CORPORATE INVERSIONS AND THE COST OF EQU ITY: A TALE OF TWO STRATE GIES By Tianpeng Zhou A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Business Administration Financ e Doctor of Philosophy 2018 ABSTRACT CORPORATE INVERSIONS AND THE COST OF EQU ITY: A TALE OF TWO STRATE GIES By Tianpeng Zhou Firms invert either through a pure inversion strategy or by merging with a foreign entity. I document that the impact of corpo rate inversions on the cost of equity is significantly different between the two strategies. I find that pure inversions increase the cost of equity by 10%, whereas inversions through mergers decrease it by 13%. Although both inversion strategies increase the However, before the tax reform of 2004, which eliminated the tax savings from pure inversions, most inversions were pure, whereas after the tax reform most were done through mergers. This finding suggests that the tax reform had an unintended consequence of reducing a managerial agency problem by eliminating the less beneficial inversion option. iii This dissertation is dedicated to my wife, Xiaohui Lin. iv ACKNOWLE DGEMENTS I would like to express the deepest appreciation to my committee chair Professor Naveen Khanna , who convincingly conveyed a spirit of adventure in research and scholarship and an excitement in teaching. And equally important is his earnest teachi ng with regard to the philosophy of life , which will be one of my most valuable assets as well. I would like to thank my committee members, Professor Jeffrey M. Wooldridge, Professor Hao Jiang, and Professor Xing Huang , as well . Without their persistent g uidance and help, this dissertation and my graduation would not have been possible. Also, words cannot express how grateful I am towards my parents , Xianfeng Zhou and Aifang Zhang, who have spare d no efforts to support my education for twenty - three years, from elementary school to graduate school. Especially, I truly appreciate their belief that it is a good thing that their son has never earned a penny from any so far . Last but definitely not the least, I sincerely thank the support from my wife , Xiaohui Lin , who has hope that tomorrow is going to be better than today . Magically, this project was started at about the same time that we knew each other, and I d efended my dissertation at the time that we got married. There fore, I would like to dedicate this dissertation to her. v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ......................... vi LIST OF FIGURES ................................ ................................ ................................ ...................... vii 1. INTRODUCTION ................................ ................................ ................................ ...................... 1 2. LITERATURE REVIEW ................................ ................................ ................................ ........... 6 3. INCENTIVES OF CORPORATE INVERSIONS ................................ ................................ ..... 9 3.1. Incentives of Corporate Inversions ................................ ................................ ...................... 9 3.2. Sect ion 7874 of the 2004 American Jobs Creation Act ................................ ..................... 13 4. EMPIRICAL METHODOLOGY ................................ ................................ ............................. 16 4.1. Implied Cost of Capital ................................ ................................ ................................ ...... 16 4.2. Matching ................................ ................................ ................................ ............................ 18 4.3. Li near Difference - in - Difference Analysis ................................ ................................ ......... 20 4.4. Nonlinear Analysis: Fractional Response Models ................................ ............................. 21 5. DATA AND SUMMARY STATISTICS ................................ ................................ ................. 23 5.1. Data ................................ ................................ ................................ ................................ .... 23 5.2. Summary Statistics ................................ ................................ ................................ ............. 24 6. RESULTS ................................ ................................ ................................ ................................ . 30 6.1. Linear Regression ................................ ................................ ................................ .............. 30 6.2. Fractional Response Model ................................ ................................ ................................ 34 6.3. Robustness Checking ................................ ................................ ................................ ......... 38 6.3.1. One - for - Four Matching ................................ ................................ ............................... 38 6.3.2. Augmented Linear Regression ................................ ................................ .................... 42 6.4. Cash Holdings, Dividend Payments, and Capital Expenditures ................................ ........ 44 7. CORPORATE INVERSION STRATEGIES AND EXISTING SHAREHOLDER VALUE . 49 7.1. The Advantage of M&A Inversions over Pure Inversions ................................ ................ 49 7.2. A Moral Hazard Implication ................................ ................................ .............................. 53 8. CONCLUSION ................................ ................................ ................................ ......................... 56 APPENDICES ................................ ................................ ................................ .............................. 58 APPENDIX A. List of Corporate Inversions Announced between 1993 and 2015. ................ 59 APPENDIX B. Corporate Inversions Used for Empirical A nalysis. ................................ ........ 65 APPENDIX C. Variable Definitions ................................ ................................ ........................ 70 BIBLIOGRAPHY ................................ ................................ ................................ ......................... 71 vi LIST OF TABLES Table 1: Inversion Announcements Across Industries . ................................ ................................ 26 Table 2: Inversion Destinations. ................................ ................................ ................................ ... 27 Table 3: Summary Statistics for the Main Variables Used in This Analysis. ............................... 28 Table 4: Comparison of Treated and Control Gr oups: One - to - One Matching. ............................ 30 Table 5: Effect of Inversion on the Cost of Equity: Benchmark Linear Regression. ................... 33 Table 6: Effect of Inversion on the Cost of Equity: Benchmark Fractional Respons e Models. ... 36 Table 7: Comparison of Treated and Control Groups: One - to - Four Matching. ........................... 38 Table 8: Robustness Check for Control Sample Size: Linear Regression. ................................ ... 39 T able 9: Robustness Check for Control Sample Size: Fractional Response Models. .................. 40 Table 10: Augmented Linear Regression. ................................ ................................ .................... 44 Table 11: Effect of Inversion on Cash Holdings. ................................ ................................ ......... 45 Table 12: Effect of Inversion on Dividend Payments. ................................ ................................ .. 47 Table 13: Effect of Inversion on Capital Expenditures. ................................ ............................... 48 Table 14: Existing Shareholder Value Change. ................................ ................................ ............ 51 T able A1: List of Corporate Inversions Announced between 1993 and 2015 .............................. 59 Table A2: Corporate Inversions Used for Empirical Analysis ................................ ..................... 65 vii LIST OF FIGURES Figure 1. Headline Marginal Corporate Tax Rates (Selected Countries). ................................ .... 10 Figure 2. After - Tax Earnings Before and After Inversion. ................................ ........................... 12 Figure 3. Inversion Volume. ................................ ................................ ................................ ......... 24 1 1. INTRODUCTION The exodus of U.S. corporations for tax - haven countries, 1 to a discussion among policy makers regarding possible tax and other regulatory reforms. In a corporate inversion, a U.S. multinatio nal corporation (UMC) either (i) is acquired by its own foreign subsidiary (pure inversion strategy) or (ii) merges with or is acquired by a foreign operating company (M&A inversion strategy). The newly incorporated firm is governed and taxed by the foreig The current literature on corporate inversions has focused on the benefits of inversions. Hines and Hubbard (1990), Atlshuler, Newlon, and Randolph (1995), and D esai, Foley, and Hines (2001) identify that the inverting firm avoids paying repatriation taxes by inverting to a country with a territorial tax regime. 2 An inverting firm will also allocate its expenses more effectively across borders and reap indirect ta x savings after inversion, 3 Additionally, Talley (2015) argues governance encourages UMCs to reincorporate overseas. Although the benefits of inversions have b een extensively investigated in the literature, not much attention has been paid to the costs associated with inversions and the differences between inversion strategies. Specifically, issues requiring further inquiry include the impact of inversion 1 Tax - haven countries are those with special tax attributes designed to attract foreign investors. They typicall y have very low tax rates. In return, they will receive large foreign investments and enjoy fast economic growth. Currently, there are about 40 tax havens, including Bermuda, the Cayman Islands, and Ireland. (see Dharmapala and Hines, 2009). 