Download The Separation of Ownership and Control and Corporate Tax

The Separation of Ownership and Control and Corporate Tax Avoidance
Brad A. Badertschera
University of Notre Dame
Sharon P. Katzb
Columbia University
Sonja O. Regoc,*
Indiana University
October 2012
*
Corresponding Author
Notre Dame University,371 Mendoza College of Business, Notre Dame, IN 46556-5646. Phone: (574)
631-5197. Email: [email protected].
b
Columbia Business School, Uris Hall, 3022 Broadway, Room 605A, New York, NY 10027. Phone:
(212) 851-9442. Email: [email protected].
c
Indiana University, Kelley School of Business, 1309 E. 10th St., Bloomington, IN 47405-1701. Phone:
(812) 855-8966. Email: [email protected].
a
We are grateful for helpful comments by Ramji Balakrishnan, Jennifer Blouin, Dan Collins, Fabrizio
Ferri, Dan Givoly, Cristi Gleason, Michelle Hanlon, Shane Heitzman, Paul Hribar, Alon Kalay, Michael
Kimbrough, Josh Lerner, Greg Miller, Doron Nissim, Tom Omer, Krishna Palepu, Gil Sadka, Jim Seida,
Joseph Weber, Ryan Wilson, and workshop participants at Baruch College at CUNY, Boston University,
Columbia University, Tel-Aviv University, 2009 Information, Markets & Organization Conference at
Harvard Business School, 2009 JAAF/KPMG Conference, 2010 London Business School Accounting
Symposium, University of Colorado at Boulder, University of Iowa, University of Minnesota, and the
Texas Tax Readings Group. We thank Michelle Shimek for her assistance with the hand-collection of tax
footnote data. All errors are our own.
The Separation of Ownership and Control and Corporate Tax Avoidance
ABSTRACT: We examine whether variation in the separation of ownership and control
influences the tax practices of private firms with different ownership structures. Fama and
Jensen (1983) assert that when equity ownership and corporate decision-making are concentrated
in just a small number of decision-makers, these owner-managers will likely be more risk averse
and thus less willing to invest in risky projects. Because tax avoidance is a risky activity that can
impose significant costs on a firm, we predict that firms with greater concentrations of ownership
and control (and thus more risk averse managers) avoid less income tax than firms with less
concentrated ownership and control. Our results are consistent with these expectations.
However, we also consider a competing explanation for these findings. In particular, we
examine whether certain private firms (i.e., those that are owned by private equity firms) enjoy
lower marginal costs of tax planning, which facilitate greater income tax avoidance. Our results
are consistent with the marginal costs of tax avoidance and the separation of ownership and
control both influencing corporate tax practices.
Keywords:
Ownership structure
Agency costs
Tax avoidance
Private equity firms
Effective tax rates
1. Introduction
In this study we investigate the impact of ownership structure on corporate tax avoidance.
Shackelford and Shevlin (2001) note that little is known about the cross-sectional differences in
the willingness of firms to minimize taxes, and point out that insider control, agency costs, and
ownership structure are important but understudied factors that impact corporate tax avoidance.
We take advantage of a unique sample of firms with privately-owned equity but publicly-traded
debt and examine whether variation in the separation of ownership and control influences
income tax avoidance at private firms.1
Our sample includes private firms that are majority-owned by the firm’s managers (i.e.,
management-owned firms) and private firms that are owned by private equity (PE) firms (i.e.,
PE-backed firms).2 As a result, our research setting exhibits substantial variation in the
separation of ownership and control, as reflected in the proportion of stock owned by the firm’s
managers.3 Due to their public debt, sample firms are required to file financial statements with
the Securities and Exchange Commission (SEC). These filings allow us to utilize audited
financial information and examine corporate tax practices while holding financial reporting
requirements constant. Moreover, because the private firms in our sample are subject to less
public scrutiny than publicly-traded firms (Givoly et al. 2010), they place less weight on
financial reporting decisions and more weight on tax reporting decisions, relative to public firms
1
For the remainder of this paper we refer to firms with private equity and public debt as “private” firms and firms
with public equity and public debt as “public” firms. We note that sample firms are on average larger, have higher
credit and earnings quality, and are financially stronger than private firms that do not issue public debt (Cantillo and
Wright 2000; Denis and Mihov 2003; Bharath et al. 2008; Katz 2009; Givoly et al. 2010).
2
PE firms, such as The Blackstone Group, The Carlyle Group, and Kohlberg Kravis & Roberts, manage investment
funds that generally buy mature businesses via leveraged buyout transactions.
3
To illustrate, the mean (median) proportion of stock owned by managers at management-owned firms is 66.4
(79.4) percent but just 9.0 (4.3) percent at PE-backed private firms (see Table 2, Panel A). In contrast, the mean
(median) proportion of stock owned by managers at S&P 1500 public firms is 5.7 (2.5) percent, with an interquartile
range of 3.8 percent (based on ownership data obtained from ExecuComp, 1992-2010).
1 (e.g., Penno and Simon 1986; Beatty and Harris 1998).4 All of these features enhance the power
of our empirical tests.
Our first analysis compares the income tax avoidance of management-owned and PEbacked private firms. We predict that management-owned firms avoid less tax than PE-backed
firms because management-owned firms have more highly concentrated ownership and control
than PE-backed firms. Our prediction is based on Fama and Jensen’s (1983) theory that when
equity ownership and corporate decision-making are concentrated in just a small number of
decision-makers, these owner-managers will likely be more risk averse and thus less willing to
invest in risky projects. Since tax avoidance is a risky activity that can impose significant costs
on firms and their managers (e.g. Rego and Wilson 2012), we assert that firms with more highly
concentrated ownership and control (and thus more risk averse managers) avoid less income tax
than firms with less concentrated ownership and control. Using a variety of measures of
corporate income tax avoidance and a propensity score matching procedure, we find robust
evidence that management-owned firms avoid less income tax than PE-backed firms, consistent
with firms with more concentrated ownership and control tolerating less corporate tax risk.
We triangulate our primary results in a variety of ways. First, we hand-collect data on the
proportion of stock owned by all named executive officers for our sample of private firms, where
available. The results based on the proportion of managerial stock ownership for this subsample
of private firms, and also within subsamples of management-owned and PE-backed firms,
confirm our main findings that firms with more highly concentrated ownership and control avoid
less income tax than firms with less concentrated ownership and control. Second, we then
compare the tax avoidance of management-owned firms to the tax avoidance of firms with lower
4
Nonetheless, we acknowledge that financial reporting incentives may not be identical between management-owned
and PE-backed private firms, especially around ownership transitions periods. We conduct robustness tests to
address this possibility, as discussed in Sections 2 and 3.
2 rates of managerial stock ownership, including employee-owned private firms and propensity
score-matched public firms. In each case we find that management-owned firms exhibit
significantly lower rates of income tax avoidance than firms with less managerial stock
ownership, consistent with tax avoidance increasing in the separation of ownership and control.
We next consider a competing explanation for why management-owned firms avoid less
income tax than PE-backed firms. Specifically, PE firms may enjoy lower marginal costs of tax
planning, which facilitate greater tax avoidance at PE-backed firms relative to managementowned firms. We first examine this competing explanation by performing tests on PE-backed
firms only. Specifically, we partition PE-backed firms based on two empirical proxies for the
marginal costs of corporate tax planning, including whether the PE-backed private firm is owned
by: 1) a PE firm that owns many vs. fewer portfolio firms, and 2) a large vs. a small PE firm
(based on total capital under PE firm management). We predict private firms owned by PE firms
that own many portfolio firms (large PE firms) avoid more income tax than private firms owned
by PE firms that own fewer portfolio firms (smaller PE firms), due to economies of scale and
scope that systematically affect the marginal costs of tax avoidance at PE-backed firms.5 Our
results are consistent with our predictions. We then consider a more common proxy for
economies of scale – firm size – and examine the marginal costs of tax avoidance amongst
management-owned and PE-backed firms. To the extent that PE firms are able to reduce the
marginal costs of tax avoidance for their portfolio firms (small and large alike), we expect to find
larger differences between small-sized management-owned and PE-backed firms than between
large-sized management-owned and PE-backed firms, since large firms may enjoy economies of
5
We utilize the delegation of authority papers by Aghion and Tirole (1997) and Baker, Gibbons, and Murphy (1999)
to provide insights into why some PE firm owners might retain decision rights over tax planning at PE-backed firms
(by requiring portfolio firms to acquire tax services from a particular tax service provider), but delegate authority
over day-to-day operations (including the implementation of tax planning) to portfolio firm managers.
3 scale to tax planning independent of PE ownership (e.g., Rego 2003; Dyreng et al. 2008). Our
results are consistent with our predictions and indicate that small-sized firms experience the
greatest tax savings from PE ownership.
We then directly disentangle the dual impact of the separation of ownership and control
from lower marginal costs of tax avoidance by including empirical proxies for these constructs in
the same regression. We continue to find that while higher concentrations of ownership and
control are associated with less corporate tax avoidance, lower marginal costs of tax planning are
associated with greater corporate tax avoidance. These results hold in tests that compare tax
avoidance at management-owned and PE-backed firms, and also amongst PE-backed firms only.
Finally, we provide exploratory evidence regarding the methods that PE-backed firms
utilize to avoid more income taxes than management-owned firms. Our results broadly suggest
that the lower effective tax rates of PE-backed firms are caused – at least in part – by their use of
intangible assets, tax-exempt investments, tax credits, and the use of multi-jurisdictional tax
planning, including affiliates in low-tax rate foreign countries.
Our study extends the accounting and finance literatures in several ways. Prior
accounting research considers the impact of different organizational features, including dualclass stock and public vs. private ownership on corporate tax practices (e.g., Klassen 1997; Mills
and Newberry 2001; Chen et al. 2010; McGuire et al. 2011), but these studies provide disparate
evidence on how organizational structure influences corporate tax avoidance. In contrast, we use
Fama and Jensen’s (1983) theory on the separation of ownership and control to understand how
one specific feature of organizational structure – the separation of ownership and control –
impacts corporate tax practices. Our findings are relevant for future research on the impact of
organizational structure on corporate tax avoidance. They also increase our understanding of
4 how PE firms generate value in their portfolio firms. Prior research documents that PE firms
create value in their portfolio firms by implementing effective financial and operating strategies
and by actively monitoring top executives at their portfolio firms (e.g., Cao and Lerner 2009;
Kaplan and Stromberg 2009; Masulis and Thomas 2009). However, little is known about PEbacked firms’ tax practices. Given recent criticisms of PE firm investment practices,6 and the
growing significance of PE firms for the U.S. capital markets,7 our study provides new insights
on the extent to which PE firms increase portfolio firm value by increasing their tax efficiency
relative to other private firms.
2. Background and Empirical Predictions
2.1 The Separation of Ownership and Control and Prior Tax Research
Corporations exhibit substantial variation in the extent to which equity ownership is
separated from control over corporate decision-making. At the extremes, small closely-held
corporations have highly concentrated equity ownership and control, while large publicly-traded
corporations have nearly complete separation of equity ownership and control. The separation of
ownership and control creates well-known agency problems, including managerial incentives to
pursue non-value-maximizing behaviors such as shirking, perquisite consumption, and rent
extraction. To reduce these agency costs, firms write contracts that align managers’ incentives
with those of shareholders. These contracts incentivize managers to invest in projects that
increase firm value (e.g., Jensen and Meckling 1976; Smith and Watts 1982; Smith and Stulz
1985).
6
The rapid growth of the PE industry has raised concerns regarding anticompetitive behavior, excessive tax
benefits, and stock manipulations in this sector (see Katz 2009 and Section 2 for further discussion).
7
The cumulative capital commitments to non-venture capital PE firms in the U.S. between 1980 and 2006 is
estimated to be close to $1.4 trillion (Stromberg 2008). In addition, approximately $400 billion of PE-backed
transactions were announced in both 2006 and 2007, representing over two percent of the total capitalization of the
U.S. stock market in each of these years (Kaplan 2009). Despite a decline in PE transactions since 2007, experts
maintain that PE firms have become a permanent component of U.S. investment activity (e.g., Kaplan 2009; Kaplan
and Stromberg 2009).
5 Fama and Jensen (1983) describe the circumstances in which firms should separate or
combine decision management and decision control with residual risk sharing, where decision
management includes the initiation and implementation of decisions by “decision agents”
(typically top executives); decision control includes the ratification and monitoring of decisions
and decision agents (typically by the board of directors); and the residual claimants of a firm
(i.e., the common equity owners) share the residual risk and cash flows of the firm. The
separation of decision management from residual risk sharing is often referred to as the
separation of ownership and control. Fama and Jensen (1983) explain ownership and control
should be combined in the same decision agents (i.e., managers) in smaller organizations where
information relevant to decision-making is concentrated in a few individuals. In this case, the
benefits of low agency costs and efficient decision-making are greater than the costs of reduced
risk sharing. In contrast, decision management should be separated from residual risk sharing in
larger organizations with diffuse residual claims and where information relevant to decisionmaking is dispersed across individuals at all levels of the organization. In this case, decision
management should be delegated to individuals that possess relevant information, and decision
management should be separated from decision control to reduce the agency costs associated
with diffuse residual claims.
One key factor in Fama and Jensen’s (1983) theory is the extent to which equity
ownership is concentrated in a few decision agents (i.e., managers). When this happens, Fama
and Jensen state that it is rational for the managers to invest in less risky projects because their
portfolios are likely less diversified than that of managers in organizations with more diffuse
6 equity ownership.8,9 Consistent with Rego and Wilson’s (2012) view of corporate tax avoidance,
we argue that tax avoidance is one risky activity in which undiversified – and thus more risk
averse – managers will minimize their investments. Rego and Wilson (2012) maintain that tax
avoidance is a risky activity that can impose significant costs on firms and their managers,
including fees paid to tax experts, time devoted to the resolution of tax audits, reputational
penalties, and penalties paid to tax authorities. Thus, risk-averse managers likely prefer to
undertake less risky tax planning, while risk-neutral shareholders prefer managers to implement
all tax strategies that are expected to increase firm value, regardless of risk.
Prior accounting research has examined the impact of different organizational structures
on corporate tax practices, but no single study has examined how the separation of ownership
and control impacts tax avoidance across a broad set of firms. Instead, prior research has
investigated tax avoidance at publicly-traded, dual-class stock firms, firms with hedge fund
activists, and family-owned firms (Chen, et al. 2010; McGuire et al. 2011; Cheng et al. 2012),
and more generally at public vs. private firms (e.g., Beatty and Harris 1998; Mikhail 1999; Mills
and Newberry 2001). Klassen (1997) documents public firms that are subject to higher capital
market pressure place greater weight on financial than taxable income when divesting operating
units, relative to public firms subject to less capital market pressure.10 That is, public firms
subject to greater capital market pressure are willing to trade-off higher tax costs for the benefit
of higher financial accounting income. In this study we endeavor to take a broader perspective.
