-to-book). Second, one might be con-
cerned about a mechanical relation between financial reporting choices
and variable measurement. For example, the choice to use GAAP instead of
tax-basis accounting affects the way variables are measured. Using Compus-
tat data, we construct independent variables to proxy for profitability (ROA
and ROA DISP), debt (LEV), asset tangibility (PPE TO ASSETS and IN-
TAN TO ASSETS), and growth opportunities (GROWTH, R&D TO SALES,
and MTB). We use decile ranks of the variables to facilitate comparisons
table is likely driven in part by their small sample size (790 observations), noisy data points
generated by survey data, and the very different (i.e., small business) setting.
26 We also re-estimate table 5 using a logit model and find similar results. As an exploratory
analysis, we also re-estimate the regression for each industry and tabulate the results in the
online appendix. The results are generally consistent across industry (e.g., the coefficients on
size, ownership dispersion, and debt are almost always of the same sign and statistical signifi-
cance).
across the variables and include all three-digit NAICS industries with at least
three public firms.27
We first note that, in contrast to the firm-level results of table 5, the
industry-level results in column 1 of table 6 reveal little association be-
tween industry-level profitability and the use of audited GAAP financial
statements.28 However, we find that industries with high levels of profitabil-
ity dispersion (ROA DISP) have higher rates of audited GAAP use. We infer
from the dispersion results that audited GAAP statements are more useful
when earnings are less certain.
Next, we see no relation between industry-level leverage and private
firms’ audited GAAP production. However, this result is likely driven by
omitted variable issues, as industries with more leverage typically have
more collateralizable assets. Note that this result highlights a central point,
though: in debt contracting for private firms, alternative mechanisms can
substitute for financial statements, diminishing their benefit. In fact, the co-
efficient on industry-level PPE TO ASSETS is significantly negative, which
suggests that the use of audited GAAP statements is muted in the pres-
ence of significant tangible assets that are more easily verified upon visual
inspection.
Finally, we find evidence that industries with high levels of booked
intangible assets (INTAN TO ASSETS), unrecognized intangibles
(R&D TO SALES), and growth opportunities (GROWTH) have signif-
icantly higher rates of audited GAAP statements. MTB is positive, but not
significant at standard levels. These findings are particularly intriguing
given the repeated calls that GAAP’s treatment of intangible assets does
not meet a cost–benefit threshold (FAF [2011]) and that the way GAAP
treats research and knowledge-based firms is not informative to financial
statement users (Lev and Gu [2016]).29 We speculate that growth op-
portunity firms benefit from audited GAAP statements in two ways: (1)
they anticipate raising external capital in the future, and audited GAAP
statements facilitate this process; and (2) financial statement auditors can
27 Specifically, to transform the variable into decile ranks, we create deciles for each variable,
subtract 1 from the decile rank value and divide the result by 9. As such, the variable values
range from 0 to 1.
28 Interestingly, the linear results mask a nonlinear relation. In untabulated results, when
we include a second-order ROA term, the first-order coefficient is significantly negative and
the second-order coefficient is significantly positive. Differentiating with respect to ROA, we
find a minimum point outside of the ROA distribution, such that the relation between indus-
try profitability and audited GAAP statements is monotonically negative through the sample
distribution of ROA, with the relation becoming weaker with profitability.
29 An important caveat—and a detail that future research can consider—is that we are un-
able to measure qualified audit opinions. We suspect that intangible assets are common causes
of qualified opinions. Moreover, firms engaging in activities creating booked intangibles (e.g.,
acquisitions) are likely to benefit from audited GAAP statements for reasons other than their
intangible assets. Our results simply find that booked intangibles are positively rather than
negatively related to audited GAAP statements. We also caution against drawing causal infer-
ences.
TABLE 6
Firm Characteristics of Audited GAAP Financial Statement Production: Industry Level
30
Dependent Variable: Industry Fixed Effect from table 5, column 1
(1) (2) (3) (4) (5) (6) (7) (8) (9)

ROA −0.037 −0.017
(−0.88) (−0.36)
ROA DISP 0.073∗ 0.045
(1.92) (0.83)
LEV −0.036 0.002
(−0.75) (0.04)
∗∗
PPE TO ASSETS −0.084 −0.003
(−2.46) (−0.06)
INTAN TO ASSETS 0.117∗∗∗ 0.056
(3.31) (1.15)
∗∗
GROWTH 0.087 0.054
(2.11) (1.48)
∗∗
R&D TO SALES 0.140∗∗∗ 0.098
(4.60) (2.62)
MTB 0.060 −0.006
(1.36) (−0.11)
∗∗ ∗∗
CONSTANT 0.010 −0.045 0.009 0.033 −0.066∗∗∗ −0.051∗ −0.074∗∗∗ −0.038 −0.119
(0.43) (−2.16) (0.36) (1.61) (−2.79) (−1.96) (−3.65) (−1.56) (−2.49)
Adjusted R2 −0.1% 4.1% −0.3% 5.9% 12.9% 6.4% 23.1% 2.2% 23.0%
Observations 64 64 64 64 64 64 64 64 64
This table analyzes industry-level rates of audited GAAP financial statement production using a linear probability model regression for the 64 nonfinancial three-digit NAICS
industries with at least three firms with sufficient data to calculate all variables. The dependent variable is the estimated fixed effect coefficient for each industry from the firm-level
regression results in table 5, column 1. All independent variables are industry-based measures for publicly held firms sourced from Compustat or CRSP. ROA is net income scaled
by total assets for the average firm in the industry. ROA DISP is the within-industry interquartile range of ROA. LEV is the sum of total short- and long-term debt scaled by total
assets for the average firm in the industry. PPE TO ASSETS is the percentage of total assets composed of net property, plant, and equipment for the average firm in the industry.
INTAN TO ASSETS is intangible assets divided by total assets for the average firm in the industry. GROWTH is the percentage revenue growth for the average firm in the industry.
R&D TO SALES is research & development scaled by total sales for the average firm in the industry. MTB is the market value of assets (market value of equity plus book value of total
debt) divided by the book value of total assets for the average firm in the industry. To facilitate comparability across the coefficients, all variables are placed into deciles and scaled
between [0, 1]. Heteroskedasticity robust t-statistics reported below the coefficient estimates. Continuous variables are winsorized at the 1st and 99th percentile levels. ∗∗∗ , ∗∗ , and ∗
denote significance at the 1%, 5%, and 10% levels, respectively.
serve as outsourced internal auditors, allowing external capital providers to

