Pre-IPO Trusts, Private Information, and Corporate Spillover Michael Dambra University at Buffalo Matthew Gustafson Pennsylvania State University Phillip Quinn University of Washington March 18, 2017 Abstract: Twenty-three percent of CEOs place equity in tax-advantaged trusts prior to an IPO. Tax-advantaged trust ownership positively predicts one-year post-IPO abnormal returns, but only for firms with high informational asymmetry. Because trusts generate larger tax benefits when trust assets increase in value, this finding is consistent with CEOs having private information prior to their IPO. We also document a positive relation between trust ownership and corporate tax avoidance, suggesting that CEOs’ personal tax preferences spill over into corporate tax policy. Thus, trust establishment is a personal finance decision that reveals information about subsequent firm performance and policies. *Contact the authors at: Michael Dambra (corresponding author), [email protected]; Matthew Gustafson, [email protected]; and Phillip Quinn, [email protected]. We thank Brad Badertscher, Shane Heitzman, Jeff Hoopes, Bill Kross, Steven Savoy, Jake Thornock, Erin Towery, Steven Utke, David Weber, John Wertz, and Ryan Wilson for helpful comments and suggestions. We also thank Yan Zhao for excellent research assistance and Scott Dyreng for sharing subsidiary location data. The authors gratefully acknowledge the generous financial support from their respective institutions. Investors are focusing on Facebook’s offering price as the company prepares to go public as soon as next week. Tax specialists are paying attention to something else: how half a dozen of the firm’s luminaries, including founder Mark Zuckerberg, appear to be using a perfectly legal maneuver called a grantor-retained annuity trust, or GRAT, to avoid at least $200 million of estate and gift taxes. ─ Laura Saunders, The Wall Street Journal, May 11th 2012 1. Introduction The initial public offering (IPO) process is a watershed event for most corporate executives that has significant implications on their personal wealth. During this process, executives work closely with investment bankers, lawyers, the Securities and Exchange Commission (SEC), and equity analysts, as the company is marketed to and scrutinized by potential investors. At the same time, executives face some of the most important personal finance decisions of their lives. In this paper, we hand-collect data on one such decision: the percentage of a chief executive officer’s (CEO) shares in his firm that he places into a taxadvantaged trust. As suggested by the quote above, establishing a tax-advantaged trust prior to an IPO can result in substantial personal tax savings. Establishing a trust, however, comes with costs, such as legal fees, trustee fees, and potentially a reduction in control. We exploit this rare ability to observe a large scale personal finance decision of CEOs at the time of the IPO to provide new evidence on (1) whether CEOs have private information at the time of the IPO, and (2) whether CEOs’ large scale personal tax decisions are predictive of corporate tax strategy. To investigate these questions, we hand-collect a sample of CEO characteristics, including trust establishments, from over 1,600 IPOs between 1997 and 2013. We develop a classification system which bifurcates our sample into two main types of trusts, non-taxadvantaged (e.g., living wills) and tax-advantaged (e.g., GRATs). Twenty-three percent of CEOs transfer some portion of share ownership into a tax-advantaged trust, while four percent place shares into a non-tax-advantaged trust. The average CEO with a trust places one-third of 1 their total shareholdings into it. Tax-advantaged trust use is most common when the tax benefits of doing so are highest, such as when combined federal and state estate tax rates are higher, or the CEO is a founder, and thus likely to have low tax bases. In contrast, most firm characteristics, including firm size and profitability, do not predict tax-advantaged trust use. Thus, trust establishment appears to be a personal finance decision that CEOs use to reduce personal taxes. Given that an important input into CEOs’ personal finance decisions is their private information (Seyhun, 1986; Jenter, 2005), pre-IPO trust establishments have the potential to provide new evidence on the extent to which a CEO has private information at the time of the IPO. In particular, if CEOs have private information, the establishment of a pre-IPO trust should be positively related to future returns. By placing pre-IPO equity into tax-advantaged trusts, executives can bequest shares in a manner that avoids future capital gains and estate taxes. The most significant tax saving benefit of such trusts, however, is realized when the assets placed in the trust (e.g., shares of stock at IPO) appreciate in value. In contrast, if the trust assets decline in value, CEOs still bear the costs of establishing a trust, along with potential gift taxes at the time of transfer. Thus, rational CEOs will evaluate their firm’s prospects and place more shares into a trust when they believe their firm is undervalued. Although existing literature documents some cases in which CEOs possess private information regarding future firm performance,1 the extent to which they have such information at the time of the IPO is an unanswered empirical question. The IPO process contains many features designed to mitigate informational asymmetries between management and investors. Not only do underwriters spend months conducting due diligence and marketing the offering, but 1 For example, see Bartov and Mohanram (2004), Jeng, Metrick, and Zeckhauser (2003), Jagolinzer (2009), and Dittmar and Field (2015). 2 underwriters also certify the offering (e.g., Campbell and Kracaw, 1980), and IPO underpricing incentivizes investors to acquire information (Chemmanur, 1993). Empirically identifying the extent to which CEOs have pre-IPO private information is complicated by the fact that observed pre-IPO trades by CEOs do not necessarily reflect their private information. For example, Ang and Brau (2003) find that CEOs take steps to conceal their pre-IPO sales in order to avoid sending a negative signal to investors. In contrast, CEOs have no incentive to hide their trust use, because trusts should represent a positive signal to the market. We argue that this makes trust use a more reliable signal of a CEO’s pre-IPO private information. Consistent with CEOs having private information at the time of the IPO, the abnormal stock returns of IPO issuers with CEOs using a taxadvantaged trust are approximately 10 (7) percent higher in the first year compared to other IPO issuers on average (at the median). The magnitude of outperformance is approximately 7 (9) percentage points larger if we include first day returns (i.e., IPO underpricing) and is increasing in the percentage of shares that CEOs place into tax-advantaged trusts – a one standard deviation increase in the percentage of shares that a CEO places into a tax-advantaged trust is associated with 4.1 percent higher first-year returns. If these abnormal returns are attributable to the CEOs’ private information, as opposed to risks that disproportionately affect firms with trusts (and are not fully controlled for by our size and book-to-market matching procedure), then the post-IPO abnormal returns associated with trust usage should dissipate over time as pre-IPO information becomes less relevant. We find evidence of such dissipation, as the abnormal returns associated with trust usage are concentrated in the first, and depending on the empirical specification possibly the second, year following the IPO. The market appears to incorporate this information within two years, as the association 3 between the percent of shares transferred into a tax-advantaged trust and subsequent abnormal returns is not statistically significant. Finally, we investigate whether the prevalence of private information is concentrated in the CEOs governing certain types of firms. We find that trusts are associated with future abnormal returns only for firms with high levels of informational asymmetry, such as young firms, R&D intensive firms, and those that are operating at a loss. These findings suggest (1) the IPO process subsumes any value relevant information the CEO has only for non-informationally sensitive firms, and (2) for informationally sensitive firms, the trust establishment decision incorporates the CEO’s private information. Our second question builds in part off the idea that CEOs’ personal finance decisions spill over into their corporate policies (e.g., Bartov and Mohanram, 2004; Jenter, 2005). Given that trusts proxy for personal tax avoidance, this logic suggests that there may be a positive relation between trust usage and corporate tax avoidance. Alternatively, CEO trust use may be negatively related to corporate tax avoidance if trust use increases the scrutiny of a firm’s tax policy (Dyreng, Hoopes, and Wilde, 2016; Mider, 2013). Indeed, Hanlon and Heitzman (2010) highlight the relation between personal and corporate tax avoidance as an unsettled empirical question. An important reason why this question has not been thoroughly answered in the literature is the lack of data on executives’ personal finance decisions. There are few measures of personal tax avoidance in the literature (Kopczuk, 2013; Gale and Slemrod, 2001), and the measures that do exist are limited, often involve illegalities, endogenous interactions with the board of directors, or require unique compensation contracts (Armstrong and Larcker, 2009). For example, Chyz (2013), who provides the only evidence that personal and corporate tax policies 4 are positively related, proxies for personal tax avoidance with stock option exercise backdating (Dhaliwal, Erickson, and Heitzman, 2009). We fill this void in the literature by introducing trust establishments as a new proxy for a CEO’s personal tax avoidance strategy. Trust establishments represent legal, large-scale personal tax decisions which are not as susceptible to the limitations noted above. We find evidence that CEOs with tax-advantaged trusts operate firms that more actively avoid corporate taxes. CEO trust usage is associated with lower long-run GAAP effective tax rates and cash taxes paid. CEO trust usage is also positively associated with the volatility of cash effective tax rates and unrecognized tax benefits. Given that these associations hold after controlling for common determinants of tax avoidance, this evidence is consistent with the joint hypothesis that trusts proxy for a CEO that actively manages his personal taxes and that CEOs’ personal tax avoidance positively relates to the aggressiveness of their corporate tax policies. Next, we examine a specific mechanism for achieving corporate tax avoidance: tax havens. We find that CEO trust usage is associated with greater tax haven usage, and we also find that CEO trust usage is positively associated with the probability of a firm reporting a dot tax haven subsidiary. The latter result is consistent with CEOs that use trusts also engaging in aggressive tax avoidance strategies to minimize their corporate tax bill. Further, we are unaware of any other studies documenting that the CEOs who use complex legal structures to avoid personal taxes do the same for the corporations they manage. These results contribute to the emerging literature on the relation between personal tax preferences and corporate behavior by identifying a new variable for individual tax preferences. More broadly, our paper