We use Fama and Jensen’s (1983) theory on the separation of ownership and control to develop
8
The combination of decision management and decision control with residual risk sharing in a small number of
agents also generates “efficiency losses because decision agents must be chosen on the basis of wealth and
willingness to bear risk as well as for decision skills” (Fama and Jensen 1983, p. 306).
9
We assume that the diversification of a manager’s portfolio and professional reputation are decreasing in the
proportion of stock owned in the firm. Thus, greater managerial stock ownership implies greater risk aversity.
10
Klassen (1997) utilizes inside ownership concentration as his proxy for capital market pressure, where the mean
(median) inside ownership concentration for his sample of 327 public firms is 15.1 (8.2) percent.
7 empirical predictions for variation in tax avoidance amongst private firms with different
ownership structures, which are subject to less capital market pressure than public firms.11
2.2 Private Equity Firms
Our main empirical tests are based on management-owned and PE-backed private firms.
PE firms manage investment funds that generally acquire majority control of mature, profitable
businesses via leveraged buyout (LBO) transactions. We refer to these acquired businesses as
“portfolio firms” or “PE-backed firms.” Before we develop our empirical predictions, we first
discuss the organizational structure of PE firms, and then describe how PE firms manage their
portfolio firms (i.e., the PE-backed firms). This discussion provides the foundation for several
empirical predictions, and ultimately is essential to understanding the “ownership and control” of
PE-backed firms.
PE firms have received recent attention due to their substantial impact on merger and
acquisition activity and their generous tax treatment in the U.S. and other countries. PE firms are
typically organized as limited partnerships and most PE firm executive managers are partners in
the PE firm. Thus, we also refer to PE firm managers as “PE firm partners.” PE firms manage
the PE investment funds that directly acquire mature, profitable businesses via LBO (see Figure
1). PE funds primarily finance portfolio firm acquisitions with the capital contributed by limited
partners (i.e., investors in the PE fund) and substantial amounts of debt, resulting in highly
leveraged portfolio firms. PE firm partners contribute just a small proportion of the PE fund
capital (i.e., approximately 1 percent). The limited partners pay annual management fees
11
For the most part we exclude public firms from our study, since publicly-traded firms are subject to greater
financial reporting pressure due to greater scrutiny from investors, analysts, and regulators than private firms, and
prior research demonstrates that greater financial reporting pressure differentially affects tax avoidance at public and
private firms (e.g., Beatty and Harris 1998; Mikhail 1999; Mills and Newberry 2001). Nonetheless, Fama and
Jensen’s (1983) theory on how the separation of ownership and control should impact a manager’s risk aversity can
also be applied to a sample that only includes public firms. However, public firms generally exhibit substantially
less variation in managerial stock ownership compared to our sample of private firms (see Section 1).
8 (typically 2 percent of invested capital) to the PE fund as compensation for PE fund investment
operations. The PE fund also receives a 20 percent share (i.e., carried interest) of any gains
generated by the sale or IPO of portfolio firms (Kaplan and Stromberg 2009). The taxation of
PE firms and PE firm partners has been criticized as exceedingly unfair.12
The generally negative view of the tax benefits enjoyed by PE firms contrasts other
characteristics associated with their management of portfolio firms. PE firms usually obtain a
concentrated ownership stake and control of the board of directors with the intent of substantially
improving portfolio firm performance. Portfolio firm boards are typically comprised of the
CEO, PE firm partners, and outside industry experts. Portfolio firms’ boards are smaller than
comparable public firms’ boards and they meet more frequently via both formal and informal
meetings. These board members advise portfolio firm managers on strategic considerations, and
actively monitor and motivate the management team (Cotter and Peck 2001; Jensen 2007;
Cornelli and Karakas 2008; Kaplan and Stromberg 2009; Masulis and Thomas 2009). PE firm
partners use their control over the board of directors to impose performance-based compensation
on portfolio firm managers and do not hesitate to replace them when they underperform (Kaplan
and Stromberg 2009; Acharya et al. 2010). As a result, portfolio firm boards are widely
considered more effective than both public and other private company boards (Gilson and
Whitehead 2008; Masulis and Thomas 2009; Strömberg 2009). In sum, prior research indicates
that PE firms exercise substantial control over their portfolio firms’ boards of directors and
actively monitor the portfolio firm management team.
12
While the management fees are generally taxed as ordinary income (i.e., 35 percent tax rate), the carried interest is
taxed as long-term capital gain (i.e., 15 percent tax rate). This tax treatment of carried interest, as well as the fact
that some PE firms have been able to avoid corporate taxation once they file for an initial public offering (e.g., The
Blackstone Group) has provoked numerous negative press reports, proposed changes to federal income tax laws, and
academic studies on the tax treatment of PE firms (e.g. Fleischer 2007, 2008; Knoll 2007; Cunningham and Engler
2008; Lawton 2008).
9 Large PE firms often hire professionals with operating backgrounds and industry
expertise to work with portfolio firm managers (Gadiesh and MacArthur 2008; Acharya et al.
2010). To learn how PE firms influence the tax practices of their portfolio firms, we spoke with
partners at a large public accounting firm that provides tax services to PE-backed firms. The
partners indicated that PE firms frequently arrange for their portfolio companies to acquire tax
services from a specific accounting firm, with the intention of reducing portfolio firm tax costs
through more sophisticated tax strategies than would otherwise be used by the portfolio firm
(e.g., maximizing the utilization of net operating loss carryforwards and R&D tax credits). Thus,
some PE firms view tax planning as one avenue for increasing portfolio firm value.
While PE firm partners actively monitor portfolio firm operations through their control of
portfolio firm boards, they generally do not assume management roles in PE-backed firms (e.g.,
Cao and Lerner 2009; Kaplan and Stromberg 2009; Masulis and Thomas 2009). Instead PE firm
partners act as advisors to the portfolio firm management team. In addition, PE firms typically
acquire majority equity stakes in their portfolio companies. This separation of equity ownership
(by PE firms) and decision management (by portfolio firm managers) at PE-backed firms leads
to an organizational structure that also separates decision management from decision control (by
portfolio firm boards).13 In contrast, private firms that are owned by the firm’s management
often combine decision management, decision control, and equity ownership in a few
individuals, which provides the basis for our empirical predictions.
2.3 Empirical Predictions
Utilizing a variety of settings where the separation of ownership and control exhibits
substantial variation, we empirically test one specific implication of Fama and Jensen’s (1983)
13
This organizational structure is consistent with the prediction of Fama and Jensen (1983) that “when venture
equity capital is put into a small entrepreneurial organization by outsiders, mechanisms for separating the
management and control of important decisions are instituted” (footnote 9, page 306).
10 theory. In particular, we examine whether firms with more concentrated ownership and control
avoid less income tax than firms with less concentrated ownership and control. We also consider
a competing explanation for these findings, that being whether PE-backed firms enjoy lower
marginal costs of tax planning, which in turn facilitate greater income tax avoidance.
2.3.1 Predictions for the Separation of Ownership and Control and Tax Avoidance
To test this hypothesis we utilize a unique sample of private firms with privately-owned
equity but publicly-traded debt. This sample holds financial reporting requirements constant,
since all sample firms are required to file financial statements with the SEC, but are subject to
less capital market pressure than public firms (e.g. Givoly et al. 2010).14 Our sample also
exhibits substantial variation in the separation of ownership and control, making it a powerful
setting to test our empirical predictions.
Our primary tests are based on management-owned and PE-backed private firms. As
demonstrated in later analyses, top executives at management-owned firms own greater
proportions of company stock than top executives at PE-backed firms. As a result, managementowned firms exhibit higher concentrations of ownership and control than PE-backed firms.
Consistent with Fama and Jensen (1983), we assume that the diversification of a manager’s
portfolio (and professional reputation) is decreasing in the proportion of stock owned in the firm.
Thus, the higher concentrations of ownership and control at management-owned firms should
cause their owner-managers to be more risk averse and tolerate less tax risk than managers at PEbacked firms, which leads to our first empirical prediction:
14
To address concerns that PE-backed firms are subject to different financial reporting incentives than managementowned firms, since they are typically sold or taken pubic via IPO within five to seven years of being purchased, we
re-run our main tests separately for the sub-groups of private firms that eventually go public (Private → Public) and
that once were public but then go private (Public → Private). Specifically, we compare the tax avoidance of
management-owned and PE-backed firms during the first five private firm-years (if available) after transitioning
from public ownership. Similarly, we compare the tax avoidance of management-owned and PE-backed firms
during the last five private firm-years (if available) prior to transitioning to public ownership. Our results
(untabulated) confirm that management-owned firms avoid less income tax than PE-backed firms.
11 P1. Management-owned firms avoid less income tax than PE-backed private firms.
Prior research provides insights into variation in the separation of ownership and control
at firms with different organizational structures. Katz (2009) documents that the proportion of
stock owned by top executives at management-owned firms is significantly greater than
managerial stock ownership at PE-backed firms. Amongst PE-backed firms the proportion of
stock owned by top executives at minority-owned, PE-backed firms is significantly greater than
managerial stock ownership at majority-owned, PE-backed firms. With respect to other types of
firms, Kaplan and Stromberg (2009) and Acharya and Kehoe (2010) assert that CEOs at PEbacked firms typically own larger proportions of portfolio firm stock than CEOs of public firms,
while Bova et al. (2012a) and Bova et al. (2012b) state that employee-owned private firms
generally have stock ownership that is diffused across many individuals. Taken together, these
studies suggest that management-owned firms exhibit the highest concentrations of ownership
and control and lead to the following empirical predictions that build on P1:
P1a. Management-owned firms avoid less income tax than firms that are majority-owned
by PE-backed firms.
P1b. Management-owned firms avoid less income tax than firms that are minority-owned
by PE-backed firms.
P1c. Minority-owned, PE-backed firms avoid less income tax than majority-owned, PEbacked firms.
P1d. Management-owned firms avoid less income tax than employee-owned firms.
P1e. Management-owned firms avoid less income tax than public firms.
All of these predictions are based on Fama and Jensen’s (1983) theory that managers at
firms with high concentrations of ownership and control likely have less diversified portfolios
and thus should be more risk averse than managers at firms with less concentrated ownership and
12 control, all else equal. We predict greater managerial risk aversity should lead to less income tax
avoidance.
2.3.2 Predictions for the Marginal Costs of Tax Avoidance at PE-Backed Firms
It is possible that PE-backed private firms are fundamentally different from managementowned firms (beyond the differences in ownership and control) and these differences influence
the tax practices at management-owned and PE-backed firms. One specific attribute that would
allow PE-backed firms to avoid more income taxes than management-owned firms involves the
marginal costs of tax avoidance at PE-backed firms. Prior theoretical research examines the
circumstances in which a principal is likely to delegate authority (either formal or informal) to an
agent. These studies find that the principal is likely to retain authority over decision making
when the principal is better informed than the agent (Aghion and Tirole 1997; Baker, Gibbons,
and Murphy 1999).15 In our research setting, PE firms (and PE firm general partners) can be
considered the “principals” in the authority literature, while portfolio firm management teams are
the “agents.” From this perspective, we can evaluate the extent to which PE firms are likely to
“retain authority” over tax planning at their portfolio companies. Due to their financial resources
and past experience in managing portfolio companies, PE firms have access to superior thirdparty tax planning expertise, which in effect makes PE firms better informed about tax planning
strategies relative to portfolio firm managers. Thus, extant theory on formal and informal
authority in organizations would suggest that PE firms are likely to retain authority over tax
planning at their portfolio companies. In contrast, PE firms are not likely to retain authority over
15
Aghion and Tirole (1997) claim that asymmetric information is the key to understanding the delegation of
authority. They also explain that formal authority is likely to be delegated for decisions that are: 1) relatively
unimportant for the principal, 2) but important to the agent, 3) for which the principal can trust the agent, and 4) are
sufficiently innovative that the principal does not have substantial experience or competency.
13 most portfolio firm operating decisions, since portfolio firm managers are typically better
informed than PE firm partners with respect to day-to-day operating decisions.
Our understanding is that many PE firms effectively retain decision rights over tax
planning at PE-backed firms by arranging tax service providers for their portfolio firms. Thus,
because PE firms typically own more than one portfolio firm, PE firms should be able to reduce
the marginal costs of tax avoidance at PE-backed firms by negotiating lower tax fees for their
portfolio firms, and/or by applying similar tax planning strategies at more than one PE-backed
firm. For example, the same tax service provider could maximize the R&D tax credits at all
portfolio companies that are eligible for these credits. Thus, PE firms have the ability to generate
economies of scale and scope for tax avoidance at PE-backed firms. Conversations with tax
partners at a large public accounting firm are consistent with this assertion. These partners
explained that some (but not all) PE firm clients effectively retain decision rights with respect to
tax planning at PE-backed firms by arranging a particular tax service provider for most or all of
their portfolio firms.16 The centralization of tax accounting services should reduce the marginal
costs of tax planning at PE-backed firms, resulting in greater tax avoidance at PE-backed firms
relative to management-owned firms.17 Thus, we also examine the extent to which variation in
the marginal costs of tax planning impact tax avoidance at private firms.
To hold the separation of ownership and control relatively constant, we first restrict our
analyses to PE-backed firms. Within this sub-sample, we assert that firms owned by PE firms
with “many” portfolio companies are likely to have lower marginal costs of tax planning than
firms owned by PE firms with “fewer” portfolio companies. We classify a PE firm as having
16
These partners also stated that PE firms similarly reduce other portfolio firm costs by centralizing certain
administrative services for their portfolio companies. For example, some PE firms require their portfolio firms to
purchase legal services from specific law firms and insurance services from specific insurance firms.
17
However, tax services must be tailored to fit the particular needs of each portfolio company and so it is not clear
that PE-backed firms truly enjoy lower marginal costs of tax planning.
14 “many” portfolio firms if they own more than 200 portfolio firms and their ratio of equity
invested-to-number of portfolio firms is greater than $30 million. We classify all PE firms not
meeting these two requirements as having “fewer” portfolio firms. PE firms that own many
portfolio companies should enjoy economies of scale and scope with respect to tax planning
costs at their portfolio firms, since the same tax planning strategies can potentially be utilized at
a larger number of portfolio firms. Thus, our next empirical prediction is:
P2a. Firms that are owned by PE firms with more portfolio firms avoid more income tax
than firms that are owned by PE firms with fewer portfolio firms.
Consistent with the discussion above, we also partition PE-backed firms based on
whether they are owned by large or small PE firms, where “large” PE firms include the fifteen
largest PE firms as measured by total capital under PE firm management during our sample
period.18 We classify all other PE firms as “small” PE firms. We expect firms that are owned by
large PE firms to have lower marginal costs of tax planning than firms that are owned by small
PE firms, since large PE firms should enjoy economies of scale and scope with respect to tax
planning costs at their portfolio firms. Indeed, prior research shows that large PE firms regularly
outperform smaller PE firms, consistent with a greater ability to create financial value through
operational improvements at portfolio firms (e.g., Kaplan and Schoar 2005; Acharya et al. 2010).