ensure their investments are not being misallocated by management. We
view these empirical results—which stand in contrast to assertions made
about the diminishing benefits of accounting in public firms, especially
among knowledge-intensive firms—as motivating further research.30 For
completeness, in the final column of table 6 we include all industry
variables. Results become weaker, but multicollinearity is a clear issue with
so few observations.31 In all, the industry-level results complement the
firm-level inferences by showing that industry asset tangibility is negatively
related to audited GAAP financial statement production.
4.2.2. Time-Series Analyses. Our final set of tests exploit the panel struc-
ture of the data to identify the factors associated with changes in the firm-
level decisions to produce audited GAAP financial statements. We start by
keeping only those firms that are included in the IRS data set in all three
years, 2008–2010, and then examine a firm’s decision to begin producing
audited GAAP statements. The dependent variable BEGIN GAAP AUDIT
equals 1 if the firm begins producing audited GAAP statements in 2009,
and 0 otherwise. The control firms are those that do not produce audited
GAAP financial statements (either because they do not follow GAAP or do
not have their financial statements audited, where the latter reason is the
most prominent) in any of the three years. The independent variables in-
clude both the initial (i.e., 2008) levels of each of the variables, as well as
the changes (measured from 2008 to 2010, thus we measure the change sur-
rounding the change decision), making the unit of observation a firm, not
a firm-year. Approximately 5% of firms choose to begin producing audited
GAAP statements in 2009 (268 of 6,541).32
Results are reported in table 7. Consistent with our earlier cross-
sectional tests, in column 1 we find that larger firms with more own-
ers, debt, and trade credit are more likely to initiate audited GAAP
30 Note that our proxies identify both types of intangibles identified by Skinner [2011]:
(1) intellectual capital and lack of physical substance, and (2) identifiable and recognized.
Although our finding that firms in industries with more intangibles are more likely to have
audited GAAP statements suggests that the current approach to accounting for intangibles is
net cost-beneficial for these firms, we are cautious in making this inference. Nevertheless, our
results suggest that the private firm U.S. setting could be a fruitful setting to investigate the
costs and benefits of intangible accounting.
31 For example, INTAN TO ASSETS, GROWTH, and R&D TO SALES are all positively corre-
lated and each of the three variables remains significant in column 9 if the other two are not
included.
32 Although the frequency of beginning audited GAAP statements may seem low, note that
this is for only one year and firms typically do not change this decision frequently. If this rate
is cumulated over a number of years, a substantial portion of the firms would experience a
change. See the online appendix for a Markov transition matrix of changes in both type of
standards and whether the financial reports are audited. We also repeat our analysis for firms
ceasing the production of audited GAAP financial statements. Results are reported in the
online appendix; inferences mirror those reported here.
TABLE 7
Firm Characteristics of Audited GAAP Financial Statement Production: Firm Level Changes
Dependent Variable: BEGIN GAAP AUDIT
(1) (2) (3)

∗∗
LOG REVENUE 0.006 0.005 0.005
(2.29) (1.55) (1.58)
LOG NUM OWNER 0.010∗∗∗ 0.015∗∗∗ 0.014∗∗∗
(3.32) (4.24) (3.76)
∗∗ ∗∗
LOG DEBT 0.009∗∗∗ 0.010 0.010
(2.66) (2.55) (2.57)
LOG ACCT PAY 0.018∗∗∗ 0.023∗∗∗ 0.022∗∗∗
(3.32) (4.17) (3.99)
PROFIT MGN −0.001 0.006 0.010
(−0.23) (0.80) (1.28)
∗∗ ∗∗ ∗∗
LOSS 0.017 0.018 0.018
(2.52) (2.27) (2.22)
∗∗ ∗∗ ∗∗
INTAN 0.014 0.013 0.012
(2.54) (2.17) (2.18)
PE TO ASSETS −0.012 −0.006 −0.002
(−0.67) (−0.27) (−0.10)
PE TO WAGES −0.000 0.000 0.000
(−0.79) (0.24) (0.20)
−0.062∗∗∗ −0.062 −0.059
∗∗ ∗∗
LAND TO ASSETS
(−2.96) (−2.19) (−2.17)
C-CORP 0.034∗∗∗ 0.025∗∗∗ 0.021∗∗∗
(3.48) (2.99) (2.77)
GROWTH 0.030∗∗∗ 0.030∗∗∗ 0.029∗∗∗
(3.61) (3.58) (3.48)
PROFIT MGN −0.019 −0.006 −0.004
(−1.38) (−0.43) (−0.30)
LOG ACCT PAY 0.001 0.001 −0.000
(0.08) (0.12) (−0.02)
PE TO ASSETS
∗
0.096 0.070 0.068
(1.96) (1.18) (1.15)
PE TO WAGES −0.002∗∗∗ −0.002 −0.002
∗∗ ∗∗
(−2.80) (−2.45) (−2.49)