contributes to the literature linking personal attributes or incentives to corporate outcomes (e.g., Jensen and Murphy, 1990; Bertrand and Schoar, 2003; 5 Malmendier and Tate, 2005; Dyreng, Hanlon, and Maydew, 2010; Ge, Matsumoto, and Zhang, 2011). Importantly, trust usage is publicly available, directly observable to all researchers, and not related to compensation decisions made by the board of directors. In this manner, our findings relate to the recent literature suggesting that CEO pilot licenses relate to corporate risk taking (Cain and McKeon, 2016) and innovation (Sunder, Sunder, and Zhang, 2017), and CEOs’ previous employment and off-the-job behaviors impact financial and reporting policies (Custodio and Metzger, 2014; Davidson, Dey, and Smith, 2015). Our paper also contributes to the literature on the extent to which CEOs have private information regarding future firm performance. Jeng, Metrick, and Zeckhauser (2003), Bartov and Mohanram (2004), and Jagolinzer (2009) all provide evidence that insiders earn abnormal returns on their personal trades, and Dittmar and Field (2015) show that firms successfully time their open market share repurchases. We show that this information advantage extends to the IPO setting, even though the IPO process is designed to mitigate such informational asymmetries. Finally, our paper sheds new light on executives’ gift and estate planning strategies. Anecdotal evidence suggests that tax-advantaged trusts have cost the federal government over $100 billion in lost tax revenues since 2000 (Lund, 2013). The CEO of Las Vegas Sands, Sheldon Adelson, alone saved $3 billion in gift and estate taxes from using tax-advantaged trusts (Mider, 2013). Given the lack of available data, however, research has been scarce in this area (Heitzman and Hanlon, 2010; Tsoutsoura, 2015). We observe that CEOs engage in more complex tax planning strategies in the presence of future estate tax liabilities, providing new evidence that personal tax liabilities appear to have a first-order effect on CEO behavior. 2. Trusts 6 This section describes how trusts operate in the broader context of estate planning and the benefits of transferring share ownership to trusts. When individuals die without formalizing plans to distribute their assets, state laws dictate the distribution of the individuals’ estates through court-supervised probate procedures (American Bar Association, 2012). Estate planning allows individuals, rather than state laws, to determine the distribution of their estates. Estate planning involves formalizing plans to transfer assets to preferred beneficiaries. 2.1 Background on trusts A grantor forms a trust with a trustee through a trust instrument, which establishes the trust’s beneficiaries and assets.2 The costs of establishing a trust include legal fees to draft the trust instrument, trustee fees, and tax preparation fees for preparing the trust returns. Anecdotal evidence suggests annual fees typically range from 0.5 to 2.0 percent, depending on the trust’s assets and complexity (Feldman, 2015). Transferring shares to tax-advantaged trusts may cause the CEOs to forfeit significant control rights, and in instances when the trust retains income, that income faces high tax rates. In our paper, we categorize trusts into two types. Tax-advantaged trusts (e.g., irrevocable trusts) move assets out of the immediate control of the grantor and exclude the assets from the grantor’s estate. If a gift exceeds certain monetary thresholds, gift taxes may be assessed based on the fair market value of the assets transferred at the time of the gift. Therefore, grantors who expect that their assets will appreciate in value often establish taxadvantaged trusts to not only remove the assets from their estate, but also remove all future asset appreciation. In non-tax-advantaged (e.g., revocable) trusts, the grantor retains ownership of the assets, which remain part of the grantor’s estate. Non-tax-advantaged trusts provide no immediate taxable event, but do offer other benefits to grantors, such as financial privacy. 2 The general term for a person who establishes a trust is a grantor or a settlor, and a trustee is the person or company that manages the trust’s assets. 7 Wealthy grantors use tax-advantaged trusts to achieve considerable savings on estate and gift taxes. One popular estate planning strategy to reduce future estate taxes is a grantor-retained annuity trust (GRAT). Under a GRAT, the grantor first places assets, such as shares of stock, in the GRAT in exchange for a note payable, with a minimum interest rate set by the IRS. During the term of the GRAT, annuity payments flow to the grantor in accordance with the parameters of the trust. If the value of the trust assets appreciates at a rate in excess of the interest rate over the life of the GRAT, then the net appreciation passes to the heirs free of gift or estate taxes. If the value of the shares declines or increases at a rate below the interest rate, however, the shares pass back to the grantor, and the heirs incur no loss or tax. The net result is that GRATs provide grantors with the opportunity to benefit from substantial tax savings, but only when the trust assets appreciate.3 Price and Donaldson (2015, §9.10) note several tax-advantaged trusts provide tax benefits similar to GRATs. 2.2 Trusts and initial public offerings As previously noted, tax-advantaged trusts are most useful when the shares that the CEO places in the trust appreciate in value. This aspect of tax-advantaged trusts makes establishing a trust before an IPO especially attractive. Serafin (2013) notes “For every company that goes public… many more employees and investors can find themselves sitting on a windfall -- and trusts can be a key tool in minimizing the taxes owed when cashing in on that windfall.” Before the IPO, private share valuations receive a liquidity discount of up to 30 percent versus an IPO 3 As an example, assume a CEO places $100 million of shares in a two-year GRAT with equal annuity payments. The shares appreciate 10% annually and the interest rate is 4.0%. The annuity that the CEO receives equals the 1−(1+.04)−2 present value of a two-period annuity with an interest rate of 4.0% (i.e., 1.8861 = ). After the first year, .04 the shares are worth $110 million, and the CEO receives an annuity payment of $53,019,458 (i.e., $100 million / 1.8861) and the remaining shares are worth $56,980,542. After the second year, the remaining shares, which have growth of 10% in the second year, are now worth $62,678,596, and the CEO receives the second annuity payment of $53,019,458. The remainder, $9,659,138 of shares, passes to the heirs free of gift or estate taxes. 8 offer price (Brady, 2006; Pratt, 2009). Further, immediately following the IPO, public shares generate a significant day one price appreciation on average. In contrast to the significant benefits of establishing a tax-advantaged trust before the IPO, there is no clear change in the costs of establishing a trust prior to an IPO. Legal fees are unlikely to depend on when the trust is established. Given that estate and gift taxes are assessed based on the fair market value of stock transferred to descendants, we view the pre-IPO venue as a powerful setting to investigate how executives utilize trusts to minimize personal taxes.4 3. Data and Setting 3.1 Sample Our sample begins with all IPOs in Thompson’s SDC database from 1997, when prospectuses became available on EDGAR, through 2013. We employ sample filters common to the IPO literature. We exclude financial industries (including REITs), IPOs with proceeds below $5 million, best efforts offerings, rights offerings, shell companies, limited partnerships, and firms with issue prices less than $5 per share or missing stock returns data from CRSP.5 Unlike most IPO studies, we further restrict our sample by requiring non-missing financial statement data for the fiscal year immediately prior to going public from Compustat. We also obtain subsidiary disclosures from Exhibit 21 of the 10-K from Scott Dyreng’s website (e.g., Dyreng and Lindsey, 2009), federal and state estate tax rate information from Wolters Kluwer, and IPOrelated information from SDC. Finally, to be included in our sample, we also require that the 4 A more general setting of how CEOs use trusts after IPOs may provide additional insights, but trust usage after IPOs will be less likely to capture personal tax avoidance as opposed to other factors. 5 Shobe (2016) notes that the tax arbitrage benefits of supercharging an IPO only exist when the pre-IPO firm is a partnership. Thus, our results are not confounded by supercharged IPOs (e.g., Edwards, Hutchens, and Rego, 2016). 9 CEO holds a non-zero share position in the company prior to the IPO. Our final sample contains 1,611 IPOs. 3.2 Identifying CEO Trust Use From this sample, we read the most recent prospectuses available on EDGAR as of the IPO date (typically Form 424B). We hand-collect CEO characteristics, including the number of common shares owned pre-IPO, age, and whether the CEO was listed as a founding member of the issuer.6 The notes of the ‘Principal and Selling Stockholders’ table contains trust information for each CEO. Although trusts are confidential agreements, the SEC requires that executives disclose all beneficial ownership of shares held by the executive and certain members of the executive’s family. We record the number of shares the CEO transfers into a trust. This disclosure allows us to identify the number of shares transferred to a trust by the CEO. Next, we use the name of the trust to determine whether the trust is likely to provide tax advantages. Per discussion with estate planning consultants, the name of the trust typically (but not always) indicates the type of the trust established. This procedure allows us to separate trusts that lack immediate tax advantages (e.g., living wills) from trusts with significant tax advantages (e.g., irrevocable trusts). We classify all trusts with names containing ‘revocable’, ‘living’, ‘marital’, and ‘community property’ as non-tax-advantaged trusts (NTA Trust). All other types of trusts, including annuity trusts, children’s trusts, generation-skipping trusts, gift trusts, descendent trusts, qualified terminable interest property trusts, and GRATs, are classified as taxadvantaged trusts (TA Trust). Table 1 contains detailed definitions for all variables used in our analysis. 6 If the issuing firm did not list a CEO in the prospectus, we collected the information on the first executive listed in the ‘Management’ section of the prospectus (such as the President). We refer to such all individuals as CEOs, even though their titles may differ slightly. 10 Appendix A provides an excerpt from the pre-IPO prospectus of Greenlight Capital Re, Ltd. (GLRE) from Form 424 dated May 24, 2007 to illustrate our hand-collected trust data. For GLRE, the CEO, David Einhorn, transferred 3,623,370 shares into the David M. Einhorn 2007 GRAT. This amount represents 100 percent of the executive’s total shareholdings. The Form 424 indicates that the transfer of stock into the trust occurred on April 20, 2007. The transfer of assets into the trust shortly before the IPO date is consistent with Brady (2006), who notes that the tax savings from trusts are concentrated before the shares become liquid at the initial public offering. 