2 The tax - savi inversion decision. For instance, in its announcement of inverting to Canada, Burger King predicted that it could save $117 million annually. Similarly, Walgr eens estimated that it could dodge up to $4 billion in U.S. taxes over five years by inverting to Switzerland. 3 For example, the surviving tax - haven parent company can lend to the U.S. subsidiary, thereby generating interest deductions against U.S. taxabl e income while the interest income of the parent company is tax - free (Desai and Hines, the funds borrowed from subsidiaries are taxed as repatriated dividends. 2 on a f increasing existing shareholder value; whether there is evidence that the choice of strategies may be driven by the managerial agency problem; and whether the 2004 ta x reform aimed at curbing inversions has been effective. This paper attempts to answer some of these questions. First, I believe this paper is the first to document changes in the cost of equity caused by a corporate inversion. I also find differential imp acts of inversions on the cost of equity between the two inversion strategies: pure inversions increase the cost of equity by around 10% of pre - inversion levels, whereas M&A inversions decrease it by around 13%. Second, I document that an M&A inversion is more beneficial to existing shareholders than a pure inversion and on average delivers an additional amount of around $4.6 billion to existing shareholders over a 5 - year window. Third, before the 2004 tax reform, most inversions were pure, whereas after th e tax reform, they were done through M&A. Although this may have been an unintended consequence of the tax reform, my results suggest that the tax reform appears to have reduced a managerial agency problem by making the less efficient strategy unattractive and, thus, inducing managers to choose the more efficient inversion strategy. The differential impacts of inversions on the cost of equity between the two strategies can be explained by the salient differences between them. The preferred pure inversion de stinations are usually offshore tax - haven destinations such as Bermuda and the Cayman Islands, where corporate tax rates are usually zero. Moreover, there is usually no change in the location of the ions of inverting firms under pure inversion. 4 There is a change, however, in the jurisdiction of the firm because it is now governed 5 Political and economic 4 In fact, the company is not required to conduct any meetings or even have an office in the newly incorporated country. 5 For example, the U.S. corporate laws are more protective against hostile take overs than the corporate laws i n most 3 risks are larger in these dest ination jurisdictions as well. are likely to change if there are efficiency gains to be achieved in an M&A inversion. Also, the destination jurisdictions of M&A invers ions are in better developed countries such as Ireland and the United Kingdom, and their corporate law systems and political risks are similar to those in the United States. Therefore, ceteris paribus, the cost of equity increase caused by the change in co rporate law jurisdiction in an M&A inversion should be smaller than the increase following a pure inversion. Second, in an M&A inversion, a UMC merges with a foreign firm that is usually of similar size, economically profitable, and internationally well kn own. All else equal, merging in an M&A inversion are general ly in the same industry, an M&A inversion decreases the global positive effect of M&A inversions on the cost of equity indicates that the benefits created through corporate laws. Though an M&A inversion strategy is more favorable than pure inversions in terms of reducing the cost of equity, the tax savings through an M&A inver sion strategy are usually less than those through pure inversions because the corporate tax rates in the destination jurisdictions of M&A inversions are usually higher than those in the offshore tax - haven destination jurisdictions of pure inversions. This begs the question of which strategy is more beneficial to existing shareholders. I look at the impact of the 2004 tax reform, which restricts the tax benefits from pure of the tax havens. 4 inversions and induces inverting firms to invert through mergers. By estimating the cha nges in existing shareholder value after a firm inverts, either through a pure inversion strategy or by merging with a foreign entity, I am able to document that the M&A inversion strategy is more beneficial to existing shareholders in that it delivers an additional amount of around $4.6 billion to existing shareholders. This finding suggests that prior to the 2004 tax reform, managers do not appear to maximize shareholder value. Moreover, it appears that the tax policy change had an unintended consequence of reducing the managerial agency problem. To test the hypothesis that UMCs invert in order to have access to the cash held in foreign subsidiaries, I look at changes in the cash holdings of inverting firms and find a statistically significant decrease in the cash holdings of pure - inverting firms. I further show that the repatriated cash holdings are mainly used to invest in potentially profitable projects located in the United States instead of used to distribute more dividends: the capital expenditure of the pure - inverting firm significantly increases, whereas the dividend payment is unchanged after a pure inversion. However, the cash holdings and investment policies of the foreign acquirer affect the post - inversion cash holdings, dividend payments, and ca pital expenditure in an M&A inversion, which makes the changes insignificant. In this paper, I investigate the impact of inversions on the cost of equity and shareholder value by causal inference. The difference - in - difference (diff - in - diff) method is used to eliminate factors that can simultaneously affect inverting and non - inverting firms, thus leaving inversion to be the only policy change between the inverting firm and its matched non - inverting control pair. groups - two periods" linear diff - in - diff model, I use - multiple periods" individual - level linear and nonlinear diff - in - diff models to capture more unobserved firm and time heterogeneities and hence to increase explanatory power. 5 Since the incenti ve to invert comes from the less costly benefit of having profitable foreign operations and repatriating cash held in foreign subsidiaries, I choose the control sample from the pool of U.S. multinationals. To find the appropriate match for the inverting fi rm, I first define the inversion. Then within the same industry 6 and measurement quarter, I identify up to four control firms for each inverting one by matching on three observable firm characteristics using the Mahalanobis distance 7 : firm size (log of total assets), leverage (leverage ratio), and profitability (return on assets). My matching results show that there is no statistical difference in the distribution of these matched variables between the treated and control sample. Since I find matches for each inverting firm, I essentially estimate the average treatment effect on the treated (ATT) of corporate inversions. Robustness checking results show that the analy sis in this paper is robust to the size of the control sample. The rest of the paper is organized as follows. Section 2 discusses the related literature. Section 3 briefly discusses the benefits of corporate inversions and the legislative actions taken tha t aim to curb (pure) inversions. Section 4 gives my empirical methodology. Section 5 describes the data and provides the summary statistics. Section 6 discusses the empirical testing results of the impact of inversions on the cost of equity. Sections 7 com pares changes in existing shareholder value after a firm inverts either through a pure inversion strategy or by merging with a foreign entity. Section 8 concludes. 6 Industry is an important factor to control because there is significant heterogeneity in the concentration of inversions (Babkin, Glover, and Levine, 2016). 7 The Mahalanobis distance of two vectors x i and x j is defined as d(i; j) = ( x i - x j - 1 ( x i - x j ), where is the variance - covariance matrix of x i and x j . For a fixed vector x i , the vector x j * that gives the shortest d(i; j) is the best match for vector x i . 6 2. LITERATURE REVIEW An important strand of literature on corporate inversions is investiga ting the causes of inversions. Hines and Hubbard (1990), Atlshuler et al. (1995), and Desai et al. (2001) identify two components of the tax benefits of inversion. First, by inverting to a country with a territorial tax regime, the firm can avoid paying re patriation taxes as well as circumvent taking costly actions to avoid such taxes had the firm not inverted. Second, an inverting company can save taxes by reallocating expenses, including the allocation of interest expense to foreign source income. Desai a nd Hines (2002) find that large firms, those with extensive foreign assets, and those with considerable debt are most likely to expatriate, which suggests that U.S. taxation of foreign income, including the interest expense allocation rules, significantly affects inversions. Seida and Wempe (2003a) further confirm that the reduction in taxes post - inversion is partly due to the reduction in U.S. taxable income by shifting more expenses to the United States after inversion. Talley (2015) argues that besides t law and corporate governance has encouraged UMCs to reincorporate overseas. Over the last 15 traditionally the domain of state law. These mandates have displaced state law as a primary source of governance regulations for U.S. - listed issuers, and this displacement has gradually unbundled domestic tax law from corporate governance, which erodes the U.S. market power in regulatory competition. The second strand of literature on inversion, albeit small, analyzes the costs of inversion. Cortes, Gomes, and Gopalan (2015) find that inverting firms have higher bid - ask spreads and their investors put a low er value on the cash on their balance sheets. They also find that inverting firms have a more concentrated institutional share