LAND TO ASSETS −0.008 −0.045 −0.049
(−0.05) (−0.19) (−0.21)
∗∗
OWNER INC 0.025 0.011 0.003
(2.37) (0.93) (0.32)
RAISE EQUITY 0.024∗∗∗ 0.026∗∗∗ 0.015∗
(3.38) (3.21) (1.81)
DEBT INC 0.002 0.004 0.004
(0.47) (0.71) (0.59)
−0.010 −0.010
∗∗ ∗∗
LOG DEBT FROM SH
(−2.44) (−2.54)
DEBT FROM SH INC −0.007 −0.005
(−0.89) (−0.65)
YOUNG 0.082∗∗∗ 0.026
(3.94) (1.19)
YOUNG∗ OWNER INC 0.143∗
(1.96)
(Continued)
Dependent Variable: BEGIN GAAP AUDIT
(1) (2) (3)

∗∗
YOUNG∗ RAISE EQUITY 0.109
(2.09)
YOUNG∗ DEBT INC 0.023
(0.59)
YOUNG∗ DEBT FROM SH INC −0.018
(−0.32)
Industry FE? Yes Yes Yes
Adjusted R2 0.042 0.058 0.065
Observations 6,541 4,928 4,928
Where Dep. Var. = 1 268 213 213
This table reports linear probability model regression estimates in which the dependent variable = 1 if
the firm begins preparing audited GAAP financial statements in 2009 and = 0 if the firm never prepares
audited GAAP financial statements. The samples in all columns require the firm to exist in all three years
and report not using audited GAAP statements in 2008. Thus, the regressions compare firms that begin
preparing audited GAAP financial statements in 2009 (and continue doing so in 2010) to those that never
prepare audited GAAP financial statements 2008–2010. Changes specifications of all variables are measured
as the difference between 2008 and 2010. Young is defined as firms less than four years old. All models use
robust standard errors clustered by three-digit NAICS industry code. Nonindicator variables are winsorized
at the 1st and 99th percentile levels. Please see the appendix for all variable definitions. The model speci-
fications exclude firms with number of owners >100. ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and
10% levels, respectively.
statements, while those with more land are less likely. Moreover, rein-
forcing one of the main findings of this paper, we continue to find ev-
idence that high-growth, intangibles-based firms are more likely to view
audited GAAP statements as cost-beneficial. In addition, we find that au-
dited GAAP rates are increasing in property and equipment relative to
total assets, but decreasing in the ratio of property and equipment to
Wages.
In column 1, we also assess the importance of financing-based reasons for
preparing audited GAAP statements by including three indicator variables
measuring changes in equity and debt financing. OWNER INC (DEBT INC)
equals 1 if the number of shareholders (sum of short and long-term ex-
ternal debt) increases from 2008 to 2010, and 0 otherwise. Because equity
transactions can occur among owners (e.g., one owner could sell to one or
more other owners) or between owners and the firm (e.g., the firm could
raise equity by selling to new owners), we also include RAISE EQUITY, a
variable equal to 1 if the amount of capital stock plus additional paid in
capital (as measured on Schedule L) increases between 2008 and 2010,
and 0 otherwise. The results in table 7, column 1, suggest that increases
in owners and equity are positively related to the decision to begin pro-
ducing audited GAAP statements, while increases in debt are not. The co-
efficients on both OWNER INC and RAISE EQUITY are significantly posi-
tive (and economically significant relative to the unconditional mean of
5%), whereas the coefficient on DEBT INC is small and statistically in-
significant. We obtain an almost identical estimate using an indicator
variable for NEW DEBT, equal to 1 if the firm moves from 0 to positive debt
(untabulated).
In column 2 of table 7, we focus only on corporations to use proxies
for inside debt and age. Because we noted in the previous analysis that
the relation between firm age and production of audited GAAP statements
mutes after approximately four years and because we also conduct two-way
sorts based on age in subsequent tests, we dichotomize age as an indica-
tor variable YOUNG equal to 1 if the firm is less than four years old, and 0
otherwise. Consistent with our cross-sectional tests, we find that firms with
more inside debt are less likely to begin producing audited GAAP state-
ments (though changes in inside debt do not coincide with changes in fi-
nancial reporting), while younger firms are more likely to begin producing
audited GAAP statements.
In column 3, we further explore how a firm’s reputation (i.e., age) and
potential for having built relationships could mediate the usefulness of
financial statement production in the presence of raising debt and eq-
uity. Results show that young firms that are adding owners (YOUNG ×
OWNER INC) or raising equity (YOUNG × RAISE EQUITY) are substantially
more likely to initiate audited GAAP statement production. Furthermore,
conditioning on age has no such relation with respect to increases in out-
side or inside debt (i.e., the coefficients on both YOUNG × DEBT INC and
YOUNG × DEBT FROM SH INC are insignificant).33
We further illustrate the economic significance of the interactions be-
tween age and capital raising using contingency tables. In table 8, we parti-
tion the sample from column 3 of table 7 along two dimensions: firm age
and either ownership increases (panel A) or debt increases (panel B).34
Overall, three insights emerge from this analysis. First, audited GAAP state-
ments are not necessary for expanding ownership or attracting external
debt—that is, all cells report values less than 100%. Second, noting that the
right-hand column of cells of the contingency tables in both panels have
the larger values, increases in ownership dispersion and increasing debt
are associated with initiating audited GAAP statements—but, noting that
the lower right hand cell values are substantially larger, this association is
much stronger for young firms. Third, although increasing debt is associ-
ated with increased audited GAAP statement production, this difference is
generally less than half the value of increasing ownership. Overall, these
33 We conduct several additional analyses to further investigate the relation with debt
changes. First, we consider whether investigating the extensive margin matters. We condition
the sample only on firms with zero debt in 2008 and create an indicator variable for those
with at least some amount of debt in 2010. We continue to find no significant results on the
indicator (or with an interaction with firm age). Second, we re-estimate table 7, column 2, but
interact age with DEBT CHANGE (continuous amount of change in debt). Again, we do not
find that conditioning on age produces significant results between debt changes and audited
GAAP changes.
34 See the online appendix for additional analyses partitioning on age and raising new
equity.
TABLE 8
2 × 2 Contingency Tables Conditional on Age, Ownership, and Debt
Panel A: Partitioning by firm age and change in
Panel B: Partitioning by firm age and change in debt
ownership
Ownership Dispersion Debt
No change/ No change/
decrease Increase Diff decrease Increase Diff
3.5% 5.1% 1.5% 3.4% 4.4% 1%*
Old Old
4,118 514 3,346 1,286
Age
Age
12.0% 33.3% 21.3%** 13.0% 17.0% 4.1%
Young Young
266 30 208 88
Diff 8.5%*** 28.3%*** 29.8%*** Diff 9.6%*** 12.6%*** 13.6%***
This table reports the propensity for corporations to initiate audited GAAP financial statements condi-
tional on the age of the firm and increase in either ownership dispersion or the level of debt. The sample
is restricted to the balanced panel of corporations without audited GAAP financial statements in 2008. The
cells of the contingency tables report the percentage of firms that begin producing audited GAAP financial
statements in 2009. Young firms are those less than four years old. Ownership dispersion is defined as the
number of shareholders of the corporation. Debt is defined as the sum of external short-term and long-term
debt per Schedule L. Changes in ownership dispersion and debt are measured from the year 2008 to the
year 2009. The model specifications exclude firms with number of owners >100. ∗∗∗ , ∗∗ , and ∗ denote sig-
nificance at the 1%, 5%, and 10% levels, respectively, using robust standard errors clustered by three-digit
NAICS industry code.
results along with those of table 7 suggest that young firms bringing in new
owners—a setting with substantially less research than debt explanations—
have a particularly high propensity to begin producing audited GAAP
statements.
5. Conclusion and Future Directions