Similarly, large PE funds, which build superior reputations with lenders, are documented to
obtain cheaper loans and less restrictive debt covenants than other borrowers (Kaplan and
Stomberg 2009; Demiroglu and James 2010; Ivashina and Kovner 2011). Thus, our next
empirical prediction is:
18
The fifteen largest PE firms are Carlyle Group, Blackstone Group, Warburg Pincus, Kohlberg, Kravis, Roberts
and Company, Goldman Sachs and Company, Cerberus Capital Management, Fortress Investment Group, Apollo
Global Management, Bain Capital, TPG Capital, 3i Group, Apax Partners Worldwide, Thomas H. Lee Partners,
Morgan Stanley Private Equity, and Welsh, Carson, Anderson, and Stowe.
15 P2b. PE-backed firms that are owned by large PE firms avoid more income tax than PEbacked firms that are owned by smaller PE firms.
We now consider variation in the marginal costs of tax planning amongst managementowned and PE-backed firms. To the extent that PE firms reduce the marginal costs of tax
planning by centralizing tax services for their portfolio firms (small and large alike), we expect
to find greater differences between small-sized, management-owned and PE-backed firms than
between large-sized, management-owned and PE-backed firms, since large-sized firms may
enjoy economies of scale to tax planning independent of PE ownership (e.g., Rego 2003; Dyreng
et al. 2008).19 We define small-sized (large-sized) firms as those in the lowest (highest) quartile
of net sales for our sample of private firms and predict:
P2c. The difference in tax avoidance at small-sized, PE-backed and management-owned
firms is larger than the difference in tax avoidance at large-sized, PE-backed and
management-owned firms.
Lastly, we directly attempt to disentangle the dual impact of the separation of ownership
and control from the marginal costs of tax planning on corporate tax avoidance. Specifically, we
hand-collect managerial stock ownership data for our samples of management-owned and PEbacked firms, which allows us to include empirical proxies for a manager’s risk aversity and the
marginal costs of tax planning in the same regression. As previously explained, we assume that
the diversification of a manager’s portfolio is decreasing in the proportion of stock owned in the
firm and lower diversification leads to greater risk aversity. Our final empirical prediction is:
P3. Holding the marginal costs of tax avoidance constant, private firms with managers
that own larger proportions of the firm’s stock avoid less income tax than private
firms with managers that own smaller proportions of the firm’s stock.
3. Research Design
3.1 Measures of Corporate Tax Avoidance
19
Note that the term “large PE” firm refers to economies of scale and scope of the PE firm, while the term “largesized PE-backed” firm refers to economies of scale and scope of the PE-backed firm.
16 We rely on several measures of tax avoidance because different measures capture
different aspects of corporate tax planning. Our first two measures are based on effective tax
rates and include GAAP_ETR and CASH_ETR, where GAAP_ETR (CASH_ETR) is total tax
expense (cash taxes paid) summed over three years, scaled by adjusted pretax income summed
over three years.20 Both measures convey a firm’s average tax cost per dollar of pretax income
and capture a broad range of tax planning activities that can have both certain and uncertain
outcomes with tax authorities. Recent research presents evidence that both effective tax rate
measures reflect variation in tax avoidance across firms (Dyreng et al. 2008; Robinson et al.
2010; Armstrong et al. 2012).
We complement these effective tax rate measures with two additional measures designed
to capture more risky tax avoidance: Frank et al.’s (2009) discretionary permanent book-tax
difference measure (DTAX) and Wilson’s (2009) measure of tax sheltering (SHELTER). While
DTAX is the residual from a regression of permanent book-tax differences on non-discretionary
sources of those differences,21 SHELTER is the predicted value from a tax shelter prediction
model. Frank et al. (2009) demonstrate that DTAX is significantly associated with actual cases of
tax sheltering and Wilson (2009) demonstrates that SHELTER is able to predict tax shelter
activity out-of-sample. See the Appendix for details on how we calculate each of these
measures.
We acknowledge that all four measures reflect income tax avoidance with error. While
the effective tax rate measures are commonly used in accounting research and understood by a
20
Whenever possible we use three years of data to calculate GAAP_ETR and CASH_ETR. However, if data
limitations (such as transition years or missing values) prohibit us from using three years of data, we next use two
years, followed by one year of data. Results are qualitatively similar if we base our calculations on one year of data.
21
GAAP_ETR also reflects variation in permanent book-tax differences, where permanent book-tax differences are
differences between financial and taxable income that do not reverse through time (e.g., interest income from
municipal bonds is exempt from federal income taxation but included in pre-tax financial income). DTAX is distinct
from GAAP_ETR because Frank et al.’s (2009) model is designed to remove non-discretionary sources of permanent
book-tax differences from GAAP_ETR to isolate intentional, more aggressive tax avoidance.
17 broad set of financial statement users, they capture all types of tax avoidance (i.e., risky and nonrisky strategies alike). Moreover, GAAP_ETR is confounded by changes in tax reserves and the
valuation allowance, while CASH_ETR is confounded by the timing of tax payments, settlements
with tax authorities, and some types of earnings management. In contrast, DTAX and SHELTER
were designed to capture more risky tax avoidance, and in fact both measures are associated with
tax shelter transactions (Frank et al. 2009; Wilson 2009). But DTAX only captures “permanent”
tax strategies,22 and both DTAX and SHELTER are based on cross-sectional empirical models
that are subject to criticisms similar to those directed at discretionary accrual models (i.e., the
models estimate tax avoidance with error). None of the four measures are clearly superior (or
inferior) to the other three. Consequently, we rely on all four measures in our empirical tests to
evaluate the robustness of our results.23
3.2 Modeling the Impact of Separation of Ownership and Control on Corporate Tax Avoidance
To investigate whether the separation of ownership and control impacts corporate tax
avoidance, we estimate equation (1) below. P1 predicts that management-owned firms avoid
less income tax than PE-backed firms. Thus, the variable of interest in equation (1) is
MGMT_OWNED, although in some specifications we replace MGMT_OWNED with other
proxies for the separation of ownership and control:
TAXi = 0 + 1MGMT_OWNEDi + 2RNOAi + 3LOSSi + 4NOLi + 5LEVi + 6INTANGi
+ 7MNCi + 8AB_ACCRi + 9EQ_EARNi + 10SALES_GRi + 11ASSETSi + 12SOXi + 13INV_MILLSi + ji YEARi + kl INDUSi + I, (1)
The dependent variable, TAX, represents the four proxies for corporate tax avoidance:
GAAP_ETR, CASH_ETR, DTAX, and SHELTER. The indicator variable, MGMT_OWNED,
22
“Temporary” tax strategies reverse through time because they temporarily accelerate expense recognition or defer
revenue recognition, while “permanent” tax strategies affect book and taxable income differently, and in a manner
that is not expected to reverse (e.g., shifting income from a high-tax to a low-tax location).
23
To the extent we obtain results that are consistent across the four measures we can be confident that our findings
are highly robust across the various tax avoidance metrics.
18 equals ‘one’ if a firm is majority-owned by its current and past named executive officers, and
zero if otherwise. If management-owned firms avoid less tax than PE-backed firms, then the
coefficient on MGMT_OWNED should be significant and positive (negative) in regressions
where GAAP_ETR and CASH_ETR (DTAX, SHELTER) are the dependent variables. See the
Appendix for a detailed definition of each variable included in equation (1).
Equation (1) also includes controls for factors that influence a firm’s tax avoidance
activity, as documented by prior research (e.g., Manzon and Plesko 2002; Rego 2003; Dyreng et
al. 2008; Frank et al. 2009; Wilson 2009; Chen et al. 2010). The first set of control variables,
which includes RNOA, LOSS, NOL, and LEV, controls for a firm’s need to tax plan. We include
an indicator variable, LOSS, and the return on net operating assets (RNOA) as proxies for current
profitability, since profitable firms have greater incentives to tax plan. We include an indicator
variable for the presence of net operating loss carryforwards (NOL) at the beginning of the year,
since firms with loss carryforwards have less incentive to engage in current year tax planning.
We include a firm’s leverage ratio (LEV) because firms with greater leverage have less need to
tax plan due to the tax benefits of debt financing.
We include an indicator variable for foreign operations (MNC) in equation (1), since
firms with foreign operations have greater opportunities for tax avoidance by shifting income
between high and low tax rate locations (e.g., Rego 2003). MNC equals ‘one’ if a firm reports
non-zero foreign income or foreign tax expense, and zero if otherwise. We control for intangible
assets (INTANG) and equity in earnings of unconsolidated affiliates (EQ_EARN) because these
items often generate differences between book and taxable income and can thus affect our tax
avoidance measures.24 We include sales growth (SALES_GR) in equation (1) because growing
24
We note that intangible assets represent at least two different constructs. First, intangible assets are subject to
different amortization rules for financial and tax reporting purposes; thus, to some extent, intangible assets generate
19 firms likely make larger investments in depreciable assets, which generate larger temporary
book-tax differences and can thus affect some tax avoidance measures. We control for firm size
(ASSETS) because large firms likely enjoy economies of scale in tax planning. We include an
indicator variable for years following the Sarbanes-Oxley Act of 2002 (SOX), since prior
research demonstrates that the regulatory environment surrounding corporate financial and tax
reporting changed substantially in the post-SOX time period (e.g. Cohen et al. 2008). We further
include year (YEAR) and industry (INDUS) fixed-effects to control for fundamental differences
in tax planning that may exist across years and industries.
Frank et al. (2009) find a strong positive relation between financial and tax reporting
aggressiveness. Katz (2009) documents that PE-backed firms report more conservatively and
engage in less earnings management compared to non-PE-backed firms. To the extent our test
and control firms exhibit different financial reporting quality, we need to control for financial
reporting quality in equation (1). Thus, we control for both timely loss recognition and earnings
management by including AB_ACCR in equation (1).25 AB_ACCR is the amount of abnormal
accruals after controlling for conservatism in our abnormal accruals calculation (see Ball and
Shivakumar 2006).
Our last control variable is the inverse Mills ratio (INV_MILLS) from the first stage of the
Heckman (1979) sample selection correction procedure. This two-stage estimation procedure
attempts to correct endogeneity associated with PE firm investment decisions (e.g., if the same
characteristics that influence PE firm ownership are also correlated with portfolio firm tax
nondiscretionary book-tax differences that are unrelated to intentional tax avoidance. Second, intangible assets are
also frequently used to avoid income taxes; e.g., the placement of intangible assets in a low-tax jurisdiction allows
firms to shift profits from high-tax jurisdictions to low-tax jurisdictions. Thus, intangible assets also capture a firm’s
ability to engage in multijurisdictional tax avoidance. By including INTANG in our regressions, we are biasing
against finding significant results for our variable of interest (e.g., MGMT_OWNED).
25
Results (untabulated) are substantially similar if we replace AB_ACCR with the absolute value of AB_ACCR.
20 avoidance). The Heckman procedure is performed for all regression analyses that include both
management-owned and PE-backed private firms. In the first stage, we estimate the following
probit regression, which predicts whether a private company is owned by a PE firm:
PE_BACKED = 0 + 1BVE + 2RNOA + 3Q_RATIO + 4OPER_CYCLE +
5FIRM_AGE + 6CASH + 7CAP_EXP + 8BIG_AUDIT + 9LOSS +
10NOL + 11LEV + 12MNC + 13INTANG + 14EQ_EARN +
15SALES_GR + 16AB_ACCR + 17SOX + 18ASSETS + 
(2)
See the Appendix for complete definitions of the variables included in equation (2), which is
based on existing models of private investor financing and PE ownership (e.g. Chou et al. 2006;
Morsfield and Tan 2006; Katz 2009; Beuselinck et al. 2009).26 We compute the inverse Mills’
ratio for each firm-year observation, based on the estimated coefficients for equation (2), and
then include that variable in equation (1), the second stage of the Heckman estimation
procedure.27, 28
4. Sample Selection and Empirical Results
4.1 Sample Selection
Our initial sample consists of private firms that have publicly-traded debt. Because their
debt is public, these firms must file financial statements with the SEC, even though their equity
is privately-held. We follow Katz (2009) and select all firm-year observations on Compustat in
any of the 33 years from 1978 through 2010 that satisfy the following criteria: (1) the firm’s
stock price at fiscal year-end is unavailable, (2) the firm has total debt as well as total annual
26
See also Ball and Shivakumar (2005) and Givoly et al. (2010) for a similar methodological approach in the
comparison of private and public firms.
27
We estimate the Heckman (1979) two-stage procedure using Lee’s (1979) switching simultaneous equation (see
Maddala, 1983, Chapter 9). In the first-stage probit regressions, we obtain MacKelvey-Zavonia pseudo-R-squares
that range between 68 percent and 74 percent, which validates the relevance of our chosen explanatory variables.
28
To further evaluate self-selection concerns, we compare the tax avoidance of future PE-backed firms during the 5
years before they are taken private by PE firms to a sample of propensity score matched public firms that are never
private PE-backed. The results (untabulated) indicate that income tax avoidance does not significantly differ
between these two groups of public firms, which is not consistent with PE firms acquiring public firms with the
greatest potential tax savings.
21 revenues exceeding $1 million, (3) the firm is a domestic company, (4) the firm is not a
subsidiary of another public firm, and (5) the firm is not a financial institution or in a regulated
industry (SIC codes 6000-6999 and 4800-4900). To ensure that the sample includes only private
firms with public debt, we examine each firm and remove public firm observations (details
provided in Table 1, Panel A). We further categorize each firm as being in one of the following
categories: (1) management-owned, defined as firms that do not have a PE sponsor and are at
least 50 percent owned by founders, current and past named executive officers, and/or their
families, (2) PE majority-owned, defined as firms whose equity is majority-owned (i.e., more
than 50 percent) by PE firms, according to Thomson Financials VentureXpert, and (3) PE
minority-owned, defined as firms whose equity is minority-owned (i.e., less than or equal to 50
percent) by PE firms. The resulting sample consists of 2,628 private firm-year observations and
549 private firms.
[PLACE TABLE 1 HERE]
Table 1, Panel B presents the industry composition of our sample of private firms with
public debt (i.e., the 2,628 firm-year observations in Panel A). Our sample of private firms with
public debt is generally consistent with the broader Compustat population over the same time
period. Only the proportion of private firms classified as retail firms is significantly different
from the Compustat population (25.1 vs. 9.4 percent).
We hand-collect managerial stock ownership data for our sample of private firms (where
available) as an alternative proxy for the separation of ownership and control. Specifically, we
hand-collect from SEC filings the total amount of stock owned by all named executive officers.29
29
Of the 549 firms included in our sample of private firms with public debt, we are able to hand-collect managerial
stock ownership data for 374 firms. For each firm, we collect stock ownership data for only one firm-year and
assume stock ownership remains relatively constant all years the firm remains in our sample, unless we determine
22 We then calculate the proportion of all outstanding common shares owned by these executive
officers and refer to this variable as MGR_STOCK. Because managerial stock ownership data is
only available for 374 of our 549 private firms, we use MGR_STOCK as our secondary proxy for
the separation of ownership and control, while MGMT_OWNED is our primary proxy.