ownership structure, and they document the lower 7 stock liquidity resulting from inversions. Different from their paper, my paper analyzes the cost change in a more direct way by analyzing the changes in the cost of equity. Babkin et al. (2016) develop a model to value the net benefits of inversion and show that while the benefits of inversion disproportionately accrue to the CEO, fo reign shareholders, and short - term investors, many long - term investors suffer a net loss. The negative return of long - term investors results from the capital gains tax event triggered by inversion. They have to pay capital gains taxes immediately after inv ersion, thus losing the option to defer capital gains taxes on their shares. That loss outweighs the benefits generated from the reduced future corporate tax rate. This paper is also related to the literature that analyzes the share price reaction around a n inversion. Based on a sample of 19 pure - inverting firms, Desai and Hines (2002) find that share prices rise by an average of 1.7% in response to expatriation announcements. Among these 19 companies, eight experienced positive abnormal returns over the on e - day window, and ten did so over a five - day window. But their paper does not indicate whether the increase is statistically significant. Cloyd, Mills, and Weaver (2003) and Seida and Wempe (2003b) find no evidence of a positive market reaction to board ap proval announcements. Since these papers were published before the passing of Section 7874 of the 2004 American Jobs Creation Act, they all analyze pure - inverting firms with a small sample. The sample used in my paper also includes post - 2004 inverting comp anies and M&A inverting companies. Thus, the sample size in this paper is much larger than their sample size (63 versus 19). Finally, this paper is also related to the literature on cross - listing firms (Doidge, Karolyi, and Stulz, 2004; Doidge, Karolyi, an d Stulz, 2007), which suggests that U.S. listing facilitates the firm to take growth opportunities by limiting the extent to which controlling shareholders can engage in expropriation. But there is a difference between the inverting companies and the cross - 8 listing companies. Compared to the cross - listed companies discussed in this strand of literature whose significant operations are mainly outside of the United States, the inverting firms still operate mainly in the United States. This further enhances the on the inverting companies (Siegel, 2005; Shnitser, 2010; Licht, 2003; Cortes et al., 2015). 9 3. INCENTIVES OF CORPORATE INVERSIONS 3.1. Incentives of Corporate Inversions regarded as an important reason that caused companies to invert. The dominant federal tax bracket for most moderate - sized (or larger) U.S. corporations is 35%. 8 Adding state corporate taxes, which range from 0% to 9%, to this figure, most UMCs are faced w ith a marginal headline tax rate around 40%. 9 As Figure 1 illustrates, the combined headline rate in the United States (the top thick line) far exceeds the rate in most of the competitive jurisdictions, and the gap has been widening in recent effective tax rate in the United States may be well below the headline tax rates because of various kinds of accounting and financing methods, U.S. corporations are still faced with a large tax bill. In addition, the worldwide approach of the U.S. tax regime combined w ith the high U.S. tax rates give UMCs the incentive to invert. 10 Different from the territorial tax regime in most OECD countries, which exempts the distributions from controlled foreign subsidiaries for tax purposes, the U.S. tax authorities levy taxes on generated in the United States. To avoid double taxation, the U.S. tax code grants a credit for foreign taxes already paid to the foreign governments, and the foreign operations are liable only to the extent that tax liability under U.S. law would exceed that amount. Also, a UMC can defer parent. I use a hypothetical example to illustrate this point (see Pan el A of Figure 2). 8 To be more specific, these are the companies whose taxable net ea rnings exceed approximately $18.3 million per year (Talley, 2015). 9 See Corporate Tax Rates Table, KPMG, http://www.kpmg.com/global/en/services/tax/ tax - tools - and - resources/pages/corporate - tax - rat es - table.aspx. 10 Besides the United States, other OECD countries that take a worldwide approach are Chile, Greece, Ireland, Israel, South Korea, and Mexico. 10 Figure 1. Headline Marginal Corporate Tax Rates (Selected Countries). This figure depicts the headline (country plus state) marginal tax rates of several selected OECD countries. The headline marginal tax rate in the United States (the thick line at the top) far exceeds the rate in most of the competitive jurisdictions, and the gap has been widening in recent years. In 2016, the U.S. corporate headline tax rate is 39%, much higher than that of Ireland (12.5%), the United Kingdom (20% an d falling), Canada (26.5%), and the OECD average (25%). Suppose a U.S. incorporated multinational parent company has two separately incorporated subsidiaries, 11 a U.S. subsidiary and a foreign subsidiary, and each generates $100 in taxable earnings. Sup territorial corporate tax rate is 15%. Immediately when the earnings are generated, the U.S. subsidiary pays $40 to the U.S. tax authority and the foreign subsidiary pays $15 to th e foreign government. No further taxes are levied at the corporate level as long as the remaining $85 is kept in the foreign subsidiary. However, if the foreign subsidiary repatriates the profits to its U.S. parent, in theory it should pay $40 to the Unite d States, but the U.S. law permits the company to 11 The profits of unincorporated foreign business branches are taxed immediately by the United State s. Tax deferral is not allowed for foreign business branches. 11 claim $15 of foreign tax credits, which equals the amount of taxes already paid to the foreign government. So it needs to pay a $25 repatriation tax to the U.S. government before repatriating the remaining $60 to the U.S. parent. The overall after - tax earnings of the parent company is $120. There is not much a U.S. parent company can do to avoid the U.S. tax liabilities on the foreign earnings. Though it could defer the $25 U.S. tax liability indefinitely as long as the after - - incorporated subsidiaries, the parent company might want to utilize these earnings for profitable investment undertakings in the United States or for distributing more cash divi dends. A common strategy for using unrepatriated earnings is through intra - company loans from the foreign company to the U.S. parent company. But such practices are usually restricted under Section 956 of the U.S. Internal Revenue Code by deeming such loan s as dividends and thus taxable. 12 After inverting to a tax - haven country, the multinational firm can avoid paying repatriation taxes and unlock its overseas earnings. Suppose the U.S. parent company shown in Panel A of Figure 2 is acquired by its foreign subsidiary and then the surviving company is reincorporated in territorial, all $60 in after - tax earnings in the U.S. subsidiary can be repatriated to the fo reign parent company without paying any repatriation taxes. The overall after - tax earnings of the parent company are now $145 compared to the $120 pre - inversion, thereby generating $25 in tax savings. In addition, the surviving firm after an inversion can allocate its expenses more effectively across borders and further save taxes after inversion. For example, the surviving tax - haven - incorporated parent company can lend to the U.S. subsidiary, thereby generating interest 12 There are some limited exceptions under Section 956 of Internal Revenue Code, which gives criteria under which certain loans are nontaxable. But such loans are required to be short - term in nat ure and generally need to be repaid 12 deductions against U.S. taxable in come while the interest income of the parent company is tax - free. In this way, overseas funds are made available to the U.S. subsidiary, and the firm also pays less taxes. Figure 2. After - Tax Earnings Before and After Inversion. In this figure I use a hyp othetical example to show where the tax savings originate after a corporate inversion. The corporate tax rate in the United States and the foreign country is assumed to be 40% and 15%, respectively. Panel A represents the case in which the multinational co rporation is U.S. - incorporated. Because the U.S. tax regime takes a worldwide approach, the firm needs to pay a $25 repatriation taxes to the U.S. government if it wants to repatriate the $85 in foreign after - tax earnings. By flipping the corporate structu re and reincorporating in a foreign country where taxes are levied only on the earnings generated domestically (territorial regime), as indicated in Panel B, the multinational firm could save $25 in repatriation taxes and hence increase its overall after - t ax earnings by $25. 13 3.2. Section 7874 of the 2004 American Jobs Creation Act The first wave of inversion came after 1994 when Helen of Troy, a publicly traded U.S. company in the personal care industry, pure - inverted to Bermuda. From 1994 to 2003, 28 co mpanies were reincorporated abroad, among which 24 companies used pure inversions. Typically, the U.S. parent company is acquired by one of its foreign subsidiaries registered in a tax - haven country such as Bermuda in a pure inversion. After the transactio n, the foreign subsidiary becomes the new parent company and the U.S. parent company becomes a subsidiary: therefore, the corporate structure is inverted. Existing shareholders of the former U.S. parent company typically still hold 100% of the shares in th e new foreign - incorporated company. For instance, after Helen of Troy announced its inversion decision in December of 1993, it registered a subsidiary firm in Bermuda at first and then flipped it to become the new parent. Shareholders of the former U.S. pa rent firm received one share of the new firm for each share that they owned. In response to this trend, Section 7874 of the 2004 American Jobs Creation Act was p assed threshold" (following Talley (2015)) is used to determine whether an inverting firm is qualified to reap tax benefits after an inversion. 