This study contributes new insights about private firms and their produc-
tion of audited GAAP financial statements by using a comprehensive panel
of over 200,000 business tax returns for private firms in the United States
with at least $10 million of assets from 2008 to 2010. We find that nearly two-
thirds of medium-to-large private U.S. firms do not produce audited GAAP
financial statements, even though they control trillions of dollars in capital
and outnumber public firms across all industries, even after conditioning
on revenues exceeding $100 million.
Although factors, such as raising equity or debt and the use of trade
credit, explain significant variation in firms’ financial reporting decisions,
our results highlight substantial heterogeneity: thousands of firms with dis-
perse ownership and millions of dollars of debt do not produce audited
GAAP financial statements, while many smaller, closely held firms without
debt do. Moreover, firms with growth opportunities, intangible assets, and
a lack of reputation history have a higher likelihood of producing audited
GAAP statements compared to more established firms with tangible assets.
These inferences generally hold across firm-level, industry-level, and firm-
changes tests. Our results stand in contrast to prior accounting research
on small private firms or publicly traded firms that focus on theories and
empirical results related to contracting, and suggest that financial report-

ing is less relevant for intangibles-type firms (Allee and Yohn [2009], Arm-
strong, Guay, and Weber [2010], Lev and Gu [2016]).
Although our study begins to provide new insights into both the finan-
cial reporting choices and economics of private firms, it generates many
questions. For example, how can it be that so many—even among larger—
private firms do not produce audited GAAP statements while raising ex-
ternal debt? And why do young, high-growth firms raising equity capital
find audited GAAP statements so useful? Substitute mechanisms to costly
audited GAAP financial statements are available to firms, such as collateral
and relationships evident in debt contracting. Also potentially important in
firms’ financial reporting choices are alternatives to audits providing differ-
ent assurance levels, such as reviews and agreed-upon procedures, which
we are unable to measure.
Research on private companies is also timely and directly useful to stan-
dard setters. For example, together with the FASB, the PCC considers which
standards are particularly onerous for private companies. Research could
examine qualified versus unqualified audit opinions and provide evidence
on which standards actually lead to more qualified opinions. As another
example, the PCC has existed since 2012 and private companies now have
a number of alternatives for specific accounting standards. An important
question is, how effective have these changes been? Have the number of
qualified opinions decreased? Are more firms producing audited GAAP
statements? How do these deviations from public companies’ GAAP report-
ing affect private firms’ ability to raise external capital? Despite years of op-
erations, the benefits of these fairly large changes are unknown. Moreover,
although private company data are typically difficult to obtain, targeted
samples—which could be easier to obtain—can provide valuable answers
to these questions.
Finally, our finding that firms frequently do not have audited GAAP fi-
nancial statements is intriguing from another perspective: could it be that
some firms are suboptimizing on their financial reporting when it is not
mandated? If the costs of high-quality financial reporting are salient (e.g.,
the costs of an audit and hiring high-quality internal accounting person-
nel), but the benefits of financial reporting are not (e.g., lower cost of cap-
ital and better firm performance), then private firms may in fact not be
choosing the “right” kind of financial reporting. Research furthering our
understanding of frictions to produce audited GAAP statements would be
helpful to standard setters, preparers, and practitioners.