3.3 Descriptive statistics Table 2 provides descriptive statistics. Slightly over one fourth of CEOs hold some portion of their shares in trusts, which is consistent with existing evidence that a large portion of firm ownership is controlled by trusts (Villalonga and Amit, 2009; Gadhoum, Lang, and Young, 2005). Trusts hold 9.8% of the CEOs’ total shares (Trust). Conditional on using a trust, CEOs place 38.1% of their holdings into trusts, resulting in the average trust-held shares being worth $78.5 million at the IPO price (untabulated). According to our classification system, approximately 23.2 percent of CEOs transferred shares into tax-advantaged trusts, while less than 4 percent used non-tax-advantaged trusts. Figure 1 shows that trust utilization has been relatively consistent throughout our sample period with tax-advantaged trust establishments ranging from a low of 15 percent in 2006 to a high of 35 percent in 2011. The average (median) CEO in our sample is 48 (48) years old, holds $50.9 ($15.9) million worth of shares of the issuer prior to the IPO. Approximately 42 percent of CEOs are founding members of the firm. From Wolters Kluwer, we collect federal and state estate tax rates based on the IPO year and headquarter-state location of our issuers. Over our sample 11 period, federal estate taxes ranged from zero percent (in 2010) to 55 percent (from 1997 through 2001), and state estate taxes ranged from zero percent to 19 percent. Our sample provides both cross-sectional and time series variation in estate tax rates. Given these descriptive statistics, our sample provides unique insight into the tax planning strategies of high-net worth individuals. 4. Determinants of trust usage In this section we explore the determinants of trust ownership. This section serves several purposes. First, insofar as tax-advantaged trusts are significantly associated with tax incentives, this exploration provides support for our trust classification method. Second, trusts are frequently deployed by high net worth individuals and to our knowledge there is no large scale evidence on how trusts are used, especially among executives (Hanlon and Heitzman, 2010), making the determinants of trust usage interesting in their own right. Last, if trusts are used rationally, they may provide new insight into CEOs’ individual tax preferences and ultimately may relate to corporate-level outcomes. 4.1 Economic Determinants of Trust Usage Model We employ both an ordinary least squares (OLS) regression and a Tobit to investigate the determinants of trust usage: 𝑇𝑟𝑢𝑠𝑡 = 𝛽0 + 𝛽1 𝐿𝑛𝐴𝑔𝑒 + 𝛽2 𝐹𝑜𝑢𝑛𝑑𝑒𝑟 + 𝛽3 𝐿𝑛𝐸𝑞𝑢𝑖𝑡𝑦𝑊𝑒𝑎𝑙𝑡ℎ + 𝛽4 𝐸𝑠𝑡𝑎𝑡𝑒 𝑇𝑎𝑥𝑒𝑠 + 𝛽5 𝑇𝑜𝑝10𝐿𝑎𝑤𝑦𝑒𝑟 + 𝛽6 𝐷𝑒𝑏𝑡 + 𝛽7 𝑆𝑖𝑧𝑒 + 𝛽8 𝐿𝑜𝑠𝑠 + 𝛽9 𝐿𝑛𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑠 + 𝜑 𝑇 + 𝜀 (1) The benefits of using OLS include ease of interpretation and the ability to include categorical variables that can bias inferences when researchers use maximum likelihood estimation (Greene, 2004). Because our dependent variable is censored at zero and one, we additionally employ a Tobit. In addition, we control for firm leverage, size, profitability, and the 12 proceeds from the IPO sale, include year-quarter fixed effects, and winsorize our non-binary variables at the top and bottom one percent. Our first dependent variable is Trust, which is the percent of CEOs’ shares that are placed in a trust prior to the IPO. We also examine the results of our determinants model with the percent of CEOs’ shares placed in tax-advantaged trusts (TA Trust) and non-tax-advantaged trusts (NTA Trust) as dependent variables. A rational CEO will place shares into the trust only if the benefits exceed the costs. We motivate several likely determinants of trust usage, based on the premise that trust usage should be increasing in the tax benefits of establishing a trust. There are at least two reasons why the tax benefits of estate planning are likely more salient for older CEOs (LnAge). First, the uncertainty of the value of assets in executives’ future estate declines as the individual ages, which can provide a more accurate assessment of the net benefits of estate planning. Second, older individuals are typically more risk averse (e.g., Barker and Mueller, 2002; Yim, 2013; Serfling, 2014). Transferring shares to a tax-advantaged trust can generate an immediate tax liability, but mitigates the possibility of higher estate tax liabilities being owed by descendants in the future. Founder CEOs (Founder) are also likely to reap larger tax benefits from trusts than other CEOs. Founders’ long standing with the firm makes their tax basis lower on average. Because capital gains taxes are applied on the difference between the sale price of stock and the cash paid to acquire the stock, founders face substantial capital gains taxes when selling equity under normal circumstances. A tax-advantaged trust can allow a CEO to avoid capital gains (and possibly gift) taxes entirely, depending on the executive’s tax exemptions and the trust’s structure. Finally, we expect trust share transfers to be an increasing function of the CEOs’ wealth (LnEquityWealth), because estate taxes only apply when the value of an estate exceeds 13 certain thresholds and inter vivos gifts are more likely as a household’s net worth increases (McGarry, 2013; Poterba, 2001). Our most direct measure of the personal tax benefits of trust use is the combined federal and state estate tax rates (Estate Tax). CEOs that expect to incur higher estate tax rates have a greater incentive to place their ownership in a tax-advantaged trust, and thus we expect a positive relation between tax-advantaged trust holdings and Estate Tax. Finally, to examine whether top law firms are associated with the creation of taxadvantaged trusts, we include Top 10 Lawyer. Prior work finds higher deal completion (offer withdrawal) rates when the bidding (target) firm retains a top law firm, which is consistent with top law firms providing superior legal advice (Krishnan and Masulis, 2013). Insofar as top law firms provide their clients with superior legal advice, we expect top law firms will have greater incentives and ability to provide tax-advantaged trust compliance and planning services with a firm’s CEO during the IPO process. 4.2 Results for Economic Determinants of Trust Usage Table 3 displays the results from regressing overall trust usage on CEO and firm characteristics (column 1). For ease of interpretation, we focus our discussion on the OLS models, which are qualitatively similar to our Tobit analysis. Consistent with our predictions, when estate taxes are more salient, the executive transfers more private IPO shares into trusts. Both the executive’s age (coeff. = 0.121, p-value < 0.01) and equity wealth (coeff. = 0.299, pvalue < 0.01) are positive and statistically significant. Economically, a 10 percent increase in pre-IPO equity wealth results in 3 percent more of the executive’s shares being transferred into a trust. When CEOs are founding members of the issuing firm they transfer 3.5 percent more shares into trusts than non-founding members (coeff. = 0.035, p-value < 0.01). We also find 14 support for our predictions that trust utilization is increasing in applicable estate taxes (coeff. = 0.095, p-value < 0.05). In terms of economic significance, a standard deviation increase in estate taxes (approximately an 11 percent increase) results in a 1.1 percent higher proportion of shares CEOs transfer into a trust. We also explore the how firm characteristics prior to the IPO influence the CEO decision to transfer share ownership into trust entities. We find in column 1 that CEOs managing firms with less leverage (coeff. = -0.040, p-value < 0.01) are more likely to transfer their shares into trusts prior to going public. One possible explanation for this result is that leverage constrains a CEO’s ability to take advantage of investment opportunities, which makes them less willing to place shares into pre-IPO trusts (Lang, Ofek, and Stulz, 1996). Interestingly, CEO trust use is not significantly related to firm size (t-stat = 0.660), operating at a loss (t-stat = 0.612), or IPO proceeds (t-stat = -1.540). Interestingly, the only other firm characteristic that relates to trust establishment is having a top law firm (coeff. = 0.057, p-value < 0.01). Assuming that certain CEOs underutilize trusts, perhaps those individuals who are subject to informational constraints, this result is consistent with top law firms providing superior legal advice (Krishnan and Masulis, 2013). In columns (3) through (6) of Table 3, we explore how our personal tax incentives and firm characteristics influence the propensity to transfer shares to tax-advantaged trusts and nontax-advantaged trusts. As a validation of our trust classification procedure, the associations between personal tax preferences appears concentrated in our tax-advantaged trusts (TA Trust) in columns (5) and (6) and the inferences are statistically and economically similar to overall trust usage. Out of our five hypothesized predictions on personal trust planning, only one coefficient loads in the predicted direction for the non-tax-advantaged trusts (NTA Trust) in columns (3) and 15 (4). These results should be interpreted cautiously, however, given the small number of CEOs using non-tax-advantaged trusts. In sum, our results suggest that managers appear to use tax-advantaged trusts prior to an IPO when the tax-related incentives to doing so are the greatest. Given the results herein, we use TA Trust in our remaining tests to examine how personal tax strategies associate with CEOs’ private information and corporate-level tax strategies. 5. Trusts and Private Information Different strands of literature generate opposite predictions regarding the extent to which CEOs are likely to have private information prior to an IPO. On the one hand, there is a large literature establishing that managers possess private information in certain settings and that they use this private information when making both corporate and personal finance decisions. For example, Bartov and Mohanram (2004) find that executives exercise more options prior to periods of poor performance, and Dittmar and Field (2015) find that managers successfully time their share repurchases. On the other hand, there are reasons to suspect that this informational advantage may not extend to the IPO setting. Many features of the underwriting process are designed to mitigate exactly the type of informational asymmetries that lead to management having private information. The typical IPO process involves months of underwriter due diligence and an extensive marketing roadshow. Underwriters certify the offering with their reputation (e.g., Campbell and Kracaw, 1980), and significantly underprice the offering, which Chemmanur (1993) argues incentivizes investors to acquire firm-specific information. Answering this empirical question of the extent to which CEOs have private information at the time of the IPO is difficult. One reason for this difficulty is that the observable pre-IPO 16 trading behavior of CEOs is difficult to interpret. Ang and Brau (2003) find that CEOs take steps to conceal their pre-IPO sales in order to avoid sending a negative signal to investors. CEOs’ use of pre-IPO trusts offers a cleaner setting to investigate this question. The establishment of a tax-advantaged trust is a costly personal finance decision that CEOs benefit more from when they expect future stock performance to be favorable. Holding tax rates constant, the tax benefits of a tax-advantaged trust are maximized when the fair market value of their shares is lower now than it will be in the future. If CEOs understand this and have accurate private information about the firm’s future performance, then they will transfer more shares into a trust when they believe their firm to be undervalued, leading