13 First, the 60% thresh old kicks in when the existing shareholders of the inverting firm hold less than 60% of the shares in the surviving (combined) company. If so, the surviving company will be fully treated as a foreign corporation by the U.S. tax authorities. Second, if the existing shareholders of the inverting firm hold greater than 60% but 13 acquires substantially all of the U.S. corporation three criteria, it can reap the benefits of tax savings. But criteri a (i) and (ii) are usually satisfied for almost all pure 14 less than 80% of the shares in the surviving company, the surviving company is partly recognized e gains that it received in this transaction in the following 10 years after inversion. Third, if the existing shareholders of the inverting firm hold more than 80% of the shares in the surviving company, the surviving company is still regarded as a U.S. i ncorporated company for tax purposes and thus reaps no tax benefits from this inversion. Section 7874 effectively put a leash on pure inversions because the existing shareholders of a pure - inverting firm hold almost 100% of the shares in the surviving firm , and therefore the firm incurs no tax - saving benefits from the inversion. But the trend of corporate inversions was not stopped by Section 7874, and M&A inversions took the place of pure inversions, becoming the most favorable inversion strategy after 200 4. Different from pure inversions, a UMC merges with or is acquired by a foreign company of similar size in an M&A inversion and changes its incorporation country to where the foreign acquirer is located. For example, in August of 2014 the then Delaware - in corporated fast food chain Burger King Worldwide, Inc., announced that it agreed 14 The new company is now governed by Canadian corporate laws while it still lists on t he New York Stock Exchange. An inverting company can easily bypass the 80% threshold by acquiring a foreign company even be below 60% after M&A inversions. Fo llowing thi s change, more mature and better 14 f or each share they owned. The two brands still operate independently after the merger under the new combined company, Restaurant Brands International, which is incorporated in Canada. The shares of the combined company started trading on the Toronto Stock Exchange and New York Stock Exchange under the symbol QSR on December 15, 2014. 15 developed countries such as Canada, the United Kingdom and Ireland replaced Bermuda and the Cayman Islands as the most popular countries to invert to. These new destinations have stronger economies, more stable e conomic policies, more favorable corporate law structures, and similar political risks than the pure inversions. Those advantages will possibly be reflected in the inverting 16 4. EMPIRICAL METHO DOLOGY 4.1. Implied Cost of Capital In this paper, I use the implied cost of capital (ICC), computed using earnings forecasts and market price, to measure the cost of equity. Fama and French (1997) conclude that the cost of equity estimates derived from th e capital asset pricing model and three - factor models using in identifying the right asset pricing model, imprecision in the estimates of factor loadings, and the imprecision in the estimates of factor risk premia. Pastor, Sinha, and Swaminathan (2008) further show in simulations as well as empirically that ICC outperforms the realized rate of return in detecting a risk - return trade - off. The empirical cons truction of the ICC in this paper closely follows Pastor et al. (2008). The firm - level quarterly ICC is the value of r e that solves the empirically tractable finite - horizon model: (1) where P t is the (average) stock price at quarter t , FE t+k and b t+k are the forecasts of the earnings per share (EPS) and the plowback ratio in k y ears after quarter t , and T is the forecasting horizon ( T = 15). Notice that represents the free cash flow to equity in k years after quarter t . The first term in equation (1) captures the total present value of free cash fl ow to e quity up to the terminal period T , and the second term captures the total present value of all cash fl ows beyond the terminal period. The EPS forecasts FE t+k ( k = 1 ; 2 T+1 ) are obtained based on the following steps. (i) The 1 - and 2 - year - ahead EPS fore casts are the median I/B/E/S 1 - and 2 - year - 17 forecasts. (ii) The 3 - year - ahead EPS forecast is computed as FE t+3 = FE t+2 (1 +LTG) , where LTG is the long - term earnings growth rate obtained from I/B/E/S. If LTG is missing in the I/B/E/S databa se, I extrapolate the growth rate in the first two years ( FE t+2 / FE t+1 - 1 ) for another year. Following Pastor et al. (2008), firms with growth rates above 100% (below 2%) are assigned with 100% (2%). (iii) By imposing an exponential rate of decline to mean - revert the 3 - year - ahead growth rate to the steady - state growth rate, I compute the 4 - to ( T +1) - year - ahead EPS as follows: (2) (3) where is the k - year ahead earnings growth rate, and g is the steady - state growth rate starting in T+ 2 periods after t . I use the 10 - year rolling average of the annual nominal GDP growth rate as a proxy for g . (iv) I assume that the dividend paid to the shareholders in year t+k is zero if the forecasted earnings in that year are negative. The last thing needed to compute the ICC is the forecasted plowback rate b t+k , which is computed in two stages: (i) I explicitly forecast the plowback ratio in the nearest two years, b t+1 and b t+2 income) 15 ; (ii) I mean - revert the plowback ratios between t+2 and t+T +1 linearly to the steady - state plowback ratio based on the following equation: (4) where b is the steady - state plowback ratio. The variable b is computed from the sustainable growth 15 Following Gerbardt, Lee, and Swaminathan (2001), for firms with negative earnings, I divide dividends by long - term earnings. The long - term earnings are estimated to be 6% of total assets since the long - run return on assets in the United States is 6%. 18 rate formula, g = ROI b , where ROI is the steady - state return on investment. Following Gerbardt et al. (2001) I use the 10 - - state ROI . Finally, wit h the forecasted EPS and plowback ratio readily available, ICC can be computed from equation (1). 4.2. Matching I test the change in the cost of equity by comparing the inverting firms to a matched sample of UMCs and computing the diff - in - diff estimate of the effect of inversion on the cost of equity. I select the control sample from the pool of UMCs instead of the entire group of U.S. incorporated companies for two reasons. First, as illustrated in Section 3, only multinational firms can benefit from inve rsions, and the incentive to invert comes from the less costly benefit of having profitable foreign operations and repatriating the cash held within foreign subsidiaries. So companies with 100% business in the United States will not construct a good contro l sample, and therefore these firms should be excluded from our consideration. Second, because firms in my treated sample are all incorporated in the United States and are publicly traded in U.S. markets before they invert, to make a good comparison, I sel ect the control sample only from the firms that have remained incorporated and publicly traded in the United States since their inception. So, diff erent from Cortes et al. (2015), I do not consider the foreign private issuer (FPI) and American foreign corp orations (AFC) in this paper. 16 To find the appropriate control sample by matching, I first defi ne the measurement quar ter 16 FPIs are foreign incorporated and have business mainly outside of the United States. Though they are cross - listed in the United States and are governed under the federal securit ies law, they are usually faced with less stringent disclosure and corporate governance rules (Siegel, 2005; Shnitser, 2010; Licht, 2003). AFCs include household names such as Michael Kors Holdings Ltd and Carnival Corporation, which have substantial busin ess in the United States but were incorporate outside of the United States from the time they were built. Since they started out as a foreign company, there is no need to invert out of the United States for them. 19 to be the last quarter of the calendar year prior to the announcement of inversion. Then, for each inverting firm, I match on three ob servable firm characteristics while controlling the measurement quarter and industry: firm size ( log(total assets) ), leverage ( leverage ratio ), and profitability ( return on assets , or ROA ). These matching variables are chosen because the literature (e.g., Desai and Hines (2002) and Cortes et al. (2015)) has shown that large, profitable multinational firms with high leverage have a higher probability of inverting. Industry is another important factor to control because some industries, such as the pharmaceut ical products industry, tend to have more inversions compared to other industries. Specifically, for each inverting firm in the treated sample, I identify the best match in the same industry and the same measurement quarter in terms of those three observab le firm characteristics using the Mahalanobis distance. It might be instructive to look at an example to understand the matching used here. If firm A announces that it will invert to Bermuda in March of 2010, I define the measurement quarter to be the four th quarter of 2009. I then compute the Mahalanobis distance for all the UMCs in the measurement quarter based on log(total assets) , leverage ratio , and return on assets , and double - sort the firms by distance and industry. Lastly, I use the firm with the sm allest distance in the same industry as firm A to be the control firm. To ensure that the control firm is the closest match, I match with replacement and drop the duplicates. In addition, the first four closest matches are used for a robustness check. This matching method guarantees the three matching variables to be exogenous to the treatment (inversion) because they are measured in the quarter before the inversion an nouncement. In the commonly used matching method, a match is found for each firm - quarter o bservation before and after the treatment, so that these matching firm characteristics will be changed by an inversion, especially an M&A inversion, therefore weakening the analysis. In this sense, my matching 20 method is more advantageous than the commonly used matching method and di ff ers from that of Cortes et al. (2015). 