Overall, there is significant room to better understand and revise be-
liefs over the value of financial reporting in the setting of private U.S.
firms. As the number of private firms is steadily increasing in the U.S.
economy—resulting in an increasing proportion of firms exercising their
right to “remain silent” when it comes to publicly disclosing their finan-
cial statements—we call on additional research to critically examine these
issues.
APPENDIX
Variable Definitions
IRS Form Source
1120 1120S 1065

Financial reporting
choices
GAAP = 1 if Schedule M-3 1 if Schedule M-3 1 if Schedule M-3
Part I Line 4b Part I Line 4b Part I Line 4b
accounting accounting accounting
standard standard standard
checkbox is checkbox is checkbox is
checked “GAAP”; checked “GAAP”; checked
0 otherwise. 0 otherwise. “GAAP”; 0
otherwise.
AUDIT = 1 if Schedule M-3 1 if Schedule M-3 1 if Schedule M-3
Part I Line 1b Part I Line 1a Part I Line 1b
Audit use Audit use Audit use
checkbox is checkbox is checkbox is
checked “Yes”; 0 checked “Yes”; 0 checked “Yes”; 0
otherwise. otherwise. otherwise.
GAAP AUDIT = 1 if GAAP = 1 and 1 if GAAP = 1 and 1 if GAAP = 1 and
AUDIT = 1; 0 AUDIT = 1; 0 AUDIT = 1; 0
otherwise. otherwise. otherwise.
Firm characteristics
LOG REVENUE = log(1+Total log(1+Total log(1+Total
Income), where Income), where Income), where
Total Income is Total Income is Total Income is
Total Income Total Income Total Income
(Line 11) + (Line 6) + COGS (Line 8) +
COGS (Line 2) in (Line 2) in $M COGS (Line 2)
$M from Page 1 from Page 1 in $M from
Page 1
LOG NUM OWNER = log(1+ # of log(1+ # of log(1+ # of
shareholders shareholders partners from
from Schedule K from Page 1 Line Page 1 Line I).
Line 10; =101 if I).
missing (it is
missing if
#shareholders is
>100)
LOG DEBT = log(1+ Mortgages, log(1+ Mortgages, log(1+ Mortgages,
notes, bonds notes, bonds notes, bonds
payable in less payable in less payable in less
than one year than one year than one year
[Schedule L Line [Schedule L Line [Schedule L
17 column d) + 17 column d) + Line 16 column
Mortgages, notes, Mortgages, notes, d) + Mortgages,
bonds payable in bonds payable in notes, bonds
one year or more one year or more payable in one
(Continued)
A P P E N D I X—Continued
IRS Form Source
1120 1120S 1065

[Schedule L Line [Schedule L Line year or more
20 column d], in 20 column d], in [Schedule L
$M) $M) Line 19 column
d], in $M)
LOG DEBT FROM SH log(1+ Loans from log(1+ Loans from Not Available.
Shareholders Shareholders
[Schedule L Line [Schedule L Line
19 column d], in 19 column d], in
$M) $M)
LOG ACCT PAY = log(1 + Accounts log(1 + Accounts log(1 + Accounts
Payable), where Payable), where Payable), where
Accounts Payable Accounts Payable Accounts
from Schedule L from Schedule L Payable from
Line 16 column d, Line 16 column d, Schedule L Line
in $M in $M 15 column d, in
$M
PROFIT MGN = Taxable Income Ordinary Business Ordinary Business
(Loss) before Income (Loss) Income (Loss)
NOL from Page 1 from Page 1 Line from Page 1
Line 28, in 21, in $M/Total Line 22, in
$M/Total Income Income ($M); $M/Total
($M); bounded to bounded to [–1, Income ($M);
[–1, 1]. 1]. bounded to [–1,
1].
LOSS = 1 if PROFIT MGN < 1 if PROFIT MGN < 1 if PROFIT MGN
0; 0 otherwise 0; 0 otherwise < 0; 0 otherwise
PE TO ASSETS = Net Depreciable Net Depreciable Net Depreciable
Assets (Buildings Assets (Buildings Assets
and other and other (Buildings and
depreciable assets depreciable assets other
on Schedule L on Schedule L depreciable
Line 10b column Line 10b column assets on
d) divided by d) divided by Schedule L Line
Total Assets (Page Total Assets (Page 9b column d)
1, Line D). 1, Line F). divided by Total
Assets (Page 1
Line F).
PE TO WAGES = Net Depreciable Net Depreciable Net Depreciable
Assets (Buildings Assets (Buildings Assets
and other and other (Buildings and
depreciable assets depreciable assets other
on Schedule L on Schedule L depreciable
Line 10b column Line 10b column assets on
d) divided by d) divided by Schedule L Line
(Continued)
A P P E N D I X—Continued
IRS Form Source
1120 1120S 1065