We further hand-collect data on board composition and CEO characteristics from SEC
filings and the BoardEx database to determine if our sample of management-owned and PEbacked firms is similar to samples in prior research. To minimize the hand-collection process,
we randomly select three minority PE-backed firms for each year in our sample and match them
with both majority PE-backed and non-PE-backed private firms in the same year and the same
four-digit SIC code. If a match is not available in the same four-digit SIC code, we then find a
match in the same three- (or two-) digit SIC code. Thus, our sample of hand-collected data
includes 38 firms that are majority PE-backed, 38 firms that are minority PE-backed, and 38
firms that are non-PE backed.30 We also hand-collect tax footnote information from SEC
financial statement filings for these same three sets of firms to gain a better understanding of the
types of tax strategies adopted by sample firms (see Section 4.6).
4.2 Descriptive Statistics on Ownership, Board Composition, and CEO Characteristics
Table 2, Panel A presents the statistics regarding the proportions of stock owned by PE
firms, managers, and CEOs for 8 different types of private firms, including different types of PEbacked and employee-owned private firms.31 Asterisks indicate significant differences between
the mean and median amounts of stock owned by PE firms / management / CEOs at
that the firm experienced a change in ownership structure. In this case we collect stock ownership data for at least
one year after the change in ownership structure.
30
After removing 7 employee-owned firms from the non-PE-backed sub-sample, we are left with 31 managementowned firms.
31
Employee-owned firms are private firms that do not have a PE sponsor and whose equity is more than 50 percent
owned by employees. These firms are excluded from most analyses, except for Table 2, Panel A, Table 5, Panel D,
and Table 9, row 4.
23 management-owned firms (row 1) compared to other types of private firms (rows 2-8). Overall,
the results in Panel A indicate that managers not only own the majority of stock in managementowned firms (by definition), but their percentage stock ownership (mean = 66.4 percent, column
2) is also substantially larger at management-owned firms compared to all other types of firms.
Amongst PE-backed firms, managers own greater proportions of stock at firms that are minorityowned by PE firms (mean = 29.9 percent, row 4) than at other PE-backed firms. In fact, mean
(median) managerial stock ownership at majority-owned, PE-backed firms (row 3) is just 7.0
(3.7) percent. We also note that CEOs (column 3) account for the vast majority of stock
ownership by the management team (column 2) regardless of firm type. We conclude that our
sample of private firms exhibits substantial variation in managerial stock ownership, making it a
powerful setting to examine the impact of the separation of ownership and control on corporate
tax avoidance.
[PLACE TABLE 2 HERE]
Panel B contains statistics regarding board composition and CEO characteristics. The
results indicate that while 57 percent of board members at management-owned firms are insiders,
the proportions of insiders on boards at minority- and majority-owned, PE-backed firms are
significantly smaller at 45 and 30 percent, respectively. Panel B also indicates that PE firms
have 62 (39) percent representation on their majority-owned (minority-owned) portfolio firms’
boards. The chairman of the board is a representative of the PE firm owner 29 (48) percent of
the time, and the CEO is either nominated by or is affiliated with the PE firm owner 58 (44)
percent of the time at majority- (minority-) owned PE-backed firms. All of these statistics
24 clearly demonstrate PE firms’ abilities to monitor and control portfolio firms’ management and
boards of directors, consistent with the discussion in Section 2.2.32
Lastly, Panel B indicates that majority-owned, PE-backed firms have larger boards of
directors than both minority-owned, PE-backed firms and management-owned firms, and their
CEOs are less likely to serve as the chairman of the board. CEOs at PE-backed firms are also
often younger and have fewer years with the firm than CEOs at management-owned firms.
Moreover, the CEOs of both majority- and minority-owned, PE-backed firms are more likely to
receive stock option compensation than the CEOs of management-owned firms, consistent with
PE firms tying management compensation to portfolio firm performance. Overall, we conclude
that our sample of PE-backed firms is similar to samples examined in prior research.
4.3 Results for the Separation of Ownership and Control and Tax Avoidance
The evidence in Table 2 indicates that management-owned firms differ from PE-backed
private firms with respect to the proportion of managerial stock ownership and control of boards
of directors. Fama and Jensen (1983) assert that when equity ownership and corporate decisionmaking are concentrated in just a small number of decision-makers, these owner-managers will
likely be more risk averse and less willing to invest in risky projects, which we argue includes
income tax avoidance. To examine the impact of the separation of ownership and control on
corporate tax avoidance, we perform a propensity score matching procedure to mitigate concerns
that our results are driven by fundamental differences between management-owned and PEbacked firms.33 We first calculate propensity scores derived from a probit model, where the
32
The Pearson correlation between PE firm ownership and PE representation on the board of directors is 61.4
percent (p-value <0.001). We find no instances where PE firms have minority ownership but majority
representation on the board of directors. We find only four instances (out of 38) where PE firms have majority
ownership but minority representation on the board of director; however, in all four instances the chairman of the
board represents the PE firm (two firms) and/or the CEO was nominated by the PE firm (three firms).
33
Indeed, comparisons of management-owned and PE-backed firms reveal significant differences in many
characteristics. In particular, management-owned firms are significantly more profitable (e.g., RNOA, LOSS, and
25 dependent variable is a PE-backed indicator variable (PE_BACKED), and the model includes
variables that are significantly different between management-owned and PE-backed firms,
including RNOA, LOSS, NOL, LEV, MNC, INTANG, AB_ACCR, SALES, and ASSETS. We then
match each management-owned firm-year, one-to-one, to the PE-backed firm-year with the
closest propensity score without replacement.34 To ensure that each management-owned firmyear and its match are similar to each other, we restrict the two firms to have propensity scores
within 0.10 of each other. Descriptive statistics in Table 3, Panel A for the propensity score-matched managementowned and PE-backed private firms indicate that the matched samples do not differ with respect
to the control variables. However, in Panel B we observe significant differences for all four
measures of tax avoidance. These results uniformly suggest that management-owned firms
avoid less income tax than PE-backed private firms. Specifically, the mean and median amounts
of GAAP_ETR and CASH_ETR are statistically higher, while the mean and median amounts of
DTAX and SHELTER are significantly lower for management-owned firm-years. These results
are consistent with P1, which predicts that management-owned firms avoid less income tax than
PE-backed private firms.
[PLACE TABLE 3 HERE]
Panel C presents Pearson and Spearman correlations between the MGMT_OWNED
indicator variable and each measure of tax avoidance. Consistent with Panel B, the correlations
in Panel C indicate that management-owned firms avoid less tax than PE-backed firms. In
addition, most of the correlations between the measures of tax avoidance are as expected. In
NOL), have significantly lower leverage ratios, are less likely to have foreign operations (MNC), report lower total
and intangible assets (ASSETS and INTANG), but higher abnormal accruals than PE-backed firms. We also conduct
our analysis without propensity score matching and our conclusions are similar to those presented in this paper.
34
For additional insight into the propensity score matching procedure, see Marosi and Massoud (2008), Angrist and
Pischke (2009), or Armstrong et al. (2010).
26 particular, the ETR measures are positively correlated with each other, while DTAX and
SHELTER are positively correlated with each other. However, the CASH_ETR measure is not
highly correlated with DTAX, perhaps because the latter measure is designed to capture more
risky tax avoidance.
Table 4, Panel A presents results for our primary tests of P1, which predicts managementowned firms avoid less income tax than PE-backed private firms.35 The coefficients on all four
measures of tax avoidance are in the predicted directions and are statistically significant based on
two-tailed p-values, providing support for P1.36,37 The coefficient on MGMT_OWNED in the
CASH_ETR regression indicates that management-owned firms pay on average 13.3 percent
more income tax per dollar of adjusted pre-tax income than PE-backed private firms. This result
suggests a large economic difference in tax avoidance between management-owned and PEbacked firms.38 This economic difference is similar to findings in Kaplan (1989) and Kaplan and
Stromberg (2009) regarding the impact of PE firm owners on net income margins (e.g., 10
percent increase) and on the ratio of cash flow to sales (e.g., 40 percent increase) when public
firms are taken private. However, Guo et al. (2011) document more modest increases in
operating and cash flow margins for PE-backed firms. Thus, the economic significance of the
MGMT_OWNED coefficients should be interpreted with caution. We also note the coefficients
35
The number of observations differs across most regressions due to different data requirements. The GAAP_ETR
and CASH_ETR regressions are based on fewer observations (334 and 304, respectively) because these measures
require firms to have positive pretax income over a three-year time period.
36
Regressions where DTAX (SHELTER) is the dependent variable do not include INTANG and EQ_EARN (RNOA,
LEV, MNC, AB_ACCR, and ASSETS) because those variables are included in the estimation of DTAX (SHELTER),
and thus are orthogonal to DTAX (SHELTER), by design.
37
We include LOSS in the GAAP_ETR and CASH_ETR regressions because GAAP_ETR and CASH_ETR are scaled
by the sum of pretax net income over years t, t-1, and t-2, while LOSS captures whether year t’s net income is less
than zero.
38
However, these cash tax savings are not received in perpetuity, since PE-backed firms are generally sold or taken
public through an initial public offering within 5-7 years of the private equity acquisition.
27 on INV_MILLS are not significant, consistent with sample selection bias having little impact on
our estimates.39
Panel B presents results for our secondary tests of P1, where the proportion of stock
owned by managers (MGR_STOCK) is the variable of interest. The results are similar to those
for MGMT_OWNED in Panel A. Overall, the results in Table 4 are consistent with the
separation of ownership and control having a significant impact on the tax avoidance practices of
private firms.
[PLACE TABLE 4 HERE]
In Table 5, we summarize results for supplemental tests that further evaluate the impact
of the separation of ownership and control on corporate tax avoidance (control variables included
but not tabulated). In each of these tests we estimate equation (1) but vary the sample
composition and/or utilize a different proxy for the separation of ownership and control. Panel A
(B) compares the tax avoidance of management-owned and propensity score-matched, majorityowned (minority-owned), PE-backed firms. As expected, the coefficients on MGMT_OWNED
consistently indicate that management-owned firms avoid less income tax than majority-owned
(minority-owned), PE-backed firms, although the results are somewhat weaker in Panel B.
These findings support empirical predictions P1a and P1b. Given the greater managerial stock
ownership percentages in Table 2, Panel A for minority-owned, PE-backed firms relative to
majority-owned, PE-backed firms, we compare the tax avoidance of these two types of private
firms in Panel C. We find that minority-owned, PE-backed firms avoid significantly less income
tax than majority-owned, PE-backed firms, consistent with P1c. The coefficients on
MGMT_OWNED in Panel D indicate that management-owned firms avoid less income tax than
39
Stolzenberg and Relles (1997) argue that if selection bias is moderate then the two-step estimation approach can
generate estimates that are inferior to those from ordinary least squares estimation. In untabulated results we reestimate equation (1) after excluding INV_MILLS and our primary inferences are unchanged.
28 employee-owned firms, which have more diffuse stock ownership. These results support P1d.
Panel E compares the tax avoidance of management-owned firms and propensity score-matched
public firms. Consistent with empirical prediction P1e, the coefficients on MGMT_OWNED in
Panel E suggest that management-owned firms avoid less income tax than similar public firms.40
Lastly, we extend Table 4, Panel B and use MGR_STOCK as the proxy for the separation of
ownership and control in regressions based on management-owned (Panel F) and PE-backed
firms (Panel G) only. The results in Panels F and G indicate that within our samples of
management-owned and PE-backed private firms, respectively, tax avoidance is decreasing in
the proportion of stock owned by managers (MGR_STOCK). In sum, Table 5 provides
convincing evidence that our findings for P1 are robust to different proxies for the separation of
ownership and control and for firms with different ownership structures.
[PLACE TABLE 5 HERE]
4.4 Results for the Marginal Costs of Tax Avoidance at PE-Backed Firms
Empirical predictions 2a-2c consider alternative explanations for our findings in Table 4
that management-owned firms avoid less income tax than PE-backed private firms. In particular,
2a (2b) predicts that private firms owned by PE firms with many portfolio firms (large PE firms)
have lower marginal costs of tax avoidance and, as a result, avoid more income tax than other
PE-backed firms. We test these predictions by estimating equation (1) based on a sample that
only includes PE-backed firms and replace MGMT_OWNED with MANY_PE and LARGE_PE,
which are indicator variables, respectively, for whether a PE-backed firm is owned by a PE firm
with many portfolio firms or by a large PE firm. To be consistent with our empirical predictions,
the estimated coefficients on these variables should be negative (positive) in the GAAP_ETR and
40
We acknowledge that public and private firms are subject to substantially different financial reporting incentives,
which may influence our results.
29 CASH_ETR (DTAX and SHELTER) regressions. Table 6, Panel A (B) summarizes the results for
tests where MANY_PE (LARGE_PE) is the variable of interest (control variables included but
not tabulated). The estimated coefficients on MANY_PE in Panel A and on LARGE_PE in Panel
B are all statistically significant in the predicted directions.41 These results are generally
consistent with private firms that are owned by large PE firms and PE firms with many portfolio
firms having lower marginal costs of tax avoidance and, as a result, avoiding more income tax
than other PE-backed firms.
[PLACE TABLE 6 HERE]
Empirical P2c predicts that the difference in tax avoidance between small-sized, PEbacked and management-owned firms is larger than the difference in tax avoidance between
large-sized, PE-backed and management-owned firms, since large firms may enjoy economies of
scale to tax planning that are independent of PE ownership. We perform a difference-indifference analysis based on our sample of management-owned and PE-backed private firms to
empirically test this prediction. We first classify firms as small- vs. large-sized, where smallsized (large-sized) firms are those in the lowest (highest) quartile of net sales for all private
firms. We then remove firms that are not small- or large-sized and estimate the following
equation:
TAXi = 1PE_BACKED×SMALLi + 2MGMT×SMALLi + 3PE_BACKED×LARGEi + 4MGMT×LARGEi + 5RNOAi + 6LOSSi + 7NOLi + 8LEVi + 9INTANGi + 10MNCi + 11AB_ACCRi + 12EQ_EARNi + 13SALES_GRi + 14ASSETSi + 15SOXi + 16INV_MILLSi + ji YEARi + kl INDUSi + I, (3)
In this model specification, the coefficients on the four indicator variables (1 – a4)
capture the average value for each tax avoidance measure for each type of firm (e.g., small-sized
41
In untabulated analyses we estimate equation (1) based on our sample of PE-backed firms and include both
MANY_PE and LARGE_PE in the same regression. While all four coefficients on LARGE_PE remain significant in
the predicted directions, the coefficients on MANY_PE are no longer significant.
30 vs. large-sized, PE-backed firms), after controlling for numerous firm characteristics. We predict
that the difference in coefficients between small-sized, PE-backed and management-owned firms
is larger than the difference in coefficients between large-sized, PE-backed and managementowned firms. Table 6, Panel C summarizes the results of our difference-in-difference analyses
(control variables included but not tabulated). The differences are in the predicted directions,
although the results based on DTAX are not significant. For example, the difference in
CASH_ETR between small-sized firms (-0.108) is significantly larger than the difference in
CASH_ETR between large-sized firms (-0.010); P-value for F-test is 0.009. The results in Panel
C are consistent with small-sized firms experiencing the greatest tax savings from PE ownership.