4.3. Linear Di ff erence - in - Di ff erence Analysis Having identified the treated and control groups, I test the ef f ects of inversion on the cost of equity by estimating the following linear mo del: (5) where is the risk premium (the cost of equity minus the risk - free rate) of company i at time t , and is the incorporation status dummy variable of company i at time t . If company i is incorporated in the United States at time t , equals 0. Otherwise, if it is incorporated in a foreign country at time t, equals 1. Notably, equals 0 for all control firms. For treated fi rms, equals 0 before they invert, and 1 afterward. Since no companies in my sample reincorporate back to the United States after they invert, = 1 whenever = 1 for all s < t . Log(total assets) , leverage ratio , and ROA are used as controls and included in . The term denotes a full set of unobserved heterogeneity ef f ects, which will absorb the impact of any time - invariant firm characteristics, and denote a full set of time - fixed ef f ects. The error term includes all other time - varying unobservable shocks to the cost of equity. We assume that the incorporation indicator is strictly exogenous: after controlling for log(total assets) , leverage ratio , ROA , unobserved firm - specific heterogenei ty, and time - fixed ef f ects, is conditionally independent of incorporation status, . This assumption is plausible because companies invert mainly because of tax savings and expense reallocation, and there is no evidence indicating that sh ocks to the cost of capital af f decision. 21 Following Imbens and Wooldridge (2007), equation (5) is essentially an individual - level dif f - in - dif f model, and is the diff - in - diff estimator. Compared to traditional linear diff - in - diff re gressions with only two periods around the treatment, the individual - level diff - in - diff regression controls more unobservable time - fixed effect s by using the entire time series path issue to notice is that no systematic factors are included in equation (5) . There is little doubt that systematic factors such as the market risk premium, size, and book - to - market effect s significantly affect the cost of equity (Pastor et al., 2008). They are omitted from equation (5) because they tend to affect both treated and control observations, and their net contribution to the diff - in - diff estimate is zero. 4.4. Nonlinear Analysis: Fractional Response Models To make the estimation of the effect s of inversion more precise, I use a fractional response model to capture the p roperty that the cost of equity is generally between 0 and 1: (6) where is the cost of equity. Assuming , we have (7) To ensure lies between 0 and 1, I choose three functional forms of : (i) ; (fractional logit) (ii) , where is the cumulative function of the standard normal distribu tion, or ; (fractional probit) (iii) ; (fractional complementary log - log). All of these models are estimated using quasi - max imum like lihood estimation, and the stan dard 22 errors are fully robust to firm - level heteroskedasticity and serial correlation. The av erage treatment effect on the treated (ATT) estimated in the fractional response models is t he diff - in - diff estimate of the 23 5. DATA AND SUMMARY STATISTICS 5.1. Data I obtain price data from CRSP, accounting data from COMPUSTAT (U.S., quarterly), analyst forecast data from I/B/E/S, and the one - month T - bill rate of return from Kenneth website. Additionally, nominal GDP growth rates are available from the Bureau of Economic Analysis. To compute the ICC, we need to merge COMPUSTAT with I/B/E/S first. However, we cannot merge them directly because COMPUSTAT uses GVKEY as the perma nent identifier to track each company, while I/B/E/S uses I/B/E/S TICKER. So I start with merging I/B/E/S and CRSP by CUSIP, CUSIP dates, and company names to build a one - to - one match between I/B/E/S TICKER and CRSP PERMNO, 17 and then merge the resulting fi le with CRSP/COMPUSTAT - Merged dataset by PERMNO and date. In this way, we obtain the COMPUSTAT/IBES - Merged dataset. The list of corporate inversions is obtained mainly from three sources: the existing literature (Desai and Hines, 2002; Seida and Wempe, 200 3b; Talley, 2015), the Bloomberg Corporate Expatriates list, and the SEC. Appendix A lists 88 companies that announced plans to approval. To be consistent with the existing lit screen these companies by comparing each firm with its SEC Edgar filings based on the following criteria: (i) the inversion was a publicly traded U.S. comp any before inversion; (iii) stocks of the surviving company are still publicly traded in the United States; 18 (iv) the transaction was closed/completed by December 31, 2015. We are left with 63 inversions for 17 The matched pairs whose exchange tickers match but the company names and CUSIPs do not match are omitted. 18 Hence, firms whose stocks are traded on the OTC market after inversion are not included in the treated sample in this paper. 24 empirical testing purposes (Appendix B). 5.2. S ummary Statistics Figure 3 plots the total number of inversions and the number of inversions through each strategy from 1993 to 2014. Before 2004, 28 companies inverted overseas, among which the majority (24) inverted purely. Nevertheless, after the passin g of Section 7874 in 2004, inverting by merging with a foreign entity replaces pure inversion as the most common inversion strategy. Moreover, it has become the dominant strategy especially after 2010. For instance, while there have been only 4 pure invers ion announcements since 2010, there were 13 M&A inversions in 2014 alone. Spin - of f inversion is more frequent after 2004 than before, but the number of this type is negligible. Figure 3. Inversion Volume. The figure displays the number of inversion announ cements from 1993 to 2014. Panel A displays the overall announcements. The other three panels display the number of inversions for all three kinds of inversions: pure inversion (Panel B), M&A inversion (Panel C), and spin - off inversion (Panel D). The integ ers above the bars indicate the number of inversions announced in that year. 25 Table 1 displays the number of inversions announced in each industry. Generally, inver sions occur in a wide range of industries: they exist in 26 out of 48 Fama - French industri es. That being said, inversions cluster mainly in three industries: pharmaceutical products (14), petroleum and natural gas (13), and insurance (8). A firm in the petroleum and natural gas industry inverts mainly through a pure inversion strategy, whereas a fi rm in the pharmaceu tical products industry commonly inverts through mergers. We also display 17 destination countries in Table 2, and 6 of them are tax havens (Hines and Rice, 1994; Dharmapala and Hines, 2009). The second column in Table 2 gives the to tal number of inversions for each destination; Bermuda is the most popular destination (26 inversions out of 88). The other popular destination jurisdictions are Ireland (13), the Cayman Islands (10), the United Kingdom (9), Canada (8), and the Netherlands destinations have been shifting from less developed offshore tax haven jurisdictions such as Bermuda and the Cayman Islands to more developed countries such as Ireland, the Netherlands, and the United Kingdom (see column (3) of Table 2). Moreover, this shifting has become more apparent after 2010. There have been 12 inversions to Ireland, 8 to the United Kingdom and five to the Netherlands, but only four to Bermuda since 2010 (see column (5) of Table 2). Inverting firms p refer more developed economies after 2004 because it is easier for them imposed by Section 7874 of the 2004 American Jobs Creation Act. Interestingly, even for th e firms inverting to Bermuda after 2004, M&A is also a common strategy to take. These firms were often acquired by a Bermuda - incorporated firm that pure - inverted to Bermuda before 2004. For example, the insurance company Argonaut Group inverted to Bermuda in 2007 by merging with the Bermuda - incorporated PXRE Group in 2007, and the PXRE Group pure - inverted to Bermuda 26 earlier in 1999. Table 1: Inversion Announcements Across Industries. The industry classification follows Fama - French (1997). Inversions exist in 26 out of 48 Fama - French industries and cluster mainly in three industries: pharmaceutical products (14), petroleum and natural gas (13), and insurance (8). Industries All Pure M&A Spin - off Agriculture 1 0 1 0 Automobile and Trucks 1 1 0 0 Business Services 4 1 2 1 Chemicals 3 1 2 0 Communication 4 2 2 0 Computers 2 1 1 0 Construction 2 2 0 0 Construction Materials 2 1 0 1 Consumer Goods 2 1 1 0 Electrical Equipment 2 1 1 0 Electronic Equipment 5 2 1 2 Finance Trading 4 2 2 0 Food Product s 1 0 1 0 Insurance 8 5 3 0 Machinery 3 2 1 0 Medical Equipment 4 0 3 1 Non - metallic and Industrial Metal Mining 1 1 0 0 Petroleum and Natural Gas 13 8 4 1 Pharmaceutical Products 14 2 12 0 Precious Metals 3 3 0 0 Rubber and Plastic Products 1 1 0 0 Shipping Containers 1 0 1 0 Textiles 1 1 0 0 Transportation 3 1 2 0 Wholesale 1 1 0 0 restaurants, hotels, motels 2 1 1 0 Total 88 41 41 6 27 Table 2: Inversion Destinations. This table displays the inversion destination jurisdictions. Overall, Be rmuda, Canada, the Cayman Islands, Ireland, the Netherlands, and the United Kingdom are the most popular inversion destinations. Before the passing of Section 7874 of the 2004 American Jobs Creation Act, firms were mainly inverting to Bermuda (17) and the Cayman Islands (7). After 2004, fi rms were inverting to more developed countries such as Canada (7), Ireland (13), the Netherlands (5) and the United Kingdom (9). Bermuda appears to lose its charm to inverting in the last column i ndicates that the country is a tax - haven