to a positive relation between trust usage and future abnormal returns. In Tables 4 through 7, we examine whether tax-advantaged trusts predict post-IPO returns to corroborate whether managers have some information on post-IPO value that is not incorporated in the IPO offer price. Table 4 presents descriptive statistics on the average postIPO returns in the first, second, and third years following the IPO, partitioned by whether the CEO places shares in a tax-advantaged trust. The first row of Table 4 shows that the first year post-IPO returns, beginning at the IPO offer price, are 16 percent higher for firms with CEOs using tax-advantaged trusts, compared to other firms, and that this difference in means is significant at the 5 percent level. The second row in Table 4 shows that the relation continues to be statistically significant at the 10 percent level after removing IPO underpricing from the first year returns. The positive relation between trust usage and higher average returns is consistent with either CEOs having private information regarding expected abnormal stock returns or with CEOs using trusts more when risk is high. The lack of a consistent significant relation between 17 tax-advantaged trust usage and the second and third year returns is more consistent with CEOs having private information (rather than a risk-based phenomenon). To further examine the plausibility of CEO private information as a driver of the positive association between trust usage and returns, we investigate whether the positive association persists using risk-adjusted abnormal returns. To accomplish this, we follow the large post-IPO performance literature, which begins with Ritter (1991), and compute post-IPO returns relative to a matched firm that is already public. Specifically, as in Brau, Couch, and Sutton (2012) and Lyon, Barber, and Tsai (1999), we compute post-IPO returns as the difference between an IPO issuer’s returns and the returns to a matched firm that has been public for at least 5 years, has market capitalization within 30 percent of the IPO firm, and has the most similar book-to-market ratio as of the first post-IPO quarterly filing. If a control firm delists, we splice in the second best match as of the original match date. The first row in Panel B of Table 4 shows that the positive relation between taxadvantaged trust usage and first year returns is robust to using a risk-adjusted measure of returns. The average trust firm outperforms its matched counterpart by 22 percentage points in the first year, which is 17 percentage points more than the average non-trust firm. This difference is statistically significant at the 1 percent level and is qualitatively similar when measuring median, instead of mean, returns or when excluding IPO underpricing from the first year returns. In this descriptive analysis, these differential abnormal return patterns continue into the second year following the IPO, but dissipate thereafter. One challenge in computing post-IPO abnormal returns is handling IPO issuers that leave the sample. The statistics in Table 4 require a firm to exist in CRSP until the end of a given year to be included in the statistic. This is not too important for one year returns as only 41 of 1,611 18 firms do not have CRSP returns at the end of the first year. However, longer-run returns are subject to more selection bias. Figure 2 descriptively investigates the relation between trusts and post-IPO returns using a different method. Here, we include all IPO issuers through the end of the third year. If an IPO issuer leaves the sample for any reason, we assume that the issuer earned zero abnormal returns for the remainder of three years.7 The line in Figure 2 plots the difference between the median performing trust issuer (in terms of abnormal stock returns) and the median performing non-trust issuer. Figure 2 suggests that virtually all of the outperformance for the typical trust issuer occurs in the first year. During the first year trust issuers outperform non-trust issuers by 20%, and over the next two years the two groups experience similar abnormal returns. In Table 5, we investigate whether this relation between trusts and post-IPO returns can be explained by other observable differences between firms that have CEOs with tax-advantaged trusts and those that do not. To this end, we regress post-IPO returns on the percentage of personal shares a CEO places into a tax-advantaged trust along with all of the firm- and CEOlevel control variables used in Table 3. The positive and statistically significant coefficient estimate on TA Trust indicates that trusts provide information about future abnormal returns. Taken together, columns 1 and 2 suggest that this information is incremental to and not highly correlated with other CEO characteristics sources. Adding control variables has little effect on the coefficient on TA Trust. In both columns the coefficient on TA Trust of between 0.19 and 0.22 indicates that a one standard deviation (or 0.213 unit) increase in the TA Trust variable is associated with a four to five percent increase in first year post-IPO returns. In columns (3) and (4) of Table 5, we find that the results are qualitatively similar with or without including underpricing in our measure of post-IPO abnormal returns. 7 In unreported tests, we find similar sample attrition rates between our trust sample and non-trust sample. 19 In Table 6, we examine whether CEOs’ private information persists in the second and third year following the IPO for our sample. Across the four columns of Table 6, only one coefficient is positive and statistically significant, providing further support to our conjecture that the abnormal returns we observe are more consistent with private information, rather than a riskbased result. The results thus far are consistent with CEOs having private information at the time of the IPO and that the market does not fully integrate this information into prices for some time, presumably because market participants do not fully understand the signaling value of trust establishment.8 We now turn to the natural question of which CEOs actually possess information that the market does not immediately incorporate into the IPO price. Specifically, we investigate whether this outcome is more likely for informationally sensitive firms. Table 7 presents a series of regressions in which we regress first year abnormal returns on the percentage of shares that a CEO places into a trust and its interaction with several proxies for informational sensitivity, including firm size, age, profitability, and R&D intensity. Columns 1 through 4 indicate that the positive relation between trust usage and future returns is concentrated in non-profitable, young, and R&D intensive firms. In Columns 5 and 6 we run a multiple regression and find that the profitability result becomes insignificant after controlling for the age and R&D intensity interactions. Taken together, these findings are consistent with CEOs possessing private information that is not incorporated into the IPO price, only for young, R&D intensive firms. 8 These results contrast with Edwards, Hutchens, and Rego (2016), who find that supercharging an IPO (an IPO structure which generates future tax deductions for shareholders) is fully incorporated into the IPO price for 49 issuers and does not generate statistically significantly different future abnormal returns versus non-supercharged IPOs. 20 Overall, the evidence is consistent with CEOs having some private information about future stock performance, when informational asymmetry is high, and rationally increasing their trust usage when this information is positive. Our evidence also suggests that executives rationally consider personal tax strategies during the IPO process. This finding complements existing literature that suggests there are costs to CEOs selling their shares at the IPO (Ang and Brau, 2003; Brau and Fawcett, 2006). 6. Trusts and Corporate Tax Behavior Given that CEOs appear to be acting rationally by establishing trusts to minimize future individual income taxes, our setting is well suited to identifying how individual tax preferences spill over into corporate tax behavior. The prior literature suggests that there is an association between managers’ private and corporate decision making (e.g., Bartov and Mohanram, 2004; Jenter, 2005), but given the paucity of individual-level data (beyond externally assigned compensation) these associations have been difficult to demonstrate empirically. Here, we use CEOs’ decisions to transfer shares into trusts as a proxy for their personal tax avoidance preferences to provide new evidence on whether CEOs’ personal tax preferences relate to future corporate tax strategy. Specifically, we conduct the following regression: 𝐶𝑜𝑟𝑝𝑜𝑟𝑎𝑡𝑒 𝑇𝑎𝑥 𝑂𝑢𝑡𝑐𝑜𝑚𝑒 = 𝛽0 + 𝛽1 𝑇𝐴 𝑇𝑟𝑢𝑠𝑡 + 𝛽2 𝐿𝑛𝐴𝑔𝑒 + 𝛽3 𝐹𝑜𝑢𝑛𝑑𝑒𝑟 + 𝛽4 𝐿𝑛𝐸𝑞𝑢𝑖𝑡𝑦𝑊𝑒𝑎𝑙𝑡ℎ + 𝛽5 𝐸𝑠𝑡𝑎𝑡𝑒 𝑇𝑎𝑥𝑒𝑠 + 𝛽6 𝑇𝑜𝑝 10 𝐿𝑎𝑤𝑦𝑒𝑟 + 𝛽7 𝑃𝑃𝐸 + 𝛽8 𝑅𝐷 + 𝛽9 𝑁𝑂𝐿 + 𝛽10 𝑀𝑢𝑙𝑡𝑖 + 𝛽11 𝐷𝑒𝑏𝑡 + 𝛽12 𝑆𝑖𝑧𝑒 + 𝛽13 𝐿𝑛𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑠 + 𝜑 𝑇 + 𝜀. (2) The independent variable of interest is TA Trust. In our empirical design, we conduct regressions including our full slate of control variables for all observable determinants for establishing a trust (e.g., CEO age, wealth, shareholding, etc.) and corporate tax-related controls from the prior literature. A potential concern with this empirical design is that by controlling for 21 determinants of trust use, we are only measuring abnormal or unobservable trust use, rather than actual trust use. Therefore, in Tables 8 and 9, we report additional results that exclude the determinants of trust usage and other control variables from the regression. Similar to our determinants regression, we control for year-quarter fixed effects. 6.1 Trusts and corporate tax strategy Behavioral consistency theory predicts that the individual preferences of executives will influence their behavior in implementing corporate strategies (e.g., Cronqvist, Makhija, and Yonker, 2012). This could result in a positive association between personal and corporate tax avoidance strategies. Alternatively, Dyreng, Hoopes, and Wilde (2016) suggest that public scrutiny of tax positions may reduce a firm’s propensity to engage in tax avoidance. Insofar as media scrutiny of CEO trust use (e.g., Mider 2013) leads to increased regulatory scrutiny of a firm’s tax avoidance, CEO trust use may be negatively related to corporate tax avoidance. This idea is consistent with the Chen, Chen, Cheng, and Shevlin (2010) finding that family firms are less tax aggressive than non-family firms because family firms view their firms as legacies to pass on to future generations and as such are concerned about IRS audits of aggressive corporate tax positions. Thus, the relation between personal and corporate tax avoidance strategies is an empirical question. One reason that there is limited evidence on whether executives’ personal tax preferences carry over to their firm’s corporate tax policy is a lack of data on executives’ tax decisions. Most existing proxies for executive tax behavior are based on incentives generated by a CEO’s compensation scheme (Desai and Dharmapala, 2006; Rego and Wilson, 2012; Gaertner, 2014; Armstrong, Blouin, Jagolinzer, and Larcker, 2015) or a firm’s ownership structure (Chen, Chen, Cheng, and Shevlin, 2010; Badertscher, Katz, and Rego, 2013; McGuire, Wang, and Wilson, 22 2014), both of which are influenced by the board of directors. For example, the only existing evidence of a link between personal and corporate tax strategy must be inferred