Moreover, the results in Table 6 suggest that certain PE-backed firms may have lower marginal
costs of tax avoidance than other private firms, and thus may contribute to the findings of lower
tax avoidance at management-owned firms relative to PE-backed firms in Table 4.
4.5 Disentangling Ownership Concentration from the Marginal Costs of Tax Avoidance
We attempt to empirically disentangle the impact of the separation of ownership and
control from the marginal costs of tax avoidance by including proxies for both constructs in the
same regression. Table 7 summarizes the results for these analyses (control variables included
but not tabulated). While Panels A and B are based on our sample of PE-backed firms only,
Panel C (D) is based on our full sample of management-owned and PE-backed private firms for
which MGR_STOCK (MGMT_OWNED) is available. In Panel A, the proxy for the separation of
ownership and control is MINORITY_PE, since Table 2, Panel A reveals that managers at
minority-owned, PE-backed firms own significantly more stock than managers at majorityowned, PE-backed firms. We also include our proxies for the marginal costs of tax avoidance,
MANY_PE and LARGE_PE, in these regressions. The results indicate that among PE-backed
31 firms only, firms with relatively high managerial stock ownership rates (as proxied by
MINORITY_PE) avoid less income tax than other PE-backed firms, while firms with lower
marginal costs of tax avoidance (as proxied by LARGE_PE) avoid more income tax than other
PE-backed firms; however, the coefficients on MINORITY_PE are not significant for
CASH_ETR (t-statistic = 1.594) and SHELTER (t-statistic = -1.543) at the 10 percent significance
level. Panel B contains results for tests that include MGR_STOCK, rather than MINORITY_PE,
and thus has smaller sample sizes due to limited data availability. Inferences from these results
are similar to those shown in Panel A. However, the coefficients on MGR_STOCK are not
significant for GAAP_ETR (t-statistic = 1.308) and CASH_ETR (t-statistic = 1.548) at the 10
percent significance level.
[PLACE TABLE 7 HERE]
Table 7, Panel C summarizes results for regression specifications that are nearly identical
to those in Panel B. However, the sample includes both management-owned and PE-backed
firms and we add an additional control variable, PE_OTHER, to control for any systematic
impact that PE ownership may have on tax avoidance at private firms. We again find consistent
evidence that firms with lower marginal costs of tax avoidance (as proxied by both MANY_PE
and LARGE_PE) avoid more income tax than other private firms. We also find that firms with
less separation of ownership and control, as proxied by MGR_STOCK (Panel C) and
MGMT_OWNED (Panel D) avoid less income tax. The coefficients in Panel D provide insights
into the relative impacts of the separation of ownership and control (MGMT_OWNED) and PE
ownership (MANY_PE and LARGE_PE) on income tax avoidance. For example, all else equal,
management-owned firms have CASH_ETRs that are 4.9 percent higher than the CASH_ETRs of
PE-backed firms, while firms that are owned by large PE firms have CASH_ETRs that are 12.6
32 percent lower than the CASH_ETRs of other private firms. Together, these findings indicate that
the separation of ownership and control, along with the marginal costs of tax avoidance, both
significantly influence corporate tax avoidance at private firms.
4.6 Tax Avoidance Strategies and the Utilization of Foreign Subsidiaries
To gain a better understanding of the tax strategies used by our sample of private firms,
we hand-collected detailed income tax data from SEC filings for 107 private firms (see sample
selection procedures in Section 4.1). Specifically, we collected data from Form 10-K statutory
reconciliation schedules, which reveal material sources of differences between effective and
statutory tax rates, and thus sources of variation in our tax avoidance measures. The results in
Table 8 indicate that compared to management-owned firms, PE-backed firms report more
negative statutory reconciliation items related to foreign taxes, intangible assets, tax-exempt
income (e.g. corporate-owned life insurance policies), and tax credits, consistent with PE-backed
firms relying on a variety of tax reduction strategies. [PLACE TABLE 8 HERE]
We further investigated the use of tax avoidance strategies that involve foreign
subsidiaries. Multinational corporations commonly reduce their worldwide tax burdens by
strategically locating operations in low tax countries, including “tax havens.”42 Following the
methodology in Dyreng et al. (2011), we calculate the number of countries in which firms
operate and the number of subsidiaries located in tax havens for various subsets of sample
firms.43 The first four rows compare management-owned firms to other private firms, while the
last three rows include comparisons amongst different PE-backed firms. The results in Table 9
42
In this paper, the term “tax haven” refers to a country that has been designated a “tax haven” by the Organization
for Economic Cooperation and Development (OECD), due to its exceptionally low income tax rates and other
favorable tax attributes relative to other countries.
43
We thank Scott Dyreng for allowing us to use his database.
33 indicate that management-owned firms have significantly fewer subsidiaries – and fewer
subsidiaries in tax haven countries – than other private firms. In addition, our evidence also
suggests that minority-owned, PE-backed firms have fewer subsidiaries in tax haven countries
than majority-owned, PE-backed firms. Lastly, the results in the bottom row indicate that PEbacked firms owned by large PE firms have significantly more subsidiaries – and more
subsidiaries in tax haven countries – than PE-backed firms owned by small PE firms. Overall,
the results in Table 9 suggest that tax avoidance through foreign operations is an important tax
planning tool for PE-backed firms.44
[PLACE TABLE 9 HERE]
4.7 Supplemental Analyses
4.7.1 Deletion of Firms with Negative Pre-Tax Income Although our calculation of GAAP_ETR and CASH_ETR require the deletion of firmyears if the sum of pre-tax income over years t-2 to year t is negative, we do not impose a similar
data requirement on the other measures of tax avoidance (i.e., DTAX and SHELTER). To further
evaluate whether our results are sensitive to the exclusion of firms with negative pre-tax income,
we impose a 3-year, positive pre-tax income data requirement on regressions where DTAX and
SHELTER are the dependent variables. Our results (untabulated) are qualitatively similar for this
smaller, more profitable sample of firms relative to those shown in all tabulated analyses. We
also note that the correlations between our four tax avoidance measures strengthen when we
44
We also hand-collected data regarding fees paid to auditors from SEC filings to determine whether PE-backed
firms paid more or less tax fees to their auditors compared to non-PE-backed firms. Tax fees typically include fees
for tax compliance, tax planning, and tax advice, which include assistance with tax audits and appeals, tax advice
related to mergers and acquisitions, employee benefit plans, and requests for rulings or technical advice from tax
authorities. The untabulated results indicate that PE-backed firms have a significantly higher mean value for tax
fees paid to auditors (0.00187, scaled by lagged total assets) than non-PE-backed firms (0.00069), consistent with
PE-backed firms investing more resources in tax planning.
34 require all sample observations to have positive, cumulative pre-tax income over a three year
time period.
4.7.2 Tax Benefits from Employee Stock Options
Graham et al. (2004) find that employee stock options (ESOs) generate significant tax
savings and reduce marginal tax rates for large firms, and thus are important non-debt tax
shields. While tax deductions related to ESOs reduce cash effective tax rates, they are not
directly reflected in GAAP_ETR, DTAX, or SHELTER. Consistent with PE firms tying portfolio
firm management compensation to performance, the CEOs of PE-backed portfolio firms more
frequently receive stock options than the CEOs of non-PE-backed firms (e.g. Table 2, Panel B
and Katz 2009). However, as pointed out by Kaplan and Stromberg (2009), the equity stake of a
portfolio firm manager is illiquid because the manager cannot sell portfolio firm equity or
exercise stock options until the firm is publicly-traded. Therefore, we do not expect stock
options to generate tax benefits for PE-backed firms.
Compustat data regarding ESO tax benefits (TXBCO and TXBCOF) is available for fiscal
years 2005 and thereafter. Although less than 15 percent of our sample observations report nonzero ESO tax benefits, the amounts that are reported are not statistically different between
management-owned and PE-backed firms. Overall, we conclude that ESO tax benefits do not
significantly influence our results.
5. Conclusions In this study we investigate the impact of organizational structure on corporate tax
avoidance. We take advantage of a unique sample of firms with privately-owned equity but
publicly-traded debt and examine whether variation in the separation of ownership and control
influences the tax avoidance of private firms with different ownership structures. Fama and
35 Jensen (1983) assert that when equity ownership and corporate decision-making are concentrated
in just a small number of decision-makers, these owner-managers will likely be more risk averse
and thus less willing to invest in risky projects. Because income tax avoidance is a risky activity
that can impose significant costs on a firm, we predict that firms with greater concentrations of
ownership and control (and thus more risk averse managers) avoid less income tax than firms
with less concentrated ownership and control. Our results are consistent with these expectations.
However, we also consider a competing explanation for these findings. In particular, we
examine whether certain private firms (i.e., those that are owned by private equity firms) enjoy
lower marginal costs of tax planning, which facilitate greater income tax avoidance. Our results
are consistent with the marginal costs of tax avoidance and the separation of ownership and
control both influencing corporate tax practices. Overall, these findings increase our
understanding of whether and how organizational structure influences corporate tax practices.
Our findings are subject to several limitations. First, corporate tax avoidance is difficult
to measure and, like those used in prior research, each of our four tax avoidance measures have
their own strengths and weaknesses and none are superior or inferior to the other three. The fact
that our results are consistent across all four measures implies that our findings are highly robust.
Second, although our multivariate regression models control for numerous firm characteristics
that account for variation in tax avoidance across firms, it is not possible to control for all
sources of variation. Thus, our results should be interpreted with caution in the event that we
have inadequately controlled for any variable that is correlated with ownership structure. Lastly,
our main results are based on a sample of management-owned and PE-backed private firms that
are required to file financial statements with the SEC. We assert that this research setting is ideal
for testing our primary hypothesis, because our sample of private firms exhibits substantial
36 variation in the separation of ownership and control but holds financial reporting requirements
constant across all firms. Although our sample of private firms is subject to less financial
reporting pressure than public firms, we acknowledge that PE-backed firms are likely subject to
somewhat greater financial reporting pressure than management-owned firms, since PE-backed
firms are typically sold or taken public 5-7 years after they are taken private. Nonetheless, the
negative association between managerial stock ownership and tax avoidance holds when we
repeat our tests based on samples of management-owned and PE-backed firms only, and when
we compare management-owned and employee-owned firms. We conclude that differences in
financial reporting incentives at management-owned and PE-backed firms do not drive our main
results.
Our study seeks to understand the fundamental firm characteristics that influence
corporate tax practice by relying on principle-agent theory to build a framework for
understanding how one specific feature of organizational structure, namely the separation of
ownership and control, impacts corporate tax practices. Our findings contribute toward a better
understanding of the impact of insider control and organizational structure on corporate tax
avoidance (e.g., Shackelford and Shevlin 2001) and complements recent research that examines
how agency costs and managerial incentives influence corporate tax practices (e.g., Desai and
Dharmapala 2006; Rego and Wilson 2012).
37 References
Acharya, V., M. Hahn, and C. Kehoe. 2010. Corporate governance and value creation evidence
from private equity. Working paper, London Business School.
Aghion, P., J. Tirole. 1997. Formal and real authority in organizations. Journal of Political
Economy 105 (1):1-29.
Angrist, J., and J. Pischke. 2009. Mostly harmless econometrics. Princeton, NJ: Princeton
University Press.
Armstrong, C., A. Jagolinzer, and D. Larker. 2010. Chief executive officer equity incentives and
accounting irregularities. Journal of Accounting Research 48 (2): 225-271.
Armstrong, C., J. Blouin, and D. Larker. 2012. The incentives for tax planning. Journal of
Accounting & Economics 53 (1-2): 391-411.
Baker, G., R. Gibbons, and K. Murphy. 1999. Informal authority in organizations. Journal of
Law, Economics, & Organization 15 (1): 56-73.
Ball, R., and L. Shivakumar. 2005. Earnings quality in UK private firms: Comparative loss
recognition timeliness. Journal of Accounting and Economics 39 (1): 83–128.
Ball, R. and L. Shivakumar. 2006. The role of accruals in asymmetrically timely gain and loss
recognition. Journal of Accounting Research 44 (2): 207–242.
Beatty, A. and D. Harris. 1998. The effects of taxes, agency costs and information asymmetry on
earnings management: A comparison of public and private firms. Review of Accounting
Studies 4 (3/4): 299-326.
Beuselinck, C., M. Deloof, and S. Manigart. 2009. Private equity involvement and earnings
quality. Journal of Business Finance and Accounting 36 (5/6): 587-615.
Bharath, S. T., J. Sunder, and S. V. Sunder. 2008. Accounting quality and debt contracting. The
Accounting Review 83 (1): 1–28.
Bova, F., Y. Dou, and O.-K. Hope. 2012a. Employee ownership and firm disclosure. Working
paper, University of Toronto.
Bova, F., K. Kolev, J. Thomas, and F. Zhang. 2012b. Non-executive employee ownership and
corporate risk-taking. Working paper, University of Toronto.
Cantillo, M., and J. Wright. 2000. How do firms choose their lenders? An empirical
investigation. Review of Financial Studies 13 (1): 155–189.
38 Cao, J. and J. Lerner. 2009. The performance of reverse leveraged buyouts. Journal of Financial
Economics 91 (2): 139-157.
Chen, S., X. Chen, Q. Cheng, and T. Shevlin. 2010. Are family firms more tax aggressive than
non-family firms? Journal of Financial Economics 95 (1): 41-61.
Cheng, A., H. Huang, Y. Li, and J. Stanfield. 2012. The effect of hedge fund activism on
corporate tax avoidance. Forthcoming in The Accounting Review.
Chou, D. W., M. Gombola, and F. Y. Liu. 2006. Earnings management and stock performance of
reverse leveraged buyouts. Journal of Financial and Quantitative Analysis 41 (2): 407437.
Cohen, D., A. Dey, and T. Lys. 2008. Real and accrual-based earnings management in the preand post-Sarbanes Oxley periods. The Accounting Review 83(3): 757-787.
Cornelli, F., and O. Karakas, 2008. Private equity and corporate governance: Do LBOs have
more effective boards? Working paper, London Business School.
Cotter, J. and S. Peck. 2001. The structure of debt and active equity investors: The case of the
buyout specialist. Journal of Financial Economics 59 (1): 101–147.
Cunningham, N. and M. Engler. 2008. The carried interest controversy: Let's not get carried
away. Tax Law Review, Cardozo Legal Studies Research Paper No. 211.
Demiroglu, C., and C. James. 2010. The information content of bank loan covenants. Review of
Financial Studies 23 (10): 3700-3737.
Denis, D., and V. Mihov. 2003. The choice among bank debt, non-bank private debt and public
debt: Evidence from new corporate borrowings. Journal of Financial Economics 70 (1):
3-28.
Desai, M., and D. Dharmapala. 2006. Corporate tax avoidance and high-powered incentives.
Journal of Financial Economics 79 (1): 145-179.