country, following the de fi nition in Hines and Rice (1994). Destination All Before 2004 After 2004 After 2010 Tax Haven Antigua 1 1 0 0 Australia 1 0 1 1 Austria 1 0 1 0 Bermuda 26 17 9 4 Y British Virgin Is lands 1 0 1 0 Y Canada 8 1 7 2 Cayman Islands 10 7 3 0 Y Denmark 1 0 1 0 Ireland 13 0 13 12 Y Israel 1 0 1 1 Jersey 1 0 1 0 Luxembourg 2 0 2 1 Marshall Islands 3 0 3 0 Y Netherlands 7 2 5 5 Switzerland 2 0 2 1 Y United Kingdom 9 0 9 8 To tal 88 28 60 36 6 Table 3 displays the summary statistics for the main variables used in this analysis. Panel A and Panel B report the statistics for the group of inversions and the group of 13,152 UMCs separately. All the UMCs reported in Panel B have b een incorporated in the United States since their inception. We observe that inverting companies tend to have a lower average cost of equity than UMCs. They are also larger in size, have higher leverage ratios, and are more profitable on average, which is consistent with the findings in the literature. Panel C and Panel D give the summary statistics for firms taking pure and M&A inversion strategies. Slightly diff erent from 28 pure - inverting firms, M&A - inverting firms typically have a lower leve rage than UMCs. But they are both larger in size and more profitable when compared to an average UMC. Table 3: Summary Statistics for the Main Variables Used in This Analysis. Panel A: Treatment Group (Firms that Reincorporated from U.S. to Overseas) Number of Firms: 63 Variable Obs. Mean Std. Dev. Min Max Cost of Equity 4529 0.108 0.018 0.002 0.149 Risk Premium 4529 0.017 0.009 - 0.025 0.035 Total Assets (in Million Dollars) 4458 7211.73 13052.23 12.74 135840.7 Total Assets (in Log Points) 4458 7.598 1.787 2.545 1 1.819 Leverage Ratio 4452 0.572 0.261 0 2.508 Return on Assets 4447 0.005 0.059 - 1.296 0.444 Panel B: U.S. Multinational Firms Number of Firms 13152 Variable Obs. Mean Std. Dev. Min Max Cost of Equity 341244 0.110 0.017 0.00008 0.291 Risk Pre mium 341244 0.018 0.008 - 0.021 0.066 Total Assets (in Million Dollars) 337447 5361.56 47309.95 0.076 3281222 Total Assets (in Log Points) 337447 6.228 2.006 - 2.577 15.004 Leverage Ratio 337145 0.546 0.425 - 0.004 95.163 Return on Assets 336997 - 0.007 0. 151 - 28.427 10.053 Panel C: Pure Inversion Number of Firms 31 Variable Obs. Mean Std. Dev. Min Max Cost of Equity 2555 0.106 0.020 0.002 0.147 Risk Premium 2555 0.016 0.009 - 0.025 0.035 Total Assets (in Million Dollars) 2543 6756.00 10391.15 2 4.691 115505.4 Total Assets (in Log Points) 2543 7.647 1.758 3.206 11.657 Leverage Ratio 2543 0.612 0.281 0 2.508 Return on Assets 2536 0.002 0.061 - 1.296 0.398 29 Panel D: M&A Inversion Number of Firms 26 Variable Obs. Mean Std. Dev. Min Max Cost of Equity 1582 0.110 0.016 0.018 0.149 Risk Premium 1582 0.019 0.008 - 0.020 0.035 Total Assets (in Million Dollars) 1552 6350.10 13200.93 12.738 135840.7 Total Assets (in Log Points) 1552 7.406 1.778 2.545 11.819 Leverage R atio 1546 0.499 0.230 0.014 1.840 Return on Assets 1548 0.008 0.060 - 0.759 0.444 30 6. RESULTS 6.1. Linear Regression I start by checking whether we have a comparable control sample to the treated sample in terms of the three matching variables, log(tota l assets), leverage ratio, and ROA. I use the inverting firms as the treated sample and the closest match as the control sample (CS1). Table 4 reports the 25th percentiles, 50th percentiles, and 75th percentiles of the three matching variables, and I find that the distributions are very close between the treated and control samples. 19 The p - values of the mean and distribution comparisons reported in the last two columns of Table 4 further confirm that the treated and control groups have similar distributions . Table 4: Comparison of Treated and Control Groups: One - to - One Matching. This table compares the distributions of the matching variables, log(total assets) , leverage ratio , and ROA , in the treated sample (T) to its one - for - one matched control sample (CS 1). This table reveals that the CS1 is very similar to the treated sample. The 25th percentiles, 50th percentiles, and 75th percentiles of the three matching variables are very close. The p - values of the mean and distribution comparisons reported in the la st two columns further confirm that these three matching variables have the same distribution between the treated and control samples. Variables 25th Percentile 50th Percentile 75th Percentile P - Values for Mean P - Values for Dist. T CS1 T CS1 T CS1 l og(total assets) 6.60 6.30 7.86 7.72 8.70 8.57 0.86 0.93 Leverage Ratio 0.39 0.40 0.55 0.52 0.75 0.65 0.65 0.34 ROA - 0.01 0.00 0.01 0.01 0.02 0.03 0.32 0.90 Table 5 provides the results of the benchmark linear regressions (equation (5)) using CS1 as th e control sample. Columns (1) and ( 2 ) of Table 5 report the regression results by including all 19 The only noticeable difference is that the 25th percentile of treatment sample ROA is negative, whereas that of the CS1 is positive. But considering that their magnitudes are very small, this difference is economically insignificant. 31 63 inversions as treated sample, and I find no economically and statistically significant impact of inversions on the cost of equity. The magnitude of foreign r eincorporation dummy D it is roughly the same with and without the control variables, which indicates that the impacts of the control variables on the cost of equity have been largely captured by the reincorporation dummy. Next I look at the effects of inve rsion on the cost of equity by conditioning on inversion strategies (see columns ( 3) - ( 6 ) in Table 5). Results of pure - inverting firms are reported in columns ( 3 ) and ( 4 ). As a benchmark regression, I do not include any controls in in the first specification, and I find that a pure inversion increases the cost of equity by 23 basis points (bps) (column (3)), which is marginally significant at the 5% significance level after controlling for firm - level heteroskedasticity and serial correlation. Sin ce the mean risk premium of pure - inverting firms is 0.0116 before inverting, pure inversion increases the risk premium of these firms by 19.8% on average. In the second specification (column (4)), I include log(total assets), leverage ratio , and ROA as con trol variables and find no significant changes in the coefficient of the reincorporation dummy. The increase in the cost of equity after a pure inversion reveals that the shareholders of pure - inverting firms raise their required rate of return, which coul d be caused by the following reasons. First, after a pure inversion, the firm is governed under the corporate laws of the destination jurisdiction, thus losing the protection of U.S. corporate laws. Compared to the corporate laws in Bermuda and the Cayman Islands, the Delaware legal framework and interests (Daines, 2001), and provide more protection to the shareholders against hostile take - overs (Talley, 2015). Second, political risks and economic policy uncertainty are greater in pure inversion destinations such as Bermuda than they are in the United States. Third, pure - inverting 32 et al. (2015) show that inverting firms tend to have higher bid - ask spreads, less liquid stocks, and less institutional ownership. Investors put a lower value on the cash on their balance sheets after inversions as well. Columns (5) and (6) of Table 5 rep ort the impact of M&A inversions on the cost of equity. An M&A - inversion, which is significant at the 1% significance level, with (without) controlling matching variables . Since the mean risk premium of M&A - inverting firms is 0.018 prior to the inversion, no significant change in the estimates after controlling for industry fixed effects. The decrease in the cost of equity after an M&A inversion is a result of several mixed effects. First, the new destination jurisdictions of M&A inversions are better developed than destinations of pure inversions and have similar political and eco nomic risks as in the United States. Therefore, U.S. investors would be better protected under the corporate laws of M&A inversion destinations and are more confident about their economic developments compared to pure inversions. Second, different from pur e inversions in which the only material change is the change in the registration jurisdictions, inverting by merging with a foreign entity is bundled with the change c ompositions. An M&A - inverting firm commonly merges with a foreign firm which is usually ure economic 33 Table 5: Effect of Inversion on the Cost of Equity: Benchmark Linear Regression. This table reports the benchmark results of linear regressions investigating the impact of inversions on the cost of equity. The control samples used in this table are the closest matches for each inverting company based on industry, measurement quarter, log(total assets) , leverage ratio , and ROA (CS1). I run regressions separatel y using all inversions, pure inversions, and M&A inversions as treated samples, respectively. In columns (1), (3), and (5), we estimate the regression: , and in the other columns, I estimate the regression: , where is the risk premium of company i at time t . is the incorporation status dummy variable of company i at time t . If company i is incorporated in the United States at time t , equals 0. If the company reincorporates overseas and thus becomes foreign - incorporated at time t , equals 1. The variable denotes the unobserved firm fixed effects, and denotes the time fixed effects. The error term includes all other time - varying unobservable shocks to the cost of equity. Log(Total assets) , leverage ratio , and ROA are us ed as controls and included in . Standard errors are robust to heteroskedasticity and serial correlation and are reported in parentheses. Note that I use to indica te that the estimates are signifi cant at 1%, 5%, and 10% sign ificance levels, respectively. Variables All Pure Inversion M&A Inversion (1) (2) (3) (4) (5) (6) D - 0.0001 - 0.0001 0.0021 0.0023 - 0.0029 - 0.0033 (0.0008) (0.0008) (0.0010)** (0.0011)** (0.0009)** * (0.0011)** * Log( TA ) - 0.0005 - 0.0005 - 0.00002 (0.0002)** (0.0011)** (0.0004) Leverage - 0.0005 - 0.0009 - 0.0019 (0.0008) (0.0011)** (0.0010)* ROA 0.0002 - 0.0031 - 0.00007 (0.0010) (0.0030) (0.0011) Constant 0.0058 0.0103 0.0059 0.0116 0.0060 0.0070 (0.0010)** * (0.0022) ** * (0.0018)** * (0.0038)** * (0.0004)** * (0.0024)** * Adj. R - squared 0.82 0.81 0.74 0.73 0.87 0.87 Firms in sample 105 105 52 52 52 45 Observations 5272 5192 2410 2395 2395 2410 Firm FE Yes Yes Yes Yes Yes Yes Time FE Yes Yes Yes Yes Yes Yes Last, sin ce the two companies in an M&A inversion are generally household names in the same industry, an M&A transaction decreases the global competition of that industry and expands 34 in the cost of equity is a result of the synergy created from less intra - industry competition and geographical business diversification. Overall, the estimated negative effect of M&A inversions on the cost of equity indicates that the positive synergy created fr om an M&A worries about the political and economic risks in the new destination jurisdiction. 