Salaries and wages Salaries and wages 9b column d)
(Page 1, Line 13). (Page 1, Line 8). divided by
Salaries and
wages other
than to partners
(Page 1, Line 9).
LAND TO ASSETS Land (Schedule L Land (Schedule L Land (Schedule L
Line 12 column Line 12 column Line 11 column
d) divided by d) divided by d) divided by
Total Assets (Page Total Assets (Page Total Assets
1, Line D). 1, Line D). (Page 1, Line
D).
INTAN = 1 if End of Year 1 if End of Year 1 if End of Year
Gross Intangible Gross Intangible Gross Intangible
Assets (Schedule Assets (Schedule Assets
L Line 13a L Line 13a (Schedule L
column c) > 0; 0 column c) > 0; 0 Line 12a
otherwise. otherwise. column c) > 0;
0 otherwise.
GROWTH = Change in Change in Change in
LOG REVENUE LOG REVENUE LOG REVENUE
LOG AGE = log(1 + Age), where log(1 + Age), where Not Available.
Age is Fiscal Year Age is Fiscal Year
minus Page 1 Box minus Page 1 Box
C “Year” in Date E “Year” in Date
Incorporated Incorporated
field. Available field. Available
only for 1120 and only for 1120 and
1120S filers. 1120S filers.
YOUNG = 1 if firm age < 4; 0 1 if firm age < 4; 0 Not Available.
otherwise. otherwise.
C CORP = 1 if entity filed Form N/A N/A
1120; 0 otherwise.
LOG ASSETS = log(1 + Total log(1 + Total log(1 + Total
Assets), where Assets), where Assets), where
Total Assets from Total Assets from Total Assets
Page 1 Line D, in Page 1 Line F, in from Page 1
$M $M Line F, in $M
REFERENCES
ABDEL-KHALIK, R. “Why Do Private Companies Demand Auditing? A Case for Organizational
Loss of Control.” Journal of Accounting, Auditing, and Finance 8 (1993): 31–52.
ALLEE, K., AND T. YOHN. “The Demand for Financial Statements in an Unregulated Environ-
ment: An Examination of the Production and Use of Financial Statements by Privately Held
Small Businesses.” Accounting Review 84 (2009): 1–25.
ANANTHARAMAN, D.; V. FANG; AND G. GONG. “Inside Debt and the Design of Corporate Debt
Contracts.” Management Science 60 (2013): 1260–80.
ARMSTRONG, C.; W. GUAY; AND J. WEBER. “The Role of Information and Financial Reporting
in Corporate Governance and Debt Contracting.” Journal of Accounting and Economics 50
(2010): 179–234.
ASKER, J.; J. FARRE-MENSA; AND A. LJUNGQVIST. “What Do Private Firms Look Like?” 2011.
Available at SSRN: https://ssrn.com/abstract=1659926.
AZAR, J.; M. SCHMALZ; AND I. TECU. “Anticompetitive Effects of Common Ownership.” Journal
of Finance 73 (2018): 1513–65.
BARRIOS, J.; P. LISOWSKY; AND M. MINNIS. “Measurement Matters: Financial Reporting and
Productivity.” Working paper, University of Chicago and Boston University, 2019.
BARTON, J., AND G. WAYMIRE. “Investor Protection Under Unregulated Financial Reporting.”
Journal of Accounting and Economics 38 (2004): 65–116.
BADERTSCHER, B. A.; J. KIM; W. KINNEY; AND E. OWENS. “Verification Services and Fi-
nancial Reporting Quality: Assessing the Potential of Review Procedures” (July 26,
2018). Simon Business School Working Paper No. FR 17–17. Available at SSRN:
https://ssrn.com/abstract=3016422 or https://doi.org/10.2139/ssrn.3016422
BENSTON, G. “The Market for Public Accounting Services: Demand, Supply and Regulation.”
Journal of Accounting and Public Policy 4 (1985): 33–79.
BERGER, P.; M. MINNIS; AND A. SUTHERLAND. “Commercial Lending Concentration and Bank
Expertise: Evidence from Borrower Financial Statements.” Journal of Accounting and Eco-
nomics 64 (2017): 253–77.
BLACKWELL, D.; T. NOLAND; AND D. WINTERS. “The Value of Auditor Assurance: Evidence from
Loan Pricing.” Journal of Accounting Research 36 (1998): 57–70.
BOTOSAN, C.; H. ASHBAUGH-SKAIFE; A. BEATTY; P. DAVIS-FRIDAY; P. HOPKINS; K. NELSON; K.
RAMESH; R. UHL; M. VENKATACHALAM; AND G. VRANA. “Financial Accounting and Reporting
Standards for Private Entities.” Accounting Horizons 20 (2006): 179–94.
BRADSHAW, M.; D. BENS; C.A. FROST; E. GORDON; S. MCVAY; G. MILLER; R. PFEIFFER; M. PLUM-
LEE; C. SHAKESPEARE; W. THOMAS; AND F. WONG. “Financial Reporting Policy Committee of
the American Accounting Association’s Financial Accounting and Reporting Section: Ac-
counting Standard Setting for Private Companies.” Accounting Horizons 28 (2014): 175–92.
BROWN, G.; R. HARRIS; T. JENKINSON; S. KAPLAN; AND D. ROBINSON. “What Do Differ-
ent Commercial Data Sets Tell Us About Private Equity Performance?” Working paper,
2015.https://ssrn.com/abstract=2706556.
CASSAR, G.; C. ITTNER; AND K. CAVALUZZO. “Alternative Information Sources and Information
Asymmetry Reduction: Evidence from Small Business Debt.” Journal of Accounting and Eco-
nomics 59 (2015): 242–63.
CENDROWKSI, H.; L. PETRO; J. MARTIN; AND A. WADECKI. Private Equity: History, Governance, and
Operations, Second edition. Hoboken, NJ: John Wiley & Sons, 2012.
CHOW, C. “The Demand for External Auditing: Size, Debt and Ownership Influences.” Ac-
counting Review 57 (1982): 272–91.
COPPENS, L., AND E. PEEK. “An Analysis of Earnings Management by European Private Firms.”
Journal of International Accounting, Auditing and Taxation 14 (2005): 1–17.
COSTELLO, A. “Mitigating Incentive Conflicts in Inter-firm Relationships: Evidence from Long-
Term Supply Contracts.” Journal of Accounting and Economics 56 (2013): 19–39.