Dyreng, S., M. Hanlon, and E. Maydew. 2008. Long-Run Corporate Tax Avoidance. The
Accounting Review 83 (1): 61-82
Dyreng S., B. Lindsey, and J. Thornock. 2011. Exploring the role Delaware plays as a domestic
tax haven. Working Paper, Duke University.
Fama, E., and M. Jensen. 1983. Separation of Ownership and Control. Journal of Law and
Economics 26 (2, Corporations and Private Property: A Conference Sponsored by the
Hoover Institution): 301-325.
39 Fleischer, V. 2007. Taxing Blackstone. University of Illinois Law & Economics Research Paper
No. 07-036.
Fleischer, V. 2008. Two and twenty: Taxing partnership profits in private equity funds. New
York University Law Review 83 (1): 1-59.
Frank, M., L. Lynch, and S. Rego. 2009. Tax reporting aggressiveness and its relation to
aggressive financial reporting. The Accounting Review 84 (2): 467-496.
Fruhan, W. 2009. Role of private equity firms in merger and acquisition transactions. Harvard
Business School Technical Note # 9-206-101 (July): 1-15.
Gadiesh, O., and H. MacArthur. 2008. Lessons from private equity any company can use. Bain
& Company Inc, Boston MA, Harvard Business Press.
Gilson, R., and C. Whitehead. 2008. Deconstructing equity: public ownership, agency costs, and
complete capital markets. Columbia Law Review 108(1): 231-264.
Givoly, D., C. Hayn, and S. Katz. 2010. Does public ownership of equity improve earnings
quality? The Accounting Review 85 (1): 195-225.
Graham, J., M. Lang, and D. Shackelford. 2004. Employee stock options, corporate taxes, and
debt policy. Journal of Finance 59 (4): 1585-1618.
Guo, S., E. Hotchkiss, and W. Song 2011. Do buyouts (still) create value? Journal of Finance
66: 479-517.
Heckman, J. 1979. Sample selection bias as a specification error. Econometrica 47 (1): 153–162.
Hribar, P., and D. Collins. 2002. Errors in estimating accruals: Implications for empirical
research. Journal of Accounting Research 40 (1): 105–134.
Ivashina, V., and A. Kovner. 2011. The private equity advantage: Leveraged buyout firms and
relationship banking. Review of Financial Studies 24 (7): 2462-2498.
Jensen, M. 2007. The economic case for private equity (and some concerns). Keynote Slides,
Harvard University.
Jensen, M., W. Meckling. 1976. Theory of the firm: Managerial behavior, agency costs and
ownership structure. Journal of Financial Economics 3 (4): 305-360.
Jones, J. 1991. Earnings management during import relief investigations. Journal of Accounting
Research 29 (2): 192–228.
Kaplan, S. 1989. Management buyouts: Evidence on taxes as a source of value. Journal of
Financial Economics 24 (2): 217-254.
40 Kaplan, S. 2009. The future of private equity. Journal of Applied Corporate Finance 21(3): 8-20.
Kaplan, S., and A. Schoar. 2005. Private equity performance: Returns, persistence, and capital
flows. The Journal of Finance 60 (4): 1791-1823.
Kaplan, S., and P. Stromberg. 2009. Leveraged buyouts and private equity. Journal of Economic
Perspectives 23 (1): 121-146.
Katz, S. 2009. Earnings quality and ownership structure: The role of private equity sponsors. The
Accounting Review 84 (3): 623-658.
Klassen, K., 1997. The impact of inside ownership concentration on the tradeoff between
financial and tax reporting. The Accounting Review 72 (3): 455–474.
Knoll, M. 2007. The taxation of private equity carried interests: estimating the revenue effects of
taxing profit interests as ordinary income. Working Paper, University of Pennsylvania
Law School.
Lawton, A. 2008. Taxing private equity carried interest using an incentive stock option analogy.
Harvard Law Review 121 (3): 846.
Lee, L. 1979. Identification and estimation in binary choice models with limited (censored)
dependent variables. Econometrica 47 (4), 977-996.
Maddala, G. 1983. Limited-dependent and Qualitative Variables in Econometrics. Cambridge
University Press, Cambridge, UK.
Marosi, A. and N. Massoud. 2008. “You can enter but you cannot leave…”: U.S. securities
markets and foreign firms. Journal of Finance 63 (5): 2477-2506.
Manzon, G. and G. Plesko. 2002. The relation between financial and tax reporting measures of
income.” Tax Law Review 55 (2): 175 - 214.
Masulis, R., and R. Thomas. 2009. Does private equity create wealth? The effects of private
equity and derivatives on corporate governance. University of Chicago Law Review 76
(1): 219-259.
McGuire, S., D. Wang, and R. Wilson. 2011. Dual class ownership and tax a avoidance.
Working paper, Texas A&M University.
Mikhail, M. 1999. Coordination of earnings, regulatory capital and taxes in private and public
companies. Working Paper, Massachusetts Institute of Technology
41 Mills, L. and K. Newberry. 2001. The influence of tax and nontax costs on book-tax reporting
differences: public and private firms. Journal of the American Taxation Association 23
(1): 1-19.
Morsfield, S. and C. Tan. 2006. Do venture capitalists influence the decision to manage earnings
in initial public offering? The Accounting Review 81 (5): 1119-1150.
Nissim, D. and S. Penman. 2003. Financial statement analysis of leverage and how it informs
about profitability and price-to-book ratios. Review of Accounting Studies 8 (4): 531–559.
Penno, M. and D. Simon. 1986. Accounting choices: Public versus private firms. Journal of
Business Finance & Accounting 13 (4): 561-569.
Rego, S. 2003. Tax-avoidance activities of U.S. multinational corporations. Contemporary
Accounting Research 20 (4): 805-833.
Rego, S., and R. Wilson. 2012. Executive compensation, equity risk incentives, and corporate tax
aggressiveness, Journal of Accounting Research, 50 (3): 775-809.
Robinson, J., S. Sikes, C. Weaver. 2010. Performance measurement of corporate tax
departments. The Accounting Review 85(3): 1035-1064.
Shackelford, D. and T. Shevlin. 2001. Empirical tax research in accounting. Journal of
Accounting and Economics 31 (1-3): 321-387.
Smith, C. and R. Stulz. 1985. The determinants of firms’ hedging policies. Journal of Financial
and Quantitative Analysis. 20: 391-405.
Smith, C. and R. Watts. 1982. Incentive and tax effects of executive compensation plans.
Australian Journal of Management 7 (2): 139-157.
Stolzenberg, R. and D. Relles. 1997. Tools for intuition about sample selection bias and its
correction. American Sociological Review 62 (June): 494-507.
Stromberg, P. 2008. The new demography of private equity. Working paper, Swedish Institute
for Financial Research.
Stromberg, P. 2009. The economic and social impact of private equity in Europe: Summary of
research findings. Working paper, Stockholm School of Economics.
Wilson, R. 2009. An examination of corporate tax shelter participants. The Accounting Review
84 (3): 969-999.
42 Appendix: Variable Measurement
Measures of Tax Avoidance:
= Firm i’s GAAP effective tax rate, which equals total income tax expense
GAAP_ETR
(Compustat TXT), over years t-2 to t, divided by the sum of pre-tax income
(PI) minus special items (SPI) in year t-2 to t. If data limitations prohibit us
from using years t-2 to t, we next use years t-1 to t, followed by year t.
GAAP_ETR is set to missing when the denominator is zero or negative and
we winsorize GAAP_ETR to the range [0,1].
CASH_ETR
= Firm i’s cash effective tax rate, which equals cash taxes paid (TXPD), over
years t-2 to t, divided by the sum of pretax net income (PI) minus special
items (SPI) in years t-2 to t. If data limitations prohibit us from using years t2 to t, we next use years t-1 to t, followed by year t. CASH_ETR is set to
missing when the denominator is zero or negative and we winsorize
CASH_ETR to the range [0,1].
DTAX
= Firm i’s residual from the following regression, estimated by industry and
year: PERMDIFFit = 0 + 1 INTANGit + 2 UNCONit + 3 MIit + 4 CSTEit
+ 5 NOLit + 6 LAGPERMit + eit; where PERMDIFF = Total book-tax
differences – temporary book-tax differences = [{BI – [(CFTE +CFOR) /
STR]} – (DTE / STR)], scaled by beginning of year assets (AT); BI = pretax
book income (PI); CFTE = current federal tax expense (TXFED); CFOR =
current foreign tax expense (TXFO ); STR = statutory tax rate; DTE =
deferred tax expense (TXDI ); INTANG = goodwill and other intangible
assets (INTAN), scaled by beginning of year assets (AT); UNCON = income
(loss) reported under the equity method (ESUB), scaled by beginning of year
assets (AT); MI = income (loss) attributable to minority interest (MII),
scaled by beginning of year assets (AT); CSTE = current state tax expense
(TXS), scaled by beginning of year assets; NOL = change in net operating
loss carryforwards (TLCF), scaled by beginning of year assets (AT);
LAGPERM = PERMDIFF in year t-1. From 1980 to 1986 the STR is 46%,
for 1987 the STR is 40%, from 1988 to 1992 the STR is 34%, from 1993 to
2005 the STR is 35%.We winsorize DTAX to the range [-1,1].
SHELTER
= Probability that firm i engages in a tax shelter as defined by Wilson (2009),
where Compustat Tax Shelter = -4.86 + 5.20* Book Tax Differences +
4.08*Discretionary Accruals – 1.41*Leverage + 0.76*Size + 3.51*ROA +
1.72*Foreign Income + 2.42*R&D.
Private Firm Indicator Variables:
= 1 if the firm does not have a PE sponsor and at least 50 percent of the firm is
MGMT_OWNED
owned by founders, current and past named executive officers, and/or their
families, and 0 otherwise.
MGR_STOCK
= The ratio of stock owned by founders, current and past named executive
officers, and/or their families, to common shares outstanding.
PE_BACKED
= 1 if a PE firm has a majority or minority ownership stake in a private
company, and 0 otherwise.
43 MAJORITY_PE
= 1 if 50 percent or more of the firm is backed by PE firms, and 0 otherwise.
MINORITY_PE
= 1 if less than 50 percent of the firm is backed by PE firms, and 0 otherwise.
LARGE_PE
= 1 if the private equity firm that owns the portfolio firm is one of the
following: Carlyle Group, Blackstone, Warburg Pincus, KKR, Goldman
Sachs Private Equity, Cerberus Capital, Fortress Investment, Apollo Global,
Bain Capital, TPG Capital, 3i Group, Apax Partners, Thomas H. Lee,
Morgan Stanley Private Equity, and Welsh Carson Anderson & Stone and 0
for all other PE firms. PE firms are ranked according to total U.S. dollar
investment during the years 1980-2009. (Source: Thomson Financials,
VentureXpert.)
MANY_PE
=1 if the number of firms owned by the PE firm is greater than 200 and the ratio
of equity invested divided by number of firms owned is greater than $30
million and 0 otherwise.
EMPLOYEE_OWNED
= 1 if the firm does not have a PE sponsor and more than 50 percent of the
equity is owned by the firms’ employees, and 0 otherwise.
LARGE
= 1 if the firm’s sales are in the top quartile of net sales (SALE) for all private
firms and zero otherwise.
SMALL
= 1 if the firm’s sales are in the bottom quartile of net sales (SALE) for all
private firms and zero otherwise.
Control Variables and Other Variables of Interest:
= Firm i’s abnormal total accruals in year t computed derived from the modified
AB_ACCR
cross-sectional Jones (1991) model. To estimate the model yearly by twodigit SIC code, we require that at least 10 observations be available. The
regression is: TACCj,t / TAj, t–1 = a1*[1 / TAj, t–1] + a2*[(ΔREVj, t – ΔTRj,
t)/TAj, t–1] + a3*[PPEj, t / TAj, t–1] where: TACC is total accruals for firm j in
year t, which is defined as income before extraordinary items (IBC) minus
net cash flow from operating activities, adjusted to extraordinary items and
discontinued operations OANCF – XIDOC). For the years prior to 1988,
TACC is defined as Δ(current assets ACT) – Δ(current liabilities LCT) –
Δ(cash CHE) + Δ(short-term debt DLC) – (depreciation and amortization
DPC). To correct for measurement errors in the balance-sheet approach, we
eliminate firm-year observations with "non-articulating" events (Hribar and
Collins 2002). TA is the beginning-of-the-year total assets (lagged AT).
ΔREV is the change in sales in year t (SALE), PPE is gross property, plant,
and equipment in year t (PPEGT), and ΔTR is the change in trade receivables
in year t (RECTR). To control for the asymmetric recognition of gains and
losses, the modified Jones model is augmented with the following
independent variables: cash flow from operations in year t (CFt), a dummy
variable set to 1 if CFt <1 and 0 otherwise (DCFt), and an interactive
variable, CFt × DCFt (as suggested by Ball and Shivakumar 2006). CFt is
defined, for years after 1988, as cash from operations in year t adjusted for
extraordinary items and discontinued operations ( OANCF – XIDOC), and
prior to 1988 as funds from operations (FOPT) – Δ(current assets ACT) +
Δ(cash and cash equivalent CHE) + Δ(current liabilities LCT) – Δ(short-
44 term debt DLC). All variables are standardized by total assets at year-end t1.
ASSETS
= Natural logarithm of the total assets (AT) for firm i, at the end of year t.
EQ_EARN
= Firm i’s equity income in earnings (ESUB) in year t, scaled by lagged total
assets.
INTANG
= Firm i’s intangible assets (INTAN) in year t, scaled by lagged total assets.
INV_MILLS
= The inverse mills ratio from Heckman (1979) two-stage sample selection
correction procedure. In the first stage, we estimate the following probit
model (results not tabulated):
PE_BACKED = 0 + 1BVE + 2RNOA + 3Q_RATIO + 4OPER_CYCLE
+ 5FIRM_AGE + 6CASH + 7CAP_EXP + 8BIG_AUDIT +
9LOSS + 10NOL + 11LEV + 12MNC + 13INTANG +
14EQ_EARN + 15SALES_GR + 16AB_ACCR + 17SGA +
18ASSETS + 
BVE = book value of equity (Compustat CEQt + PSTKt + TXDITCt, scaled by
ATt-1); RNOA = profitability (defined as operating income divided by net
operating assets, see above), Q_RATIO = quick ratio [cash and short-term
investments (#CHEt) + total receivables (RECTt), scaled by current liabilities
(LCTt)], OPER_CYCLE = length of operating cycle [calculated as (yearly
average accounts receivable (RECTt)) / (total revenues (SALEt)/360) +
(yearly average inventory (INVTt)) / (cost of goods sold(COGSt)/360)],
FIRM_AGE = firm age (years since first appearance on Compustat), CASH =
cash holdings (CHEt scaled by ATt-1), CAP_EXP = capital expenditures
(CAPXt) scaled by ATt-1, LOSS = 1 if net income (NI) less than zero, and 0
otherwise; and BIG_AUDIT = an indicator variable for large accounting
firms (AUt). All other variables as defined above. We use the estimates from
the first-stage probit model to compute the inverse Mills’ ratio for each
sample firm-year. The inverse Mills’ ratio serves as a control variable in
equation (1), which is the second step of the Heckman estimation
procedure.45
LEV
= Firm i’s leverage in year t, measured as total long-term debt (DLTT) divided
by total assets;
LOSS
= 1 if firm i reports a loss, where loss is net income before extraordinary items
(IBC) and 0 otherwise.
