Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
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 + ji YEARi + kl 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 + ji YEARi + kl 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