from evidence in Chyz (2013), who finds that there is a positive relation between illegally backdating stock option exercises and the propensity for firms to engage in tax avoidance. However, it is unclear whether backdating option exercises are fundamental preferences of the executive, or rather a consequence of the corporate governance system dictated by the firm (Armstrong and Larcker, 2009; Biggerstaff, Cicero, and Puckett, 2015). The evidence in Table 3 suggests that trust ownership proxies for individual tax preferences, and is largely unrelated to firm characteristics. In addition, tax-advantaged trust usage is distinct from other commonly deployed proxies in the literature since the decision to transfer shares into a trust reflects a legal, conscious choice made by the CEO. Thus, our setting provides a unique perspective on how individual tax preferences relate to corporate tax outcomes. If personal tax preferences spill over into corporate tax strategy, TA Trust will be positively associated with corporate tax avoidance. To test our empirical predictions on the association between TA Trust and corporate tax strategy, we use five common proxies for corporate tax strategy as dependent variables in equation (2): the long-run GAAP effective tax rate, the long-run cash effective tax rate, long-run cash taxes paid relative to the statutory rate, volatility of the cash effective tax rate (σCETR), and unrecognized tax benefits scaled by assets (UTB). We measure the long-run GAAP effective tax rate (LR ETR) as the sum of the issuer’s income tax provision scaled by the sum of the issuer’s pre-tax income in the three most recent years after going public. We use LR ETR as a measure of tax avoidance, because it is a direct measurement of the tax rate on pre-tax GAAP earnings. The long-run cash effective tax rate (LR CETR) is our second measure of tax avoidance. Following 23 Dyreng, Hanlon, and Maydew (2008), we measure long-run cash tax avoidance as the sum of three years of tax expense cash taxes paid over the sum of three years of pre-tax income minus special items. The maximum LR ETR (LR CETR) is set to 1 and is missing when the denominator is negative. Our LR ETR and LR CETR measures are widely used in the literature, but both measures are unavailable for loss firms, and losses are common for IPO firms. To address this shortcoming, we use three additional measures of tax avoidance. Following Henry and Sansing (2014), we measure LR CTP as the sum of the issuer’s cash paid for income taxes less the statutory rate times the difference between pretax income and special items for the three most recent years after going public scaled by the sum of the market value of assets in the three most recent years after going public. In contrast to LR ETR and LR CETR, the LR CTP measure utilizes all firms in our sample with non-missing cash taxes paid, rather than firms with positive pre-tax income. Next, we measure σCETR as the standard deviation of the issuer’s cash income taxes paid (CETR)9 scaled by the sum of the issuer’s pre-tax income less special items for the three most recent years after going public. Prior literature uses σCETR to measure tax risk, which relates to the dispersion in tax outcomes (Guenther, Matsunaga, and Williams, 2016). Prior research finds greater tax avoidance is positively associated with tax risk, consistent with uncertain and risky tax positions being associated with a greater dispersion in realized tax outcomes (Neuman, Omer, and Schmidt, 2013). Finally, we use unrecognized tax benefits (UTB) following the IPO to examine tax avoidance activities. To measure LR UTB, we take the issuer’s average unrecognized tax benefits 9 To be included in this analysis, the CETR for each year must be non-missing. 24 scaled by total assets from the three most recent years after going public. Lisowsky (2010) reports the propensity to have a tax shelter is positively related to the contingent tax liability. If CEOs’ individual tax preferences complement their firm’s corporate tax strategy, we expect that TA Trust will be negatively associated with ETR, CETR and CTP, but positively associated with σCETR and LR UTB. If there is a substitution effect between trust establishment and corporate strategies, we would expect opposite results. We test our empirical predictions on the association between individual tax preferences and corporate tax outcomes in Table 8. We find evidence of a positive association between a CEOs’ preferences to avoid personal taxes and their long-term corporate tax strategies. In column (2) of Table 8, we find that TA Trust is significantly negatively associated with LR ETR (coeff. = -0.094, p-value < 0.05). A one standard deviation increase in TA Trust results in a 2 percentage point decrease in LR ETR. Inconsistent with CEO personal tax avoidance spilling into corporate tax avoidance, the coefficient estimate on TA Trust is not statistically significant in columns (3) and (4). In columns (5) and (6), we find tax-advantaged trusts are associated with significantly lower LR CTP. A one standard deviation increase in TA Trust results in a 0.5 percentage point decrease in the LR CTP. Given the mean value for LR CTP is 0.034 for our sample, the effect we document is equal to an economically significant fifteen percent change relative to the mean LR CTP. We find consistent evidence of a positive relation between CEO personal tax avoidance and corporate tax avoidance in columns (7), (8), and (10) of Table 8. In column (8), we find trust use is associated with significantly higher σCETR (p-value < 0.05). Finally, in column (10), we find evidence consistent with a positive relation between CEO tax preferences and corporate tax avoidance when measuring tax avoidance using LR UTB (p-value < 0.05). A one standard deviation increase in TA Trust is associated with a 13 basis point higher LR UTB, which represents an 25 economically significant 14 percent of the mean LR UTB of 90 basis points for our sample. Our collective results suggest that personal tax preferences revealed at a firms’ IPO are associated with their future corporate tax strategies. Specifically, CEOs that use trusts to manage their personal finances are also more aggressive in minimizing their corporate tax bill. 6.2 Trusts and tax haven usage In our final analysis, we examine the relation between CEO tax preferences and subsidiary locations in tax haven jurisdictions. In doing so, we provide evidence on a mechanism through which trust CEOs more aggressively reduce their corporate taxes and examine whether CEO trust establishment is associated with complex tax-based operating decisions. To empirically test the association between trust usage and corporate tax haven use, we use two proxies for haven usage: the number of subsidiaries which reside in countries reported as tax havens (Haven Subs) and, using a Probit, an indicator variable equal to one if a firm has a subsidiary in a dot tax haven jurisdiction (DOT).10 We follow Dyreng and Lindsey (2009) by using subsidiary data from Exhibit 21 disclosures and considering jurisdictions tax havens if they appear on at least three of four publically-available tax haven lists. In contrast to other havens, DOT haven jurisdictions are exceptionally small countries, such as the Cayman Islands, in which the primary benefit of locating a subsidiary there likely stems from tax savings, rather than other business purposes. In Table 9, we find a positive relation between a CEO’s preference to avoid personal taxes and tax haven usage. In column (2) we find a significantly positive relation between TA Trust and Haven Subs (coeff. = 0.215, p-value < 0.05)– a one standard deviation increase in trust 10 Refer to Table 1 for the countries identified as tax havens or as dot havens. We classify dot havens following Hines and Rice (1994) and Desai, Foley, and Hines (2006). 26 holdings leads to an approximate eight percent increase in the number of tax havens.11 Finally, consistent with CEO tax preferences being associated with more aggressive tax avoidance strategies at the firm, column (4) shows that there is a positive and statistically significant relation (p-value < 0.05) between TA Trust and DOT. Our results are similar whether or not we use our full models in columns (2) and (4) or limited models in columns (1) and (3). Overall, these results suggest managers symmetrically utilize separate legally-recognized entities, such as trusts and foreign dot tax havens, in their private and public lives as mechanisms to avoid taxation. 7. Conclusion The wealth and liquidity created during the IPO process forces executives to make some of the most important financial decisions of their life. One such personal finance decision is whether or not to place their equity into a trust prior to an IPO. Doing so can reduce taxes, but involves fees and can reduce an executive’s direct ownership of the firm. Using a hand-collected sample, we provide the first evidence that CEOs actively use trusts when incentives to avoid taxes are the highest. This finding highlights the first-order effect of taxes on CEO behavior, and sheds new light on how CEOs can utilize tax planning to avoid capital gains taxes from highly appreciated shares. More importantly, we exploit this setting to provide new evidence on two unsettled empirical questions. The first is whether or not CEOs have private information regarding future performance at the time of the IPO. Trust usage sheds light on this issue because trusts only generate net tax savings to the extent that the assets placed into a tax-advantaged trust appreciate in value. If CEOs have private information regarding future performance then rational CEOs will be more 11 Our results are qualitatively similar if we use the number of tax haven countries instead of tax haven subsidiaries. 27 likely to place undervalued assets into a trust, resulting in a positive relation between trust usage and future returns. We find evidence of exactly this, as IPOs with CEOs using trusts outperform other IPO issuers by 10 to 17 percent in the first year. This outperformance is concentrated in informationally deficient firms, such as those that are young and R&D intensive. Thus, CEOs appear to have private information that is incremental to what can be conveyed through the extensive due diligence and investor scrutiny that occurs during the IPO process. The second question we address is whether CEOs’ preference to avoid personal taxes is related to their aggressiveness in avoiding corporate taxes. Existing literature provides some evidence on the topic, but is limited due to the rarity with which researchers observe the personal tax decisions of CEOs. Trusts fill this void, as they represent an economically meaningful decision that is a legal and conscious effort to reduce personal taxes. Our results are consistent with CEOs possessing personal tax preferences, and those preferences spilling into corporate strategy. Trust usage is associated with a wide range of measures designed to capture corporate tax avoidance, such as lower long-run GAAP effective tax rates, lower long-run cash taxes paid relative to the statutory rate, and more unrecognized tax benefits. In addition, CEO trust usage is positively related to corporate tax haven subsidiaries, and dot tax havens, suggesting that managers incorporate separate legal entities in both their private and public affairs in an effort to mitigate taxes. Taken together, the evidence in this paper suggests that there is previously unrecognized information content in the personal finance decisions of executives. More importantly, these