MNC
= 1 if firm’s foreign pre-tax income (PIFO) or foreign income taxes (TXFO) is
positive or negative and 0 otherwise.
= 1 if firm i has net operating loss carryforwards (TLCF) available at the
NOL
45
Inverse Mills ratio is defined as: λ(Z) = φ(Ζ)/Ф(Z) if private or PE-backed = 1, and λ(Z) = -φ(Ζ)/(1 − Ф(Z)) if
private or PE-backed = 0, where: φ(Ζ) is the standard normal pdf, Ф(Z) is the standard normal cdf, and Z are the
estimates of the first stage probit model.
45 beginning of year t, and 0 otherwise.
RNOA
= Firm i’s operating income divided by net operating assets, where operating
income is net income (NI) + Δ(cumulative translation adjustment RECTA) +
after-tax interest expense (XINT) – after-tax interest income (IDIT) +
minority interest in income (MII). Net operating assets (NOA) are common
equity (CEQ) + debt in current liabilities (DLC) + total long-term debt
(DLTT) + preferred stock (PSTK) – cash and short-term investments (CHE)
– investments and advances (IVAO) + minority interest (MIB); (see Nissim
and Penman 2003).
SALES_GR
= Firm i’s sales growth, where sales growth is sales (SALE) at the end of year t
less sales at the beginning of year t divided by sales at the beginning of year
t.
SOX
= 1 if the fiscal year is 2004 and thereafter.
j INDUS
= 1 (0) if firm i is (is not) in industry j in year t, based on three-digit SIC codes.
j YEAR = 1 (0) if firm i is (is not) in year j.
46 FIGURE 1
Diagram of Typical Organizational Structure for a Private Equity Firm with One PE Fund and Four PE Portfolio Firms
Private Equity Firm
Investors
“Manager” (General Partner)
(Limited Partners)
LP, LLP, LLC
5 – 10%
90 – 95%
80% of gain*
20% of gain
(“carried interest”)
Private Equity
Investment Fund
(Upon Sale/IPO of
Portfolio Firm)
(Upon Sale/IPO of
Portfolio Firm)
LP, LLP, LLC
Portfolio
Portfolio
Portfolio
Portfolio
Firm #1
Firm #2
Firm #3
Firm #4
*
Approximately ten percent of the total gain is often distributed to the management team as part of performance-based compensation,
reducing the investors’ share to approximately seventy percent (Fruhan 2009)
47 TABLE 1
Sample Selection Procedures for Private Firms with Public Debt (1980 – 2010)
Panel A: Private Firms with Public Debt
a
“Potential” private firms with public debt (Compustat)
Eliminate firms that:
Do not have historical (non-prospectus) datab
Are public firms
Are subsidiaries of public firms
Are public spin-offs
Are involved in bankruptcy proceedings
Have insufficient information
Are foreign firms
Otherc
Subtotal of private firms with public debt
Eliminate firms that:
Are cooperatives, LPs, government-owned, and firms for
which ownership structure cannot be ascertained
Private firms with public debt:
Private firms that are majority-owned by PE firms
Private firms that are minority-owned by PE firms
Private firms that are owned by management
Panel B: Industry Classification for Private Firms with Public Debt
Industry Classification
Firm-Years
No. of Firm-Years
14,190
No. of Firms
3,699
(3,634)
(2,357)
(585)
(111)
(306)
(1,683)
(848)
(957)
3,709
(1,475)
(380)
(108)
(34)
(104)
(344)
(226)
(400)
628
(1,081)
(79)
2,628
1,559
312
757
549
350
71
128
Sample %
Compustat %
Agriculture
5
0.20%
0.40%
Mining & construction
22
0.80%
3.60%
Food
89
3.4%
4.20%
Textiles & Printing/Publishing
355
13.50%
10.10%
Chemicals
128
4.90%
4.90%
Pharmaceuticals
25
1.00%
3.30%
Extractive
61
2.30%
5.70%
Durable Manufacturers
751
28.60%
32.30%
Computers
86
3.30%
8.30%
Transportation
67
2.50%
4.10%
Utilities
0
0.0%
0.10%
Services
380
14.50%
13.60%
Retail
659
25.10%
9.40%
Total Observations
2,628
Industry classification is determined by primary SIC code as follows: Agriculture (0100-0999), Mining & Construction (1000-1999, excluding
1300-1399), Food (2000-2111), Textiles & Printing/Publishing (2200-2780), Chemicals (2800-2824, 2840-2899), Pharmaceuticals (2830-2836),
Extractive (2900-2999), 1300-1399), Durable Manufactures (3000-3999, excluding 3570-3579 and 3670-3679), Computers (7370-7379, 35703579, 3670-3679), Transportation (4000-4899), Utilities (4900-4999), Retail (5000-5999), and Services (7000-8999), excluding 7370-7379).
48 TABLE 2
Descriptive Statistics for Stock Ownership, Board Composition, and CEO Characteristics at Private Firms
Panel A: Average Stock Ownership Data for Management-Owned and PE-Backed Firms
Percentage of Stock Owned by:
PE Firms
Management
CEOs
(a)
(b)
(c)
Mean
Median
Mean
Median
Mean
Median
(1) MGMT_OWNED (N = 92)
0.0%
0.0%
66.4%
79.4%
52.3%
53.8%
(2) EMPLOYEE_OWNED (N = 13)
0.0%
0.0%
7.2% ***
2.6% ***
3.6% ***
1.2% ***
83.9%***
85.1%***
7.0% ***
3.7% ***
4.3% ***
1.8% ***
34.7%***
38.2%***
29.9%***
20.7% ***
19.5% ***
13% ***
83.3%***
87.8%***
6.5% ***
3.6% ***
4.6% ***
2.1% ***
79.9%***
83.2%***
9.4% ***
4.3% ***
5.8% ***
2.0% ***
85.4% ***
87.6% ***
5.1% ***
2.7% ***
3.5% ***
1.5% ***
77.0%***
80.9%***
10.6%***
5.1%***
6.6%***
2.4%***
(3) MAJORITY_PE (N = 258)
(4) MINORITY_PE (N = 24)
(5) MANY_PE (N = 49)
(6) FEWER_PE (N = 233)
(7) LARGE_PE (N = 89)
(8) SMALL_PE (N = 193)
*,**, ***
indicates significance at the 10%, 5%, and 1% level, respectively. Differences in means are tested for significance using a
two-tailed t-test; differences in medians are tested for significance using a two-tailed Wilcoxon signed rank test. An asterisk
indicates that the percentage of stock owned by PE firms (column a) / Management (column b) / and CEOs (column c) at
MGMT_OWNED firms (row 1) is significantly different than the corresponding percentage of stock owned by PE firms (column
a) / Management (column b) / and CEOs (column c) at each of the other firm types (rows 2-8). For instance, the mean amount of
stock owned by Management at MGMT_OWNED firms (row 1, column b: 66.4%) is significantly different than the mean amount
of stock owned by Management at EMPLOYEE_OWNED firms (row 2, column b: 7.2%).
49 TABLE 2 - Continued
Panel B: Board Composition and CEO Characteristics for Management-Owned and PE-Backed Firms
MAJORITY
MINORITY
MGMT
Diff
Diff
Diff
_PE
_PE
_OWNED
(1) – (2)
(1) – (3)
(2) – (3)
(1)
(2)
(3)
Number of Firms
38
38
31
Board Composition
Insiders
Mean
29.7%
44.7%
56.5%
-15.0%*** -26.8%***
-11.8%*
***
***
Median
28.6%
42.9%
50.0%
-14.3%
-21.4%
-7.1%*
PE Representatives
on Board
Mean
Median
62.4%
63.6%
39.2%
42.9%
23.2%***
20.8%***
Chair is PE Firm
Representative
Mean
28.9%
47.8%
-18.9%
CEO is Chair
Mean
48.9%
69.6%
66.7%
-20.7%*
-17.8%
2.9%
Board Size
Mean
Median
7.1
7.0
5.9
6.0
5.9
5.0
1.2**
1.0**
1.2*
2.0**
-0.1
1.0
CEO Characteristics
CEO Has an MBA
Mean
62.5%
55.6%
69.2%
6.9%
-6.7%
-13.7%
CEO Has Finance
Background
Mean
17.8%
26.1%
6.7%
-8.3%
11.1%*
19.4%*
CEO Age
Mean
53.6
56.7
56.3
-3.0
-2.6
0.4
CEO Years with the
Firm
Mean
Median
8.2
6.0
11.2
10.0
17.9
15
-3.0
-4.0*
-9.8***
-9.0***
-6.7***
-5.0**
CEO Has Stock
Options
Mean
71.1%
60.9%
30.0%
10.2%
41.4%***
30.9%**
CEO Nominated
by PE Firm Rep
Mean
57.8%
43.5%
*,**, ***
14.3%
indicates significance at the 10%, 5%, and 1% level, respectively. Differences in means are tested for significance using a
two-tailed t-test; differences in medians are tested for significance using a two-tailed Wilcoxon signed rank test. Insiders equals
the number of directors who serve as executives in the firm divided by total board size; PE Firms' Rep. equals the number of
directors who represent PE firms divided by total board size; Chair is PE is the percentage of firms for which the chairman is also
a general partner of the PE firm; CEO is Chair is the percentage of firms for which the CEO is the chairman of the board of
directors; Board Size is the total number of directors on the board; CEO has an MBA is the percentage of firms for which the
CEO hold an MBA degree; CEO has Finance Background is the percentage of firms for which the CEO has past experience as
investment banker, CFO, have a CPA or is a partners in a PE firm; CEO has Stock Options is the percentage of firms for which
the CEO received stock options as part of her/his compensation package; CEO Nominated by PE is the percentage of firms for
which the CEO is was either nominated or is affiliated with the PE firm owner.
50 TABLE 3
Descriptive Statistics that Compare the Tax and Non-Tax Characteristics of Management-Owned Private
Firms (Upper Rows, in Bold) and Propensity Score-Matched, PE-Backed Firms (Lower Rows, No Bold)
Panel A: Comparison of Non-Tax Characteristics
Num
25th
Obs
Percentile
Mean
Median
RNOA
241
0.063
0.114
0.104
241
0.081
0.124
0.117
75th
Percentile
0.162
0.164
Standard
Deviation
0.117
0.109
Difference between:
Mean
Median
-0.010
-0.013
LOSS
241
241
0.000
0.000
0.421
0.394
0.000
0.000
1.000
1.000
0.499
0.490
0.026
0.000
NOL
241
241
0.000
0.000
0.361
0.315
0.000
0.000
1.000
1.000
0.481
0.466
0.046
0.000
LEV
241
241
0.514
0.472
0.689
0.677
0.661
0.647
0.793
0.834
0.310
0.318
0.012
0.015
INTANG
241
241
0.000
0.000
0.206
0.211
0.117
0.128
0.264
0.336
0.229
0.255
-0.006
-0.011
MNC
241
241
0.000
0.000
0.461
0.473
0.000
0.000
1.000
1.000
0.499
0.500
-0.012
0.000
AB_ACCR
241
241
-0.042
-0.049
-0.011
-0.019
-0.008
-0.012
0.021
0.013
0.059
0.061
0.008
0.005
EQ_EARN
241
241
0.000
0.000
0.001
0.000
0.000
0.000
0.000
0.000
0.010
0.002
0.000
0.000
SALES_GR
241
241
-0.013
-0.055
0.263
0.225
0.029
0.017
0.125
0.101
0.869
0.854
0.039
0.012
ASSETS
241
241
5.125
5.388
5.904
5.946
5.719
6.017
6.383
6.503
1.116
1.188
-0.042
-0.298
SOX
241
241
0.000
0.000
0.241
0.241
0.000
0.000
0.000
0.000
0.428
0.428
0.000
0.000
51 TABLE 3 - Continued
Panel B: Comparison of Tax Avoidance Measures
Num
25th
Obs
Percentile
Mean
Median
GAAP_ETR
158
0.152
0.371
0.387
176
0.139
0.292
0.349
75th
Percentile
0.500
0.450
Standard
Deviation
0.275
0.181
Difference between:
Mean
Median
0.080***
0.038***
CASH_ETR
145
159
0.060
0.074
0.382
0.257
0.344
0.246
0.558
0.500
0.324
0.192
0.124***
0.098***
DTAX
241
241
-0.023
-0.021
0.026
0.062
-0.002
0.014
0.053
0.052
0.246
0.260
-0.036**
-0.017**
SHELTER
241
241
-2.181
-1.832
-1.383
-0.994
-1.391
-1.165
-0.753
-0.544
1.431
0.960
-0.389***
-0.226***
*,**,*** indicates significance at the 10%, 5%, and 1% level, respectively. Differences between means are tested for significance
using a two-tailed t-test; differences in medians are tested for significance using a two-tailed Wilcoxon signed rank test.
All variables are as defined in the Appendix. All continuous variables are winsorized at the 1st and 99th percentile.
Panel C: Pearson (Spearman) Correlation Coefficients for MGMT_OWNED and Tax Avoidance Measures
MGMT_OWNED
MGMT_OWNED
GAAP_ETR
CASH_ETR
DTAX
SHELTER
0.115
0.112
-0.078
-0.200
0.534
-0.090
-0.134
0.042
-0.235
GAAP_ETR
0.139
CASH_ETR
0.078
0.550
DTAX
-0.085
-0.081
0.046
SHELTER
-0.187
-0.149
-0.215
0.015
0.111
Bold indicates significance at the greater than 10 percent level based on a two-tailed t-test. All variables are as defined in the
Appendix.
52 TABLE 4
Results for Regressions that Compare the Tax Avoidance of Management-Owned Private Firms and
Propensity Score-Matched, PE-Backed Firms
Panel A: Proxy for Management Ownership Is MGMT_OWNED
GAAP_ETR
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
0.429
5.911
0.366
3.644
0.281
3.906
-0.807
-8.407
0.082***
3.267
0.133***
4.325
-0.032*
-1.781
-0.316***
-3.158
RNOA
0.006
0.040
-0.029
-0.166
-0.065
-2.153
LOSS
0.056
1.560
0.075
1.673
0.005
0.210
-0.964
-9.576
NOL
0.014
0.505
0.002
0.058
0.030
0.577
0.122
1.080
LEV
-0.068
-1.603
-0.022
-0.367
0.074
2.094
INTANG
-0.008
-0.153
0.045
0.755
0.544
2.345
MNC
0.114
4.621
0.111
3.572
-0.027
-0.181
AB_ACCR
-0.354
-1.557
-0.250
-0.875
0.019
0.822
EQ_EARN
-1.613
-1.022
-1.482
-0.635
-5.015
-0.464
SALES_GR
0.007
0.580
-0.024
-1.569
-0.014
-1.036
0.162
1.584
ASSETS
-0.027
-2.391
-0.030
-2.155
0.490
2.440
SOX
-0.022
-0.644
0.037
0.925
-0.039
-3.675
0.021
0.164
0.014
0.963
0.017
1.088
0.033
1.108
-0.005
-0.430
Intercept
MGMT_OWNED
INV_MILLS
2
Adjusted R
N
0.1294
0.1470
0.0759
0.2111
334
304
482
482
*,**,***
indicates significance at the 10%, 5%, and 1% level using a two-tailed t-test , respectively. Regressions include industry
and year indicator variables, which have not been tabulated. The t-stats have been adjusted to control for the clustering by
multiple firm observations. All variables are as defined in the Appendix.