6.2. Fractional Response Model Fractional response models are suitable for capturing the property that the cost of equity t ypically lies between zero and one, which makes the diff - in - diff estimate more precise. I use the cost of equity as the dependent variable in three estimation equations following equation (5): fractional logit, fractional probit, and fractional complementa ry - log - log. 20 All models are fully saturated in that firm - and time - fixed effect s are all controlled. Standard errors are fully robust and are clustered at the firm level. Panel A of Table 6 gives the results of regressions using all inverting firms as the treated group. As shown, the impact of inversion on the cost of equity is economically and statistically insignificant across all three fractional response models, which is consistent with the linear regression findings in Table 5. Panel B reports the regr ession results of pure - inverting firms. I find that a pure inversion increases the cost of equity by up to 103bps after controlling for firm - and time - fixed effects (see columns (2), (4) and (6)). 21 Moreover, the estimates are marginally significant at the 5% significance level. Since the mean level of a pure - 20 In particular, the complementary - log - log model allows for the existence of extreme values. 21 The estimates in regressions with controls (see columns (2), ( 4 ), and ( 6 )) are larger than those without any control variables (columns (1), ( 3 ), and ( 5 )). 35 before inversion, pure inversion increases the cost of equity by up to 10%, and it almost doubles the risk premium of inverting firms (from 116bps to 219b ps). Additionally, the similarity of the estimates across three models precludes the concern that the estimation over relies on the functional form of in equation (6). Panel C reports the results of M&A - inverting firms. I find that an M&A inversion decreases the cost of equity by up to 139bps on average, which is statistically significant at the 1% significance level. Since the average cost of equ ity and the risk premium before an M&A inversion are 11% and 1.8% respectively, an M&A inversion decreases the cost of equity by up to 12.6% and the risk premium by up to 77%, which indicates that an M&A inversion tends to greatly lower In sum, the nonlinear diff - in - diff estimates of fractional response regressions are larger than the estimates of linear regressions because fractional response models incorporate the [0; 1] data structur e of the cost of equity and therefore make the estimation more precise. That being said, fractional response regression results are consistent with linear regression results in terms of the statistical significance and the direction of the change in the co st of equity. 36 Table 6: Effect of Inversion on the Cost of Equity: Benchmark Fractional Response Models. This table reports the results of regressions investigating the impact of inversions on the cost of equity. The control samples used in this tab le are the closest matches for each inverting company based on industry, measurement quarter, log(total asset), leverage ratio, and ROA (CS1). Panel A uses all the inversions in the treated sample, and Panels B and C use pure inversions and M&A inversions as treated samples, respectively. Within each panel, I estimate three fractional response models: fractional logit, fractional probit, and complementary - log - log. In columns (1), ( 3 ), and ( 5 ) of each panel, I estimate the regression . In columns (2), ( 4 ), and ( 6), I estimate the regres sion , where is the cost of equity of fi rm i at time t , and represents the three functional forms. is the incorporation status dummy variable of company i at time t . If company i is incorporated in the United States at time t , equals 0. If the company reincorporates overseas and th us becomes foreign - incorporated at time t , equals 1. The variable denotes a full set of firm fi xed e ff ects, and denotes a full set of time fi xed e ff ects. The error term includes all other time - varying unobservable shocks to the cost of equity. Log(total assets), leverage ratio, and ROA are used as controls and included in . Fully robust standard errors are reported in parentheses. Note at 1%, 5%, and 10% significance levels, respectively. Model Fractional Logit Fractional Probit Fractional C - Log - Log Variables (1) (2) (3) (4) (5) (6) Panel A: Across all inverted firms D - 0.0061 - 0.0073 - 0.0023 - 0.0029 - 0.0062 - 0.0073 (0.0333) ( 0.0338) (0.0177) (0.0179) (0.0312) (0.0317) Log(TA) - 0.0203 - 0.0108 - 0.0190 (0.0091)** (0.0048)** (0.0086)** Leverage - 0.0212 - 0.0110 - 0.0200 (0.0348) (0.0184) (0.0326) ROA 0.0084 0.0037 0.0083 (0.0444) (0.0237) (0.0415) Constan t - 2.5212 - 2.3174 - 1.4453 - 1.3367 - 2.5593 - 2.3684 (0.0612)** * (0.1038)** * (0.0301)** * (0.0538)** * (0.0589)** * (0.0981)** * Impact on ICC - 0.0006 - 0.0007 - 0.0004 - 0.0005 - 0.0006 - 0.0008 (0.0032) (0.0033) (0.0033) (0.0033) (0.0032) (0.0032) Observations 5272 5192 5272 5192 5272 5192 37 Model Fractional Logit Fractional Probit Fractional C - Log - Log Variables (1) (2) (3) (4) (5) (6) Panel B: Pure Inversion D 0.0928 0.1026 0.0497 0.0550 0.0867 0.0959 (0.0506)** (0.0530)** (0 .0268) ** (0.0280)** (0.0475)** (0.0498)** Log(TA) - 0.0236 - 0.0126 0.0221 (0.0158) (0.0083) (0.0149) Leverage - 0.0408 - 0.0212 - 0.0382 (0.0534) (0.0282) (0.0501) ROA - 0.1537 - 0.0219 - 0.1442 (0.1279) (0.0683) (0.1197) Constant - 2.5323 - 2 .2679 - 1.4491 - 1.3078 - 2.5710 - 2.3236 (0.1109)** * (0.1849)** * (0.0538)** * (0.0950)** * (0.1070)** * (0.1753)** * Impact on ICC 0.0090 0.0100 0.0093 0.0103 0.0090 0.0100 (0.0050)** (0.0053)** (0.0051)** (0.0053)** (0.0050)** (0.0053)** Observations 2410 2395 2410 2395 2410 2395 Panel C: M&A Inversion D - 0.1247 - 0.1454 - 0.0658 - 0.0767 - 0.1171 - 0.1365 (0.0412)** * (0.0501)** * (0.0216)** * (0.0263)** * (0.0388)** * (0.0472)** * Log(TA) - 0.0009 - 0.0006 - 0.0008 (0.0161) (0.0085) (0.0151) Leverage - 0 .0821 - 0.0430 - 0.0772 (0.0425)* (0.0224)* (0.0399)* ROA - 0.0026 - 0.0014 - 0.0025 (0.0445) (0.0235) (0.0418) Constant - 2.4893 - 2.4492 - 1.4306 - 1.4092 - 2.5279 - 2.4905 (0.0235)** * (0.1001)** * (0.0116)** * (0.0532)** * (0.0225)** * (0.0937)** * Impact on ICC - 0.0118 - 0.0136 - 0.012 - 0.0139 - 0.0117 - 0.0135 (0.0037)** * (0.0045)** * (0.0038)** * (0.0046)** * (0.0037)** * (0.0044)** * Observations 2446 2410 2446 2410 2446 2410 Firm FE Yes Yes Yes Yes Yes Yes Time FE Yes Yes Yes Yes Yes Yes 38 6.3. Robus tness Checking 6.3.1. One - for - Four Matchin g In this section, I check whether our results are being driven by the choice of control sample size in my earlier regressions. Using the one - for - one match as the control sample might restrict the generalization of the findings. Following Abadie, Drukker, Herr, and Imbens (2004), as a robustness check, I use the first four closest matches of each inverting firm as the control sample. Summary statistics for the comparison between the inverting/treated group and its o ne - for - four matching group (CS4) are reported in Table 7. The 25th, 50th, and 75th percentiles of the matching variables are close between the treated and control samples. The p - values for the mean and distribution comparisons further confirm that the trea ted and control group s are statistically equivalent. Table 7: Comparison of Treated and Control Groups: One - to - Four Matching. This table compares the distributions of the matching variables, log(total assets) , leverage ratio , and ROA , in the treated samp le (T) to its one - for - four matched control sample (CS4). This table reveals that the CS4 is very similar to the treated sample. The 25th percentiles, 50th percentiles, and 75th percentiles of the three matching variables are very close. The p - values of the mean and distribution comparisons reported in the last two columns further confirm that these three matching variables have the same distribution between the treated and control samples. Variables 25th Percentile 50th Percentile 75th Percentile P - Values for Mean P - Values for Dist. T CS4 T CS4 T CS4 log(total assets) 6.60 6.30 7.86 7.49 8.70 8.28 0.23 0.11 Leverage Ratio 0.39 0.43 0.55 0.53 0.75 0.67 0.85 0.59 ROA - 0.01 - 0.01 0.01 0.01 0.02 0.02 0.81 0.39 39 Table 8: Robustness Check for Control Sa mple Size: Linear Regression. This table reports the results of regressions investigating the robustness of control sample size in linear regressions. The control samples used in this table are the first four closest matches for each inverting company bas ed on industry, measurement quarter, log(total assets), leverage ratio, and ROA (CS4). I run regressions separately using all inversions, pure inversions, and M&A inversions as treatment samples, respectively. In columns (1), (3), and (5), we estimate the regression: , and in the other columns, I estimate the regression: , where is the risk premium of comp any i at time t . is the incorporation status dummy variable of company i at time t . If company i is incorporated in the United States at time t , equals 0. If the company reincorporates overseas and thus becomes foreign - incorporated at t ime t , equals 1. The variable denotes the unobserved firm fixed effects, and denotes the time fixed effects. The error term