decisions contain value relevant information and are related to how CEOs manage their firm. 28 References American Bar Association, 2012. The American Bar Association Guide to Wills & Estates: Everything You Need to Know About Wills, Estates, Trusts, & Taxes. Random House Reference. Armstrong, C.S., Blouin, J.L., Jagolinzer, A.D. and Larcker, D.F., 2015. Corporate governance, incentives, and tax avoidance. Journal of Accounting and Economics, 60(1), 1-17. Armstrong, C.S. and Larcker, D.F., 2009. Discussion of “The impact of the options backdating scandal on shareholders” and “Taxes and the backdating of stock option exercise dates”. Journal of Accounting and Economics, 47(1), 50-58. Ang, J.S., Brau, J.C., 2003. Concealing and confounding adverse signals: Insider wealthmaximizing behavior in the IPO process. Journal of Financial Economics 67(1), 149-172 Badertscher, B.A., Katz, S.P. and Rego, S.O., 2013. The separation of ownership and control and corporate tax avoidance. Journal of Accounting and Economics, 56(2), 228-250. Barker III, V.L., Mueller, G.C., 2002. CEO characteristics and firm R&D spending. Management Science 48(6), 782-801. Bartov, E., Mohanram, P., 2004. Private information, earnings manipulations, and executive stock-option exercises. Accounting Review 79(4), 889-920. Bertrand, M., Schoar, A., 2003. Managing With Style: The Effect of Managers on Firm Policies. Quarterly Journal of Economics 118(4), 1169-1208. Biggerstaff, L., Cicero, D.C.., Puckett, A., 2015. Suspect CEOs, unethical culture, and corporate misbehavior. Journal of Financial Economics 117(1), 98-121. Brady, M., 2006. Addressing Personal Wealth Issues in an IPO. Journal of Wealth Management, 9(3), 31-37. Brau, J.C., Couch, R.B. and Sutton, N.K., 2012. The desire to acquire and IPO long-run underperformance. Journal of Financial and Quantitative Analysis, 47(3), 493-510. Brau, J.C. and Fawcett, S.E., 2006. Initial public offerings: An analysis of theory and practice. Journal of Finance, 61(1), 399-436. Cain, M. and McKeon, E., 2016. CEO Personal Risk-taking and Corporate Policies. Journal of Financial and Quantitative Analysis, 51(1), 139-164. Campbell, T.S., and Kracaw, W.A., 1980. Information Production, Market Signaling, and the Theory of Financial Intermediation. Journal of Finance 35(4), 863-882. Chemmanur, T.J., 1993. The Pricing of Initial Public Offerings: A Dynamic Model with Information Production. Journal of Finance 48(1), 285-304. Chen, S., Chen, X., Cheng, Q., and Shevlin, T., 2010. Are family firms more tax aggressive than non-family firms? Journal of Financial Economics 95(1), 41-61. Chyz, J.A., 2013. Personally tax aggressive executives and corporate tax sheltering. Journal of Accounting and Economics, 56(2), 311-328. Cronqvist, H., Makhija, A.K., Yonker, S.E., 2012. Behavioral consistency in corporate finance: CEO personal and corporate leverage. Journal of Financial Economics, 103(1), 20-40. Custodio, C., and Metzger, D., 2014. Financial expert CEOs: CEO’s Work Experience and Firm’s Financial Policies. Journal of Financial Economics 114(1), 125-154. Davidson, R., Dey, A., and Smith, A., 2015. Executives “Off-the-job” Behavior, Corporate Culture and Financial Reporting Risk. Journal of Financial Economics 117(1), 5-28. Desai, M.A. and Dharmapala, D., 2006. Corporate tax avoidance and high-powered incentives. Journal of Financial Economics, 79(1), 145-179. 29 Desai, M.A., Foley, C.F., Hines, J.R., 2006. The demand for tax haven operations. Journal of Public Economics 90(3), 513-531. Dhaliwal, D., Erickson, M. and Heitzman, S., 2009. Taxes and the backdating of stock option exercise dates. Journal of Accounting and Economics, 47(1), 27-49. Dittmar, A. and Field, L.C., 2015. Can managers time the market? Evidence using repurchase price data. Journal of Financial Economics, 115(2), 261-282. Dyreng, S.D., Hanlon, M., Maydew, E.L., 2008. Long-run corporate tax avoidance. Accounting Review 83(1), 61-82. Dyreng, S.D., Hanlon, M., Maydew, E.L., 2010. The effects of executives on corporate tax avoidance. Accounting Review 85(4), 1163-1189. Dyreng, S.D., Hoopes, J.L. and Wilde, J.H., 2016. Public pressure and corporate tax behavior. Journal of Accounting Research 54(1), 147 - 186. Dyreng, S.D., Lindsey, B.P., 2009. Using financial accounting data to examine the effect of foreign operations located in tax havens and other countries on US multinational firms' tax rates. Journal of Accounting Research 47(5), 1283-1316. Edwards, A., Hutchens, M., Rego, S.O., 2016. Supercharged IPOs: Rent Extraction or Signal of Future Firm Performance? Kelley School of Business Research Paper No. 16-14. Feldman, A. 2015. Trust Costs Go Up; Get Ready to Negotiate. Barron’s February 28, 2015. Gadhoum, Y., Lang, L.H. and Young, L., 2005. Who controls US?. European Financial Management, 11(3), 339-363. Gaertner, F.B., 2014. CEO After‐Tax Compensation Incentives and Corporate Tax Avoidance. Contemporary Accounting Research 31(4), 1077-1102. Gale, W.G., Slemrod, J.B., 2001. Rethinking the estate and gift tax: overview. National Bureau of Economic Research Ge, W., Matsumoto, D., Zhang, J.L., 2011. Do CFOs have style? An empirical investigation of the effect of individual CFOs on accounting practices. Contemporary Accounting Research 28(4), 1141-1179. Greene, W., 2004. The behaviour of the maximum likelihood estimator of limited dependent variable models in the presence of fixed effects. The Econometrics Journal 7(1), 98-119 Guenther, D.A., Matsunaga, S.R., Williams, B.M., 2016. Is Tax Avoidance Related to Firm Risk? The Accounting Review, Forthcoming. Hanlon, M. and Heitzman, S., 2010. A review of tax research. Journal of Accounting and Economics, 50(2), 127-178. Henry, E., Sansing, R.C., 2014. Data truncation bias and the mismeasurement of corporate tax avoidance. In: 2014 American Taxation Association Midyear Meeting. Hines Jr, J.R., Rice, E.M., 1994. Fiscal Paradise: Foreign Tax Havens and American Business. Quarterly Journal of Economics 109(1), 149-182. Jagolinzer, A.D., 2009. SEC Rule 10b5-1 and insiders' strategic trade. Management Science 55(2), 224-239. Jeng, L.A., Metrick, A., Zeckhauser, R., 2003. Estimating the returns to insider trading: A performance-evaluation perspective. Review of Economics and Statistics 85(2), 453-471. Jensen, M.C., Murphy, K.J., 1990. Performance pay and top-management incentives. Journal of Political Economy, 98(2), 225-264. Jenter, D., 2005. Market timing and managerial portfolio decisions. Journal of Finance, 60(4), 1903-1949. 30 Kopczuk, W., 2013. Taxation of Intergenerational Transfers and Wealth. Handbook of Public Economics, Vol. 5, 329 - 390. Krishnan, C., Masulis, R.W., 2013. Law firm expertise and merger and acquisition outcomes. Journal of Law and Economics 56, 189-226. Lang, L., Ofek, E. and Stulz, R., 1996. Leverage, investment, and firm growth. Journal of Financial Economics, 40(1), 3-29. Lisowsky, P., 2010. Seeking Shelter: Empirically Modeling Tax Shelters Using Financial Statement Information. The Accounting Review 85(5):1693-1720. Lund, B. 2013. The Man Who Saves CEOs Billions in Taxes. December 24, 2013. Lyon, J.D., Barber, B.M. and Tsai, C.L., 1999. Improved methods for tests of long‐run abnormal stock returns. Journal of Finance, 54(1), 165-201. Malmendier, U., Tate, G., 2005. CEO overconfidence and corporate investment. Journal of Finance, 60(6), 2661-2700. McGarry, K., 2013. The Estate Tax and Inter Vivos Transfers over Time. American Economic Review, 103(3), 478-483. McGuire, S.T., Wang, D., and Wilson, R.J., 2014. Dual Class Ownership and Tax Avoidance. Accounting Review 89(4): 1487-1516. Mider, Z. 2013. Accidental tax break saves wealthiest Americans $100 billion. Bloomberg December 17, 2013. Neuman, S.S., Omer, T.C., Schmidt, A.P., 2013. Risk and return: Does tax risk reduce firms’ effective tax rates. Working paper. Poterba, J., 2001. Estate and gift taxes and incentives for inter vivos giving in the US. Journal of Public Economics, 79(1), 237-264. Pratt, S.P., 2009. Business valuation discounts and premiums. John Wiley & Sons. Price, J.R., Donaldson, S.A., 2015. Price on Contemporary Estate Planning (2016). CCH. Rego, S.O. and Wilson, R., 2012. Equity risk incentives and corporate tax aggressiveness. Journal of Accounting Research, 50(3), 775-810. Ritter, J.R., 1991. The long‐run performance of initial public offerings. Journal of Finance 46(1), 3-27. Serafin, T. 2013. Protect IPO Windfalls With a Trust. Barrons. September 14, 2013. Serfling, M.A., 2014. CEO age and the riskiness of corporate policies. Journal of Corporate Finance 25, 251-273. Seyhun, H.N., 1986. Insiders' profits, costs of trading, and market efficiency. Journal of Financial Economics 16(2), 189-212. Shobe, G., 2016. Supercharged IPOs, the Up-C, and Private Tax Benefits in Public Offerings. University of Colorado Law Review, Forthcoming. Sunder, J, Sunder, S., and Zhang, J., 2017. Pilot CEOs and Corporate Innovation. Journal of Financial Economics 123(1), 209-224. Tsoutsoura, M., 2015. The effect of succession taxes on family firm investment: Evidence from a natural experiment. The Journal of Finance 70(2), 649-688. Villalonga, B. and Amit, R., 2009. How are US family firms controlled? Review of Financial Studies, 22(8), 3047-3091. Yim, S., 2013. The acquisitiveness of youth: CEO age and acquisition behavior. Journal of Financial Economics, 108(1), 250-273. 31 Appendix A Example of Hand Collection from IPO Prospectus This appendix provides excerpts from the beneficial ownership table of Greenlight Capital Re, Ltd., or Greenlight Re, prospectus filed on May 24, 2007. The CEO and his related trust entity are bolded for expositional purposes. For brevity, we have omitted certain extraneous detail from the prospectus not relevant to our variables of interest. Beneficial ownership of principal shareholders prior to the offering and the concurrent private placement Name and address of beneficial owner Number David Einhorn (2) Keren Ohr Lanoar “B” (3) Montpellier International Ltd. (4) Scoggin International Fund, Ltd. (5) Seneca Capital International Ltd. (6) United Congregation Mesorah (7) Tim Courtis (8) Leonard Goldberg (9) Barton Hedges (10) Alan Brooks (11) Frank D. Lackner (12) Joseph Platt (13) Daniel Roitman (14) Jerome Simon All directors and Named Executive Officers as a group (9 persons) * 3,623,370 1,500,000 2,000,000 1,100,000 1,250,000 1,500,000 94,065 283,667 147,633 61,667 68,667 75,667 145,670 — 4,500,406 % 16.73% 6.92% 9.23% 5.08% 5.77% 6.92% * 1.30% * * * * * * 20.77% Represents less than 1% of the outstanding Ordinary Shares. (2) Prior to this offering, David Einhorn owns 3,623,370 Class B Ordinary Shares…On April 20, 2007, David Einhorn transferred all of his 3,623,370 shares to the David M. Einhorn 2007 GRAT for which he retains beneficial ownership. 32 Figure 1 Proportion of CEOs with shares owned by a trust prior to public issuance This figure presents the time series average of the percentage of CEOs that transferred a proportion of their shares into a trust entity prior to their firm going public. The sample spans from 1997 through 2013 and includes all hand-collected observations from the prospectuses. The solid (dashed) line represents the proportion of CEOs transferring shares into (non-) taxadvantaged trusts. 