53 TABLE 4 - CONTINUED
Panel B: Proxy for Management Ownership Is MGR_STOCK
GAAP_ETR
Intercept
MGR_STOCK
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
0.425
4.135
0.424
3.068
0.402
4.230
-0.673
-6.015
0.072
**
2.000
0.105
**
2.403
-0.030
*
-1.690
-0.573
***
-4.658
RNOA
0.057
0.388
0.172
0.946
-0.065
-1.784
LOSS
0.043
1.003
0.081
1.692
0.032
0.983
-1.039
-9.552
NOL
0.016
0.500
0.009
0.257
0.010
0.173
0.190
1.544
LEV
-0.087
-2.030
-0.051
-0.732
0.068
1.817
INTANG
-0.013
-0.225
0.028
0.435
0.287
1.178
MNC
0.104
3.741
0.095
2.801
-0.121
-0.696
AB_ACCR
-0.420
-1.606
-0.325
-1.013
0.031
1.128
EQ_EARN
-3.799
-2.713
0.279
0.119
7.956
0.602
SALES_GR
-0.004
-0.275
-0.042
-2.944
-0.021
-1.253
0.204
1.815
ASSETS
-0.022
-1.412
-0.038
-2.068
0.545
2.222
SOX
-0.016
-0.436
0.055
1.253
-0.056
-3.952
-0.061
-0.472
INV_MILLS
0.010
0.679
0.019
1.106
0.034
1.051
-0.012
-1.044
Adjusted R2
N
0.1013
0.1152
0.0808
0.2812
277
260
388
388
*,**,***
indicates significance at the 10%, 5%, and 1% level using a two-tailed t-test , respectively. Regressions include industry
and year indicator variables, which have not been tabulated. The t-stats have been adjusted to control for the clustering by
multiple firm observations. All variables are as defined in the Appendix.
54 TABLE 5
Summary of Results for Supplemental Regressions that Examine the Impact of the Separation of Ownership
and Control on Corporate Tax Avoidance
Panel A: Management-Owned vs. Propensity Score-Matched, Majority PE-Backed Firms:
GAAP_ETR
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
MGMT_OWNED
2
Adjusted R
0.076***
2.78
0.124***
3.87
-0.043**
-2.61
-0.368***
-3.55
0.1264
0.1698
0.0452
0.2428
296
270
424
424
N
Panel B: Management-Owned vs. All Minority PE-Backed Firms:
MGMT_OWNED
2
Adjusted R
0.035*
1.92
0.033*
1.69
-0.004
-0.89
-0.280***
-3.66
0.1354
0.1087
0.0171
0.2067
739
566
1,069
1,069
N
Panel C: Minority PE-Backed vs. All Majority PE-Backed Firms:
MINORITY_PE
0.034**
2.27**
0.040**
2.28**
-0.053***
-3.67
-0.148**
-2.05
Adjusted R2
0.0784
0.0518
0.0541
0.2123
N
1,138
1,022
1,871
1,871
Panel D: Management-Owned vs. All Employee-Owned Private Firms:
MGMT_OWNED
2
Adjusted R
0.015
1.63
0.041*
1.74
-0.035
-1.48
-0.580***
-5.56
0.1285
0.1040
0.0387
0.2800
768
559
997
997
N
Panel E: Management-Owned Private Firms vs. Propensity Score-Matched, Public Firms:
MGMT-OWNED
2
Adjusted R
0.077***
3.89
0.138***
5.28
-0.038***
-2.88
-0.388***
-5.58
0.5140
0.1701
0.0598
0.3685
468
433
644
644
N
Panel F: Managerial Stock Ownership (MGR_STOCK) at Management-Owned Firms Only
MGR_STOCK
Adjusted R2
0.023*
1.68
0.065*
1.94
-0.033
-1.59
-1.08***
-7.40
0.1516
0.1474
0.0296
0.3594
410
360
561
561
N
Panel G: Managerial Stock Ownership (MGR_STOCK) at PE-Backed Firms Only
MGR_STOCK
2
Adjusted R
N
0.030*
1.74
0.054*
1.81
-0.12***
-3.09
-0.287**
-2.17
0.0436
0.0372
0.0454
0.2335
852
804
1,378
1,378
*,**,***
indicates significance at the 10%, 5%, and 1% level using a two-tailed t-test , respectively. Regressions include the
following control variables: RNOA, LOSS, NOL. LEV, INTANG, MNC, AB_ACCR, EQ_EARN, SALES_GR, ASSETS, SOX,
INV_MILLS (Panels A and B), INDUS, and YEAR variables, which have not been tabulated. The t-statistics have been adjusted
to control for clustering by multiple firm observations. All variables are as defined in the Appendix.
55 TABLE 6
Summary of Results for Regressions that Test whether Certain PE Firms Reduce the Marginal Costs of Tax
Avoidance at PE-Backed Firms
Panel A: MANY_PE as Proxy for Lower Marginal Costs of Tax Avoidance (PE-Backed Firms Only)
GAAP_ETR
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
MANY_PE
Adjusted R2
N
-0.039***
-2.89
0.0736
1,138
-0.030*
-2.03
0.0502
1,022
0.033*
1.67
0.0497
1,871
0.218***
4.05
0.2199
1,871
Panel B: LARGE_PE as Proxy for Lower Marginal Costs of Tax Avoidance (PE-Backed Firms Only)
LARGE_PE
-0.048***
-3.61
-0.051***
-3.70
0.029*
1.79
0.383***
7.75
2
Adjusted R
0.0784
0.0591
0.0514
0.2428
N
1,138
1,022
1,871
1,871
Panel C: Difference-in-Difference Regression Analyses that Compare the Tax Avoidance of ManagementOwned and PE-Backed, Small- and Large-Sized Private Firms, where Small- (Large-) Sized Private Firms
are in the Bottom (Top) Quartile of Net Sales for All Private Firms (Excludes Firms not Classified at Smallor Large-Sized)
GAAP_ETR
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
PE_BACKED×SMALL
MGMT×SMALL
PE_BACKED×LARGE
MGMT×LARGE
Adjusted R2
N
0.290*** 5.045
0.294*** 7.185
0.304*** 4.920
0.367*** 9.547
0.7449
846
0.093
1.325
**
0.116
2.075
0.103
1.416
***
0.224
4.107
0.6489
693
-0.006
0.884
**
0.083
2.268
-0.009
-0.695
-0.002
-0.059
0.0765
1,327
Tests for Differences between PE- and Non-PE-Backed Firms within Firm Size Categories:
PE_BACKED×SMALL ̶
-0.108***
0.085***
-0.073*
MGMT ×SMALL
PE_BACKED×LARGE ̶
-0.014
-0.010
0.002
MGMT ×LARGEDifference
-0.059^
-0.097^^^
0.082
F-test (p-value)
0.083
0.009
0.111
*,**,***
-0.035
-0.590
***
-1.205
-21.56
0.533***
5.937
***
-1.989
-28.99
0.7042
1,327
0.784***
-0.568***
1.353^^^
0.001
indicates significance at the 10%, 5%, and 1% level using a two-tailed t-test , respectively. ^,^^,^^^ indicates significant at
the 10%, 5%, and 1% level based on an F-test. Regressions include the following control variables: RNOA, LOSS, NOL. LEV,
INTANG, MNC, AB_ACCR, EQ_EARN, SALES_GR, ASSETS, SOX, INV_MILLS (Panel C), INDUS, and YEAR variables, which
have not been tabulated. The t-statistics have been adjusted to control for clustering by multiple firm observations. All variables
are as defined in the Appendix.
56 TABLE 7
Results for Regressions that Examine the Dual Impact of the Separation of Ownership and Control and the
Marginal Costs of Tax Planning on Corporate Tax Avoidance
Panel A: Only Includes PE-Backed Firms, where MINORITY_PE Is the Proxy for the Separation of
Ownership and Control
GAAP_ETR
CASH_ETR
DTAX
SHELTER
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
Coeff
t-stat
MINORITY_PE
0.033**
2.443
0.028
1.594
-0.061***
-3.928
-0.102
-1.543
MANY_PE
-0.021
-1.397
-0.017
-0.958
0.015
1.177
0.087
1.225
LARGE_PE
-0.033
**
-2.227
-0.043
**
-2.730
0.045
**
2.467
0.376
***
7.058
Adjusted R2
0.0841
0.0617
0.0574
0.2436
N
1,138
1,022
1,871
1,871
Panel B: Only Includes PE-Backed Firms, where MGR_STOCK Is the Proxy for the Separation of
Ownership and Control
MGR_STOCK
0.016
1.308
0.025
1.548
-0.141***
-3.422
-0.125*
-0.035*
MANY_PE
LARGE_PE
2
Adjusted R
-0.042
**
-2.005
-0.039**
-2.455
**
-0.036
-2.175
-2.056
0.017
0.047
1.098
**
2.109
-1.654
0.039
0.357
1.013
***
0.0665
0.0569
0.0486
0.2642
852
804
1,378
1,378
N
6.158
Panel C: Includes Management-Owned and PE- Backed Private Firms, where MGR_STOCK Is the Proxy for
the Separation of Ownership and Control
MGR_STOCK
0.019*
1.791
0.019*
1.966
-0.063***
-3.547
-0.854***
-8.126
MANY_PE
-0.037**
-2.047
-0.064***
-3.352
0.037
1.551
0.135**
2.210
LARGE_PE
-0.042***
-2.803
-0.089***
-3.790
0.087**
2.075
0.230**
2.343
-1.426
***
-3.348
***
2.899
***
2.820
PE_OTHER
2
-0.008
-0.075
0.046
0.111
Adjusted R
0.0589
0.0879
0.0477
0.2731
N
1,262
1,164
1,939
1,939
Panel D: Includes Management-Owned and PE- Backed Private Firms, where MGMT_OWNED Is the Proxy
for the Separation of Ownership and Control
MGMT_OWNED
0.066**
2.118
0.049**
2.293
-0.052***
-2.638
-0.253**
-2.066
MANY_PE
-0.048**
-2.339
-0.055***
-2.507
0.022
0.795
0.037
0.413
LARGE_PE
-0.079***
-3.217
-0.126***
-4.778
0.037*
1.836
0.238**
2.552
PE_OTHER
**
-2.158
***
-3.527
*
1.749
*
1.825
2
-0.063
-0.112
0.033
0.160
Adjusted R
0.0811
0.0834
0.0518
0.2132
N
1,701
1,441
2,628
2,628
*,**,***
indicates significance at the 10%, 5%, and 1% level using a two-tailed t-test , respectively. Regressions include the
following control variables: RNOA, LOSS, NOL. LEV, INTANG, MNC, AB_ACCR, EQ_EARN, SALES_GR, ASSETS, SOX,
INV_MILLS (Panels C and D), INDUS, and YEAR variables, which have not been tabulated. The t-statistics have been adjusted
to control for clustering by multiple firm observations. All variables are as defined in the Appendix.
57 TABLE 8
Analysis of Items that Cause GAAP_ETR to Differ from the Statutory Tax Rate for Management-Owned and
PE-Backed Private Firms
Statutory Reconciliation
Items:
Foreign Tax Rate Differential
Mean
Median
State Tax Rate Differential
Mean
Median
Intangible Assets
Mean
Median
Tax-Exempt Income Items
Mean
Median
Nondeductible Expenses
Mean
Median
Change in Tax Reserve
Mean
Median
Tax Credits
Mean
Median
Other Items
Mean
Median
PE-Backed Firms
(N = 76)
Management-Owned
Firms (N = 31)
T-Statistic for
Difference
-0.042
0.000
-0.002
0.000
-1.145
-0.475
0.012
0.008
0.013
0.010
-0.135
0.443
-0.020
-0.000
0.036
0.000
-1.988**
-1.886*
-0.013
0.000
0.014
0.000
-1.860*
-1.722*
0.013
0.000
-0.001
0.000
1.774*
0.153
0.010
0.000
0.002
0.000
0.518
0.898
-0.021
0.000
0.000
0.000
-2.140**
-0.337
0.014
0.000
0.001
0.003
0.655
0.472
*,**,***
indicates significance at the 10%, 5%, and 1% level. Differences between means are tested for significance using a twotailed t-test; differences in medians are tested for significance using a two-tailed Wilcoxon signed rank test.
58 TABLE 9
Comparison of the Extent to which Different Types of Private Firms Rely on Corporate Structures that
Facilitate Corporate Tax Planning through Inter-Company and Cross-Border Transactions between
Subsidiaries
Number
of Obs
Number of
Subsidiaries
Mean
Median
Number of
Subsidiaries
in Tax Havens
Mean
Median
Number of
Countries
Mean
Median
MGMT_OWNED
PE_BACKED
Difference
93
465
8.56
19.87
-11.31***
4
7
-3***
1.36
3.01
-1.65***
0
1
-1**
4.01
6.83
-2.82***
2
4
-2**
MGMT_OWNED
MAJORITY_PE
Difference
93
433
8.56
19.54
-10.98***
4
7
-3***
1.36
3.02
-1.66***
0
1
-1**
4.01
7.46
-3.45***
2
4
-2**
MGMT_OWNED
MINORITY_PE
93
32
8.56
16.64
4
11
1.36
1.53
0
1
4.01
8
2
7
-8.08**
-7***
-0.17
-1
-3.99***
-5***
Difference
MGMT_OWNED
EMPLOYEE-OWNED
Difference
93
128
8.56
11.86
-3.3
4
6
-2
1.36
1.86
-0.5
0
0
0
4.01
6.89
-2.88
2
3
-1
MINORITY_PE
MAJORITY_PE
Difference
32
433
16.64
19.54
2.9
11
7
-4
1.53
3.02
1.49**
1
1
0
8
7.46
-0.54
7
4
-3
MANY_PE
FEWER_PE
76
389
28.23
17.55
6
8
3.57
2.79
0
1
7.22
7.55
3
4
10.68***
-2
0.78
-1
-0.33
-1
25.23
13.19
12.04**
9
6
3***
3.97
1.83
2.14***
1
0
1***
8.14
5.89
2.25***
4.5
3
1.5***
Difference
LARGE_PE
SMALL_PE
Difference
194
271
*,**,***
indicates significance at the 10%, 5%, and 1% level. Differences between means are tested for significance using a twotailed t-test; differences in medians are tested for significance using a two-tailed Wilcoxon signed rank test. All variables are as
defined in the Appendix.
59