DEDMAN, E.; A. KAUSAR; AND C. LENNOX. “The Demand for Audit in Private Firms: Recent
Large-Sample Evidence from the UK.” European Accounting Review 23 (2014): 1–23.
DIAMOND, D. “Monitoring and Reputation: The Choice Between Bank Loans and Directly
Placed Debt.” Journal of Political Economy 99 (1991): 689–721.
DOIDGE, C.; G. ANDREW KAROLYI; AND RENÉ M. STULZ. “The U.S. Listing Gap.” Journal of Finan-
cial Economics 123 (2017): 464–87.
EDMANS, A., AND Q. LIU. “Inside Debt.” Review of Finance 15 (2011): 75–102.
FINANCIAL ACCOUNTING FOUNDATION (FAF). Blue-Ribbon Panel on Standard Setting for Private
Companies Report to the Board of Trustees of the Financial Accounting Foundation. Stamford, CT:
FAF, 2011.
FINANCIAL ACCOUNTING FOUNDATION (FAF). “Financial Accounting Foundation Establishes
New Council to Improve Standard Setting for Private Companies.” 2012. Available at
http://www.accountingfoundation.org/cs/ContentServer?c=FAFContent C&pagename=
Foundation/FAFContent C/FAFNewsPage&cid=1176160049265.
GAVER, J.; P. MASON; AND S. UTKE. “Financial Reporting Choices of Private Funds.” Working
paper, 2019. Available at SSRN: https://ssrn.com/abstract=3092331.
GE, W., AND S. MCVAY. “The Disclosure of Material Weaknesses in Internal Control after the
Sarbanes-Oxley Act.” Accounting Horizons 19 (2005): 137–58.
GOVINDARAJAN, V.; S. RAJGOPAL; AND A. SRIVASTAVA. “Why Financial Statements
Don’t Work for Digital Companies.” 2018. Harvard Business Review. Available at
https://hbr.org/2018/02/why-financial-statements-dont-work-for-digital-companies.
GROSSMAN, S. “On the Efficiency of Competitive Stock Markets Where Trades have Diverse
Information.” Journal of Finance 31 (1976): 573–85.
HOLTHAUSEN, R. W., AND R. W. LEFTWICH. “The Economic Consequences of Accounting
Choice Implications of Costly Contracting and Monitoring.” Journal of Accounting and Eco-
nomics 5 (1983): 77–117
HOPE, O. K.; W. THOMAS; AND D. VYAS. “Financial Credibility, Ownership, and Financing Con-
straints in Private Firms.” Journal of International Business Studies 42 (2011): 935–57.
JENSEN, M., AND W. MECKLING. “Theory of the Firm: Managerial Behavior, Agency Costs and
Ownership Structure.” Journal of Financial Economics 3 (1976): 305–60.
KAUSAR, A.; N. SHROFF; AND H. WHITE. “Real Effects of Audit Choice.” Journal of Accounting and
Economics 62 (2016): 157–81.
KOTHARI, S. P.; K. RAMANNA; AND D. SKINNER. “Implications for GAAP from an Analysis of
Positive Research in Accounting.” Journal of Accounting and Economics 50 (2010): 246–86.
LEFTWICH, R. “Discussion of: “Investor Protection under Unregulated Financial Reporting”
(by Jan Barton and Gregory Waymire).” Journal of Accounting and Economics 38 (2004): 117–
28.
LENNOX, C., AND J. PITTMAN. “Voluntary Audits Versus Mandatory Audits.” Accounting Review
86 (2011): 1655–78.
LEV, B., AND F. GU. “The End of Accounting and the Path Forward for In-
vestors and Managers.” 2016. Available at https://onlinelibrary.wiley.com/doi/
book/10.1002/9781119270041
LISOWSKY, P.; M. MINNIS; AND A. SUTHERLAND. “Economic Growth and Financial Statement
Verification.” Journal of Accounting Research 55 (2017): 745–94.
MINNIS, M. “The Value of Financial Statement Verification in Debt Financing: Evidence from
Private U.S. Firms.” Journal of Accounting Research 49 (2011): 457–506.
MINNIS, M., AND N. SHROFF. “Why Regulate Private Firm Disclosure and Auditing?” Accounting
and Business Research 47 (2017): 473–502.
MINNIS, M., AND A. SUTHERLAND. “Financial Statements as Monitoring Mechanisms: Evidence
from Small U.S. Commercial Loans.” Journal of Accounting Research 55 (2017): 197–233.
PETERSEN, M., AND R. RAJAN. “The Benefits of Lending Relationships: Evidence from Small
Business Data.” Journal of Finance 49 (1994): 3–37.
PITCHBOOK. “U.S. PE Middle Market Report 2017 Annual.” 2017. Available at
www.pitchbook.com.
SKINNER, D. “Accounting for Intangibles—A Critical Review of Policy Recommendations.” Ac-
counting and Business Research 38 (2011): 191–204.
SLEE, R. Private Capital Markets, Second edition. Hoboken, NJ: John Wiley and Sons, 2011.
SUNDARAM, R., AND D. YERMACK. “Pay Me Later: Inside Debt and Its Role in Managerial Com-
pensation.” Journal of Finance 62 (2007): 1551–88.
SUTHERLAND, A. “Does Credit Reporting Lead to a Decline in Relationship Lending? Evidence
from Information Sharing Technology.” Journal of Accounting and Economics 66 (2018): 123–
41.
WATTS, R. “Corporate Financial Statements, A Product of the Market and Political Processes.”
Australian Journal of Management 2 (1977): 53–75.
WATTS, R., AND J. ZIMMERMAN. “Towards a Positive Theory of the Determination of Accounting
Standards.” Accounting Review 53 (1978): 112–34.
WATTS, R., AND J. ZIMMERMAN. “Agency Problems, Auditing, and the Theory of the Firm: Some
Evidence.” Journal of Law and Economics 26 (1983): 613–33.
WATTS, R. L., AND J. L. ZIMMERMAN. Positive Accounting Theory. Englewood Cliffs, N.J.: Prentice-
Hall, 1986.
WURSTHORN, M., AND G. ZUCKERMAN. “Fewer Listed Companies: Is That Good or
Bad for Stock Markets?” The Wall Street Journal, January 4, 2018. https://www.wsj.
com/articles/fewer-listed-companies-is-that-good-or-bad-for-stock-markets-1515100040.