33 Figure 2 Abnormal Performance of Trust Ownership Firms vs. Non-Trust Ownership Firms This figure plots the difference between the post-IPO abnormal returns of the median performing IPO issuer whose CEOs use a trust and the median performing IPO issuer whose CEO does not use a trust. In each case, the median performing IPO issuer is defined using abnormal returns. Abnormal are computed for each IPO issuer as the monthly compounded buy-and-hold returns minus similar returns for a size and book-to-market matched firm. A matched firm is the firm with the most similar book-to-market ratio as the issuer out of the set of firms that have been public for five years and have a market capitalization of no less than 70% and no more than 130% of the IPO issuers’ market capitalization. If matched firms leave the sample, then the returns of the next closest match (as of the original match date) are spliced in going forward. If an IPO issuer leaves the sample, then abnormal returns are assumed to be zero for the remainder of three years. 34 Table 1 Data Definitions Variable Name Trust TrustD NTA Trust NTA TrustD TA Trust TA TrustD LnAge Founder LnEquityWealth Estate Tax Top 10 Lawyer PPE Debt Size Loss LnProceeds Multi NOL Variable Definition (source in parentheses) The percent of CEOs’ shares that are owned by a trust in the issuing firm’s most recent prospectus prior to going public (IPO Prospectus). An indicator variable equal to 1 if the CEO transferred shares into a trust prior to going public, and zero otherwise (IPO Prospectus). The percent of CEOs’ shares that are owned by revocable (non-tax-advantaged) trust in the issuing firm’s most recent prospectus prior to going public (IPO Prospectus). We define non-tax-advantaged trusts as those trusts specifically titled as ‘Revocable Trusts’, ‘Living Trusts’, ‘Marital Trusts’, and ‘Community Property Trusts’ (IPO Prospectus). An indicator variable equal to 1 if NTA Trust > 0, and zero otherwise (IPO Prospectus). The percent of CEOs’ shares that are owned by a trust not classified as revocable in the issuing firm’s most recent prospectus prior to going public (IPO Prospectus). An indicator variable equal to 1 if TA Trust > 0, and zero otherwise (IPO Prospectus). The natural log of the age of the CEO in the issuing firm’s most recent prospectus prior to going public (IPO Prospectus). An indicator variable equal to 1 if the CEO was documented as a founding member of the issuing firm in the issuing firm’s most recent prospectus prior to going public, and zero otherwise (IPO Prospectus). This is an empirical proxy for CEO wealth. We measure CEO wealth as the natural log of 1 plus the inflation-adjusted value of the CEOs’ pre-IPO shares × IPO offer price (IPO Prospectus, SDC). The Federal and State combined estate tax rate in place for the year of the IPO issuance. The issuer’s state is determined based on the issuing firm’s headquarters. If CEO Equity Wealth (unadjusted for inflation and log transformation) is below the Federal Exclusion, the Federal portion of the estate tax is set to zero. The Federal Exclusion is the minimum estate size for which individuals will owe estate taxes (Wolters Kluwer). Indicator equal to one when the IPO firm retains a law firm that was in the top 10 law firms by total IPO proceeds in the prior year, and zero otherwise (SDC). The issuers’ property, plant, and equipment net of depreciation scaled by total assets in the most recent year prior to going public (Compustat). The issuers’ long-term debt and portion of long term debt in current liabilities scaled by total assets in the most recent year prior to going public (Compustat). The natural log of the issuers’ inflation-adjusted total assets in the most recent year prior to going public (Compustat). An indicator variable equal to 1 if ROA is negative (Compustat), and zero otherwise. The natural log of the issuers’ inflation-adjusted IPO proceeds in the most recent year prior to going public (SDC). An indicator variable equal to 1 if the issuing firm discloses non-US geographical segments from their first annual report following the IPO, and zero otherwise (Compustat Segment). The issuers’ net operating losses scaled by total assets in the most recent year prior to going public (Compustat). 35 Variable Name RD Total Count Total Subs Year 1 Returns Year 1 Returns (w/o UP) Year 2 (3) Returns Year t Abnormal Returns Haven Subs DOT LR ETR LR CETR LR CTP σCETR LR UTB Variable Definition (source in parentheses) The issuers’ research and development scaled by total assets in the most recent year prior to going public (Compustat). The natural log of one plus the total number of countries in which the firm reported a material operation in Exhibit 21 from their first 10-K following the IPO (Scott Dyreng’s website). The natural log of one plus the total number of material subsidiaries in Exhibit 21 from their first 10-K following the IPO (Scott Dyreng’s website). The daily compounded holding period returns from the IPO offer price to the price at the close of the 250th day of trading (CRSP). Equals the Year 1 returns, excluding the returns from the offer price through the end of the first day of trading (CRSP). The daily compounded holding period returns from the close of the 250th (500th) day of trading through the close of the 500th (750th) day of trading (CRSP). Year t returns minus the returns to a size and book-to-market matched firm over the same period. Matched firms are the firm with the closest book-to-market ratio in the first quarter after the IPO that has been public for at least five years and has a market capitalization that is at least 70% and at most 130% of the IPO firm’s market capitalization. If matched firms leave the sample, then the returns from the next best match as of the original date are spliced in to compute future returns. The natural log 1 plus the total number of material subsidiaries operating in tax haven countries in Exhibit 21 from their first 10-K following the IPO (Scott Dyreng’s website). An indicator variable equal to 1 (0) if the firm reports (does not report) a material subsidiary in Exhibit 21 from their first 10-K following the IPO operating in any of the following countries: Andorra, Anguilla, Antigua and Barbuda, Bahamas, Aruba, Bahrain, Barbados, Belize, Bermuda, Cayman Islands, Cook Islands, Costa Rica, Cyprus, Dominica, Gibraltar, Guernsey, Grenada, Isle of Man, Jersey, Kiribati, Liechtenstein, Luxembourg, Macau, Malta, Martinique, Mauritius, Mauru, Nauru, Netherlands Antilles, Niue, Saint Lucia, Saint Kitts and Nevis, Saint Vincent and the Grenadines, Turks and Caicos Islands, and Vanuatu (Scott Dyreng’s website). Total GAAP tax expense per dollar of pre-tax income in the three most recent years after going public. The maximum LR ETR is set to 1 and missing when the denominator is negative (Compustat). The sum of the issuers’ cash taxes paid scaled by the sum of the issuers’ pre-tax income less special items in the three most recent years after going public. The maximum LR CETR is set to 1 and missing when the denominator is negative (Compustat). Following Henry and Sansing (2014), the sum of the issuers’ cash income taxes paid less the statutory rate × (PI – SPI) for the three most recent years after going public scaled by the sum of the market value of assets in the three most recent years after going public. For expositional purposes, we multiply the fraction by 100 (Compustat). The standard deviation of the issuers’ cash income taxes paid scaled by the sum of the issuers’ pre-tax income less special items for the three most recent years after going public. The maximum CETR for each year is set to 1 and missing when aggregate pretax income less special items is negative. To be included in the sample, the CETR for each year must be non-missing (Compustat). The issuers’ average unrecognized tax benefits scaled by total assets from the three most recent years after going public (Compustat). 36 Table 2 Descriptive Statistics This table provides the descriptive statistics for all the variables used in main empirical analysis. Variable definitions are provided in Table 1. Variable Trust TrustD NTA Trust NTA TrustD TA Trust TA TrustD CEO Age (Raw) LnAge Founder CEO Equity Wealth (Raw) in millions Equity Wealth PPE Debt Size Loss LnProceeds RD Multi NOL Estate Taxes Top 10 Lawyer Total Count Total Subs Haven Count Haven Subs DOT Haven LR ETR LR CETR LR CTP σCETR LR UTB N 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 1,611 470 470 470 470 470 563 553 840 471 406 Q1 0.000 0.000 0.000 0.000 0.000 0.000 42 3.738 0.000 5.760 0.005 0.070 0.036 2.771 0.000 3.715 0.000 0.000 0.000 0.392 0.000 0.000 0.693 0.000 0.000 0.000 0.271 0.109 -0.004 0.052 0.000 37 Mean Median 0.098 0.000 0.258 0.000 0.021 0.000 0.038 0.000 0.077 0.000 0.232 0.000 48.438 48 3.865 3.871 0.424 0.000 50.947 15.872 0.047 0.015 0.216 0.136 0.335 0.214 3.784 3.593 0.472 0.000 4.22 4.177 0.221 0.075 0.554 1.000 0.663 0.000 0.393 0.438 0.143 0.000 1.099 1.099 1.499 1.386 0.447 0.000 0.585 0.000 0.047 0.000 0.316 0.366 0.277 0.255 0.034 0.000 0.141 0.106 0.009 0.004 Q3 0.007 1.000 0.000 0.000 0.000 0.000 54 3.989 1.000 42.116 0.04 0.285 0.515 4.743 1.000 4.701 0.308 1.000 0.491 0.438 0.000 1.792 2.303 0.693 1.099 0.000 0.402 0.366 0.013 0.18 0.012 σ 0.244 0.437 0.122 0.192 0.213 0.422 8.607 0.178 0.494 111.824 0.123 0.211 0.395 1.667 0.499 0.853 0.369 0.497 1.759 0.112 0.347 0.937 1.153 0.532 0.741 0.211 0.29 0.22 0.151 0.13 0.014 Table 3 Determinants of Trust Use Trust is the percent of CEOs’ shares that are owned by a trust in the issuing firm’s most recent prospectus prior to going public. NTA Trust is the percent of CEOs’ shares that are owned by a non-tax-advantaged trust in the issuing firm’s most recent prospectus prior to going public. TA Trust is the percent of CEOs’ shares that are owned by a trust classified as a tax-advantaged trust in the issuing firm’s most recent prospectus prior to going public. Independent variable definitions are provided in Table 1. Each model includes year-quarter fixed effects and industry standard errors that are clustered by three-digit SIC industries. ***, **, * signify statistical significance at the 1%, 5%, and 10% level, respectively. Dependent Variable Model Type LnAge Founder LnEquityWealth Estate Taxes Top 10 Lawyer Debt Size Loss LnProceeds Intercept N Adjusted (Pseudo) R2 (1) Trust OLS (2) Trust Tobit (3) NTA Trust OLS (4) NTA Trust Tobit (5) TA Trust OLS (6) TA Trust Tobit 0.121*** (3.407) 0.035*** (3.356) 0.299*** (4.686) 0.095** (2.017) 0.057*** (2.986) -0.040*** (-3.607) 0.004 (0.660) 0.009 (0.612) -0.014 (-1.540) -0.391*** (-2.835) 0.359*** (2.897) 0.196*** (4.889) 0.756*** (3.753) 1.014*** (4.128) 0.161*** (3.459) -0.226*** (-3.085) -0.002 (-0.076) 0.045 (0.897) -0.010 (-0.305) -2.312*** (-3.984) 0.024 (1.201) 0.009 (0.985) 0.065 (1.195) 0.008 (0.297) 0.009 (1.385) -0.007 (-1.452) 0.002 (1.039) 0.009 (1.315) -0.007** (-1.997) -0.063 (-0.822) 1.009*** (2.779) 0.208 (1.070) 0.825 (1.204) 0.225 (0.314) 0.155 (1.194) -0.473*** (-3.606) 0.051 (0.884) 0.239* (1.723) -0.197** (-2.042) -4.723*** (-2.698) 0.098*** (3.126) 0.027** (2.049) 0.229*** (4.313) 0.086** (2.154) 0.049** (2.388) -0.034*** (-3.870) 0.001 (0.250) 0.000 (0.021) -0.007 (-0.686) -0.333** (-2.557) 0.261** (2.131) 0.173*** (4.020) 0.600*** (4.907) 1.014*** (4.855) 0.149*** (2.999) -0.207*** (-3.200) -0.011 (-0.500) 0.003 (0.058) 0.014 (0.423) -2.180*** (-3.392) 1,611 0.049 1,611 (0.0839) 1,611 0.016 1,611 0.036 1,611 (0.0830) 38 1,611 (0.166) Table 4 Post-IPO Returns Descriptive Statistics This table presents unadjusted (Panel A) and risk-adjusted (Panel B) returns during the first three years following the IPO. Year 1 Returns equals the holding period returns from the IPO offer price to the price at the close of the 250th day of trading, Year 1 Returns (w/o UP) excludes the returns from the offer price through the end of the first day of trading, Year 2 Returns equals the holding period returns