2020 - The Silent Majority-Private U.S. Firms and Financial Reporting Choices PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2020 - The Silent Majority-Private U.S. Firms and Financial Reporting Choices PDF

Uploaded by

Copyright:

Available Formats

DOI: 10.1111/1475-679X.

The Silent Majority: Private U.S.

Received 12 April 2018; accepted 27 March 2020

Guided by an agency theory framework, we find that size, ownership disper-

JEL codes: M41; M44; M49

in particular, equity capital allocation. These insights highlight a departure

2. Setting, Theoretical Framework, and Prior Research

accountant.5 Technically, these two dimensions are independent choices.

references therein for additional discussion.

presence of inside debt suggests reduced demand for high-quality financial

operations are not profitable, the residual claimants have incentives to

statement. Of the 20% that do produce financial statements, only 27%

Furthermore, despite this research, firms’ financial reporting practices

publicly traded stock.

1120 1120S 1065 Total

reporting choices are disclosed. We describe our approach in detail in the

finer partitions to prevent revealing any firm-specific information.

n Mean P25 Median P75 SD Mean Mean

n Mean P25 Median P75 SD Mean Mean

LOSS 216,898 0.33 0 0 1 0.47 0.33 0.33

$2.3 billion) that Compustat firms substantially outnumber private firms

commonly used data sets.

4.2 DETERMINANTS OF PRODUCING AUDITED GAAP FINANCIAL STATEMENTS

Total Debt ($ millions)

on alternative mechanisms to safeguard their loans, such as relationships

(1) (2) (3) (4) (5)

ࣘ$25M >$25–$100M ࣙ$100M

performance (PROFIT MGN) and asset tangibility (LAND TO ASSETS)

In column 2 of table 5, we add variables for growth, inside debt, and

We conduct a variety of robustness tests, which we tabulate in the on-

Dependent Variable: Industry Fixed Effect from table 5, column 1

(1) (2) (3) (4) (5) (6) (7) (8) (9)

serve as outsourced internal auditors, allowing external capital providers to

(1) (2) (3)

(−2.80) (−2.45) (−2.49)

(1) (2) (3)

5. Conclusion and Future Directions

empirical results related to contracting, and suggest that financial report-

1120 1120S 1065

1120 1120S 1065

1120 1120S 1065

You might also like