from the 251st through 500th days of trading, and Year 3 Returns represents the holding period returns from the 501st through 750th days of trading. The corresponding abnormal returns measures presented in Panel B are identical except that returns are benchmarked to a matched seasoned firm based on market capitalization and market to book ratio. Specifically, within the set of firms that have been public for at least five years and have market capitalization within 30% of a given IPO firm, we select the firm with the closest book-to-market ratio. Columns 1 and 2 present statistics for IPO issuers with CEOs using trusts, while Columns 3 and 4 present the same statistics for firms whose CEOs do not use trusts. Columns 5 and 6 present the differences in mean and median between the two groups. *, **, and *** represent t-statistics and Wilcoxon statistics representing statistically significant different means and medians at the 10%, 5%, and 1% levels, respectively. Panel A: Trust Use and Raw Returns Year 1 Returns Year 1 Returns (w/o UP) Year 2 Returns Year 3 Returns TA TrustD =1 (1) (2) Mean Median 0.47 0.16 0.11 -0.13 0.08 -0.19 0.08 -0.08 TA TrustD= 0 (3) (4) Mean Median 0.31 0.08 0.00 -0.17 0.01 -0.21 0.11 -0.13 Difference (5) Mean 0.16** 0.10* 0.06 -0.04 (6) Median 0.08** 0.04 0.02 0.05 Panel B: Trust Use and Abnormal Returns Year 1 Abnormal Returns Year 1 Abnormal Returns (w/o UP) Year 2 Abnormal Returns Year 3 Abnormal Returns TA TrustD =1 (1) (2) Mean Median 0.22 -0.03 -0.15 -0.39 -0.07 -0.36 -0.14 -0.33 TA TrustD= 0 (3) (4) Mean Median 0.05 -0.19 -0.25 -0.46 -0.20 -0.44 -0.15 -0.35 39 Difference (5) Mean 0.17*** 0.10* 0.13** 0.01 (6) Median 0.16*** 0.07*** 0.08** 0.02 Table 5 Individual Trust Strategy and Post-IPO Abnormal Stock Returns This table regresses risk-adjusted stock returns during the first year following the IPO on TA Trust, measured as the percentage of CEO shares in a tax-advantaged trust. In Columns 1 and 2 the dependent variable is Year 1 Abnormal Returns w/o UP, which equals the holding period returns from close of the first day of trading to the close of the 250th day of trading relative to a matched firm. To obtain a matched firm we take all firms that have been public for 5 years and have market capitalizations within 30% of the IPO firm and then select the firm with the most similar book-to-market ratio. The dependent variables in Columns 3 and 4 are identical except that they also include the first day’s returns (i.e., IPO underpricing). Standard errors are clustered by three-digit SIC industries. T-statistics are presented below the coefficients with *, **, and *** representing statistical significance at the 10%, 5%, and 1% levels, respectively. TA Trust (1) Year 1 Ab. Returns w/o UP 0.215*** (2.62) (2) Year 1 Ab. Returns w/o UP 0.192** (2.32) -0.032 (-0.38) 0.074 (1.59) -0.307 (-1.64) -0.416** (-2.43) -0.010 (-0.20) -0.037 (-1.19) 0.085*** (4.83) -0.059* (-1.70) -0.052 (-1.58) (3) Year 1 Ab. Returns 1,570 0.003 1,570 0.032 1,570 0.002 LnAge Founder LnEquityWealth Estate Taxes Top 10 Lawyer Debt Size Loss LnProceeds N Adjusted R2 40 0.216** (2.13) (4) Year 1 Ab. Returns 0.173* (1.73) -0.357*** (-4.17) 0.053 (1.15) 0.286 (1.08) -0.033 (-0.14) -0.007 (-0.13) -0.205*** (-3.56) -0.044* (-1.88) 0.011 (0.13) 0.212*** (4.97) 1,570 0.039 Table 6 Individual Trust Strategy and Post-IPO Abnormal Stock Returns This table regresses risk-adjusted stock returns during the first three years following the IPO on TA Trust, measured as the percentage of CEO shares in a tax-advantaged trust. In Columns 1 and 2 (3 and 4) the dependent variable is Year 2 (Year 3) Abnormal Returns, which equals the holding period returns from close of the 251st (501st) day of trading day to the close of the 500th (750th) day of trading relative to a matched firm. To obtain a matched firm we take all firms that have been public for 5 years and have market capitalizations within 30% of the IPO firm and then select the firm with the most similar book-tomarket ratio. Standard errors are clustered by three-digit SIC industries. T-statistics are presented below the coefficients with *, **, and *** representing statistical significance at the 10%, 5%, and 1% levels, respectively. TA Trust (1) Year 2 Ab. Returns 0.166* (1.67) (2) Year 2 Ab. Returns 0.141 (1.38) -0.056 (-0.55) 0.001 (0.04) 0.080 (0.25) -0.303 (-1.22) 0.038 (0.37) -0.034 (-0.39) 0.063** (1.99) -0.001 (-0.02) -0.055 (-0.91) (3) Year 3 Ab. Returns 0.067 (0.64) (4) Year 3 Ab. Returns 0.046 (0.44) 0.057 (0.54) -0.055 (-1.08) -0.396** (-2.10) -0.299 (-1.37) -0.015 (-0.25) -0.173*** (-3.28) 0.041* (1.74) -0.011 (-0.30) 0.066 (1.24) 1,390 0.001 1,390 0.007 1,107 0.000 1,107 0.025 LnAge Founder LnEquityWealth Estate Taxes Top 10 Lawyer Debt Size Loss LnProceeds N Adjusted R2 41 Table 7 When does Trust use Predict Post-IPO Returns? This table regresses risk-adjusted stock returns during the first year following the IPO. In Columns 1 through 5 first day returns are included in the first year returns measure and in Column 6 they are not. Abnormal Returns are computed relative to a matched firm. To obtain a matched firm we take all firms that have been public for 5 years and have market capitalizations within 30% of the IPO firm and then select the firm with the most similar book-to-market ratio. The explanatory variable of interest is TA Trust, measured as the percentage of CEO shares in a tax-advantaged trust, and its interaction with firm characteristics. We interact TA Trust with Size, defined as the natural log of total pre-IPO assets, in Column 1, Loss, an indicator for a firm with negative pre-IPO ROA, in Column 2, Young Firm, an indicator for a firm with age below the sample median of 9 years according to the Field Ritter database, and R&D Firm, defined using an indicator for a firm with an above median pre-IPO R&D to assets ratio. Standard errors are clustered by three-digit SIC industries. T-statistics are presented below the coefficients with *, **, and *** representing statistical significance at the 10%, 5%, and 1% levels, respectively. TA Trust TA Trust × Size (1) Year 1 Ab. Returns 0.273 (1.22) -0.201 (-0.63) TA Trust × Loss (2) Year 1 Ab. Returns 0.023 (0.17) (3) Year 1 Ab. Returns -0.055 (-0.44) 0.350** (2.28) TA Trust × Young Firm 0.537*** (2.81) TA Trust × R&D Firm 0.418** (2.07) Young Firm 0.052 (0.89) R&D Firm Size Loss Control Variables N Adjusted R2 (4) Year 1 Ab. Returns -0.031 (-0.19) -0.047* (-1.91) 0.012 (0.14) YES 0.039 1,570 -0.045* (-1.91) -0.016 (-0.19) YES 0.040 1,570 -0.040* (-1.74) -0.019 (-0.25) YES 0.044 1,570 42 0.148* (1.91) -0.026 (-1.11) -0.052 (-0.48) YES 0.046 1,570 (5) Year 1 Ab. Returns -0.044 (-0.20) -0.455 (-1.33) 0.106 (0.58) 0.437* (1.96) 0.428** (2.20) 0.063 (1.03) 0.156** (2.15) -0.027 (-1.14) -0.093 (-0.89) YES 0.051 1,570 (6) Year 1 Ab. Ret. w/o UP 0.102 (0.58) -0.431 (-1.30) 0.101 (0.52) 0.353** (2.12) 0.224 (1.41) 0.022 (0.40) 0.040 (0.83) 0.087*** (4.48) -0.102** (-2.04) YES 0.036 1,570 Table 8 Corporate Tax Strategy and Individual Trust Strategy This table provides the regression outputs from equation (3). In columns (1) and (2) we use GAAP tax expense per dollar of pre-tax income in the three most recent years after going public, LR ETR, as the dependent variable. In columns (3) and (4) we use the sum of the firm’s cash taxes paid scaled by the sum of the issuers’ pre-tax income less special items in the three most recent years after going public, LR CETR, as the dependent variable. In columns (5) and (6) we the use the firm’s long-term (three year) cash taxes paid from Henry and Sansing (2014), LR CTP, as the dependent variable. In columns (7) and (8) we use the standard deviation of the firms’ cash effective tax rate for the three years following the IPO, σCETR, as the dependent variable. In columns (9) and (10) we the use the firm’s average unrecognized tax benefits scaled by assets for the three most recent years after going public, LR UTB, as our dependent variable. All models include year-quarter fixed effects. Standard errors are clustered at the three-digit industry level. Independent variable definitions are provided in Table 1. ***, **, * signify statistical significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) LR ETR LR ETR LT CETR LT CETR LR CTP LR CTP σCETR σCETR LR UTB LR UTB TA Trust LnAge Founder LnEquityWealth Estates Taxes Top 10 Lawyer PPE RD NOL Multi Debt -0.084* (-1.732) -0.094** (-2.027) -0.008 (-0.094) 0.004 (0.115) 0.014 (0.290) 0.094 (0.832) 0.023 (0.488) 0.007 (0.129) -0.130** (-2.434) -0.057** (-2.579) 0.001 (0.035) -0.077* (-1.705) -0.021 (-0.542) 0.001 (0.019) -0.013 (-0.223) -0.016 (-0.713) -0.086** (-2.443) -0.067 (-0.504) -0.030 (-1.206) -0.199*** (-4.927) -0.197*** (-3.225) -0.030** (-2.072) 0.029 (1.413) -0.013 (-0.413) -0.029*** (-3.020) 43 -0.023* (-1.66) -0.018 (-0.670) -0.013 (-1.446) 0.064*** (3.268) 0.182*** (3.256) 0.007 (0.762) 0.006 (0.342) -0.073** (-2.211) 0.030*** (3.692) -0.010 (-1.251) 0.045* (1.961) 0.076** (2.308) 0.075** (2.096) -0.040 (-0.990) -0.026* (-1.852) 0.016 (0.569) -0.166* (-1.921) -0.020 (-1.462) -0.026 (-0.791) 0.035 (0.592) -0.000 (-0.002) 0.022 (0.970) -0.012 (-0.659) 0.005 (1.658) 0.006** (2.163) 0.004 (1.525) -0.003*** (-2.966) 0.009** (2.080) -0.015*** (-3.359) -0.000 (-0.220) -0.012*** (-3.622) 0.007* (1.807) -0.000 (-0.248) 0.004** (2.374) -0.007*** (-4.921) (Table 8 continued) Size 0.325*** (27.684) 0.009 (0.754) -0.002 (-0.070) 0.326 (1.055) 563 0.096 563 0.126 LnProceeds Intercept N Adjusted R2 -0.013 (-1.295) -0.007 (-0.400) 0.279*** 0.513** (29.884) (2.145) 553 0.056 0.036*** (4.999) -0.016** (-2.563) -0.050*** (-3.908) 0.308*** (2.910) 840 0.001 840 0.181 553 0.109 44 0.134*** (21.414) -0.009 (-1.005) -0.004 (-0.294) 0.423** (2.484) 0.009*** (10.866) 0.000 (0.240) 0.000 (0.309) -0.004 (-0.284) 471 0.001 471 0.023 406 -0.005 406 0.099 Table 9 Subsidiary Location and Individual Trust Strategy This table provides the regression outputs from equation (3). We investigate the association between firm subsidiary locations and individual trust strategies. In columns (1) and (2) using OLS, the dependent variable is Haven Subs, the natural log 1 + the total number of material subsidiaries operating in tax haven countries in Exhibit 21 from their first 10-K following the IPO. In columns (3) and (4) using a Probit, the dependent variable is an indicator variable equal to 1 (0) if the firm reports (does not report) a material subsidiary in Exhibit 21 from their first 10-K following the IPO operating in a DOT country. Columns (1) and (2) include year-quarter fixed effects. Standard errors are clustered at the three-digit industry level. Independent variable definitions are provided in Table 1. ***, **, * signify statistical significance at the 1%, 5%, and 10% level, respectively. TA Trust (1) Haven Subs (2) Haven Subs 0.298*** (2.865) 0.215** (2.070) -0.061 (-0.382) -0.054 (-1.430) 0.233** (2.485) 0.364 (0.851) 0.056 (0.626) 0.324** (2.056) 0.240** (2.384) -0.012 (-0.463) -0.255*** (-3.121) 0.007 (0.237) -0.021 (-0.327) 0.045 (0.915) LnAge Founder LnEquityWealth Estate Taxes Top 10 Lawyer PPE RD NOL Debt Size Loss LnProceeds Total Count Total Subs Intercept N Adj. (Pseudo) R2 (3) DOT 2.354*** (3.521) 1.970*** (6.996) 0.443*** (15.967) -0.102** (-2.374) 0.433*** (15.756) -0.223 (-0.372) 470 0.491 470 0.505 45 -6.887*** (-9.532) 470 (0.335) (4) DOT 1.873** (2.258) 1.460 (0.820) 1.672** (1.994) 0.997 (1.475) -0.920 (-0.426) -0.346 (-0.401) 4.438*** (2.852) 1.355 (1.290) -3.485** (-2.271) 0.372 (0.417) 0.212 (0.877) 0.546 (1.364) -0.031 (-0.069) 2.464*** (3.369) -16.003** (-2.305) 470 (0.455)
© Copyright 2026 Paperzz