includes all other time - varying unobservable shocks to the cost of equity. l og(Total assets) , le verage ratio , and ROA are used as controls and included in . Standard errors are robust to heteroskedasticity indicate that the estimates are significa nt at 1%, 5%, and 10% significance levels, respectively. Variables All Pure Inversion M&A Inversion (1) (2) (3) (4) (5) (6) D - 0.0001 - 0.0002 0.0017 0.0018 - 0.0021 - 0.0020 (0.0006 ) (0.0006) (0.0009)* * (0.0010)* * (0.0009)* ** (0.0009)* * Log(TA) - 0.00 03 - 0.0003 - 0.0004 (0.0002)* * (0.0002) (0.0002) Leverage Ratio - 0.0003 - 0.0006 - 0.0008 (0.0004) (0.0009) (0.0006) Return on Assets 0.0011 - 0.0003 0.0011 (0.0008) (0.0018) (0.0012) Adjusted R - squared 0.79 0.79 0.72 0.72 0.85 0.85 Firms in sample 247 247 127 127 106 106 Observations 10436 10356 4622 4607 5213 5177 Firm FE Yes Yes Yes Yes Yes Yes Time FE Yes Yes Yes Yes Yes Yes I repeat the linear and fractional response regressions using CS4 as the control group. The results ar e reported in Table 8 (linear) and Table 9 (fractional response). The linear diff - in - diff estimates are still small in magnitude and statistically insignificant if we pool all inverting firms in the treated sample (Panel A of Table 8), which is consistent with our findings in Table 5. The 40 regression results for pure - inverting firms (Panel B of Table 8) and M&A - inverting firms (Panel C of Table 8) are also consistent with our earlier findings. On average, a pure inversion increases the cost of equity, wherea s an M&A inversion decreases it. The diff - in - diff estimates are statistically significant and close in magnitude to the findings using CS1 as the control group. Table 9: Robustness Check for Control Sample Size: Fractional Response Models. This table repo rts the results of regressions investigating the robustness of control sample size in fractional response regressions. The control firms used in this table are the first four closest matches for each inverting company based on Fama - French industry, measure ment quarter, log(total asset), leverage ratio, and ROA (CS4). Panel A uses all the inversions in the treated sample, and Panels B and C use pure inversions and M&A inversions as treated samples, respectively. Within each panel, I estimate three fractional response models: fractional logit, fractional probit, and complementary - log - log. In columns (1), (3), and (5) of each panel, I estimate the regression . In columns (2), (4), and (6), I estim ate the regression , where is the cost of equity of firm i at time t , and represents the three functional forms. is the incorporation status dummy variab le of company i at time t . If company i is incorporated in the United States at time t , equals 0. If the company reincorporates overseas and thus becomes foreign - incorporated at time t , equals 1. The variable denotes a full set of firm fixed effects, and denotes a full set of time fixed effects. The error term includes all other time - varying unobservable shocks to the cost of equity. Log(total assets), leverage ratio, and ROA are used as controls and included in . Fully robust standard errors are reported in parentheses. Note that I sig nificance levels, respectively. Model Fractional Logit Fractional Probit Fract ional C - Log - Log Variables (1) (2) (3) (4) (5) (6) Panel A: Across all inverted firms D - 0.0062 - 0.0069 - 0.0024 - 0.0028 - 0.0062 - 0.0069 (0.0278) (0.0285) (0.0149) (0 .0153) (0.0260) (0.0267) Log(TA) - 0.0144 - 0.0076 - 0.0136 (0.0070)* * (0.0037)* * (0.0066)* * Leverage - 0.0151 - 0.0081 - 0.0141 (0.0182) (0.0097) (0.0170) ROA 0.0474 0.0251 0.0444 (0.0356) (0.0191) (0.0332) Impact on ICC - 0.0006 - 0.0007 - 0.0004 - 0.0005 - 0.0006 - 0.0007 (0.0027) (0.0027) (0.0028) (0.0028) (0.0027) (0 .0027) Observations 10436 10356 10436 10356 10436 10356 41 Model Fractional Logit Fractional Probit Fractional C - Log - Log Variables (1) (2) (3) (4) (5) (6) Panel B: Pure Inversion D 0.0773 0.0805 0.0418 0.0435 0.0721 0.0750 (0.0427 )** (0.0442)* * (0.0227)** (0.0235)* * (0.0399)** (0.0414)* * Log(TA) - 0.0119 - 0.0063 - 0.0112 (0.0110) (0.0057) (0.0103) Leverage - 0.0275 - 0.0149 - 0.0256 (0.0401) (0.0209) (0.0377) ROA - 0.0126 - 0.0078 - 0.0112 (0.0777) (0.0416) (0.07 25) Impact on ICC 0.0075 0.0078 0.0078 0.0082 0.0074 0.0078 (0.0042)** (0.0044)* * ( 0.0043)** (0.0045)* * (0.0042)* (0.0044)* Observations 4622 4607 4622 4607 4622 4607 Panel C: M&A Inversion D - 0.0898 - 0.0834 - 0.0477 - 0.0444 - 0.0842 - 0.0781 (0.0365 )** * (0.0391)* * (0.0193)** * (0.0206)* * (0.0343)** * (0 .0366)** Log(TA) - 0.0158 - 0.0083 - 0.0149 (0.0105) (0.0056) (0.0098) Leverage - 0.0356 - 0.0191 - 0.0332 (0.0231) (0.0123) (0.0217) ROA 0.0449 0 .0241 0.0419 (0.0489) (0.0262) (0. 0456) Impact on ICC - 0.0086 - 0.008 - 0.0088 - 0.0082 - 0.0085 - 0.0079 (0.0033)** * (0.0036)* * (0.0034)** * (0.0037)* * (0 .0033)*** (0.0035)* * Observations 5213 5177 5213 5177 5213 5177 Firm FE Yes Yes Yes Yes Yes Yes Time FE Yes Yes Yes Yes Yes Yes The fractional response regression results using CS4 as the control group are also con sistent with the results obtained using CS1. The estimate of the impact of inversion on the cost of equity 42 using all inverting firms as the treated sample (Panel A of Table 9 ) is still insignificant. Panel B and Panel C show that the impact on the cost of equity is positive (up to +82bps) for pure inversions and negative (up to - 88bps) for M&A inversions. More over, the diff - in - diff estimates in Table 9 are slightly smaller tha n the estimates obtained using CS1 because CS4 increases the control sample size by including less perfectly matched observations. In summary, my regression results are robust to the choice of control sample observations. 6.3.2. Augmented Linear Regressio n In our prior regressions, we treat the reincorporation dummy as an exogenous variable, and tability, no other factors af cost of equity are correlated with its inversion decision. In this section, I address the concern that st of equity by looking at an augmented linear model and using CS1 as the control sample. I identify four variables that might be missing in our earlier regressions: Cash Holdings, Capital Expenditure, R&D Expenditure, and Tax Rate. Cortes et al. (2015) have demonstrated that these variables are Cash holdings will affect s for two possible reasons. First, cash held inside the firm could be regarded as a high - quality collateral, which tends to lower the cost of equity. On the other hand, holding too much cash may imply that the firm lacks good investment opportunities, whic h tends to increase the cost of equity (or equivalently, the required rate of return by stock investors). Whichever the direction is, both reasons demonstrate that cash held inside the firm is correlated with the cost of equity. Section 3 has discussed tha t a firm with more cash held in a foreign subsidiary has more incentive to invert to repatriate the cash in a less costly manner. 43 Therefore, the cash holdings of a firm might correlate with both the cost of equity and the inversion decision. High expenditu res in capital and R&D in a firm have two implications. A firm with high expenditures in capital and R&D will have a high demand for equity to finance such activities, and this tends to increase the cost of equity. It might also signal to the equity invest ors that the firm has good investment opportunities, and this tends to lower the cost of equity. Both implications demonstrate that the capital expenditure and R&D expenditure are correlated with the cost of equity. They also appear to be correlated with i nversion decisions in that a multinational firm with high capital and R&D expenditures has a higher demand for repatriating its overseas cash to spend on such activities. uity is positively correlated the corporate taxes. Since avoiding high U.S. corporate taxes is the main in the augmented regression model. Technically, besides r einc orporation d ummy, log(total a ssets) , Leverage , and ROA , I in clude Cash Holdings , Capital Expenditure , R&D Expenditure , and Tax Rate in x it of equation (5 ) to eliminate the potential missing variable issue in Table 5, and I report the results in Table 10. I find that generally there are no significant diff erences in the diff - in - diff estimates with and without these extra control variables: a pure inversion still increases the inverting equity, and an M&A inversion still decreases it. The resul ts reported are statis tically significant, which implies that our earlier results are robust to the potential missing variable issue. 44 Table 10: Augmented Linear Regression. This table reports the augmented linear regression results by adding extra contr ols in of equation (5): . The control sample used in this table is CS1. I run regressions separately using all inversions (column (1)), pure inversions (column (2)), and M&A inver sions (column (3)) as treated samples. Log(total assets) , Leverage , ROA , Cash Holdings , Capital Expenditure , R&D Expenditure , and Tax Rate are included in x it as control variables. Firm and time fixed effects are both controlled. Standard errors are robust to respectively. Variables All Pure Inversion M&A Inve rsion (1) (2) (3) D - 0.0005 0.0012 - 0.0024 (0.0007) (0.0005)*** (0.0010)*** Log(total assets) - 0.0001 0.0005 0.0000 (0.0004) (0.0003) (0.0005) Leverage - 0.0002 - 0.0001 - 0.0013 (0.0007) (0.0009) (0.0015) ROA 0.0006 0.0004 0.0001 (0.0012) (0.0 022) (0.0015) Cash - 0.0004 - 0.0030 0.0002 (0.0011) (0.0011)*** (0.0011) Capital Expenditure - 0.0028 - 0.0162 0.0016 (0.0034) (0.0059)*** (0.0031) R&D Expenditure - 0.0004 0.0062 - 0.0004 (0.0021) (0.0048) (0.0025) Tax Rate - 0.0006 0.0002 - 0.0009 ( 0.0004) (0.0012) (0.0004)** Constant 0.0106 0.0080 0.0054 (0.0017)*** (0.0017)*** (0.0017)*** Adj. R - squared 0.85 0.91 0.87 Observations 1502 357 1081 Firm FE Yes Yes Yes Time FE Yes Yes Yes 6.4. Cash Holdings, Dividend Payments, and Capital Expen ditures As illustrated in Section 3, firms have an incentive to invert because of the less costly benefit of repatriating the cash held in foreign subsidiaries either to invest in projects in the United 45 States or to pay more dividends to U.S. shareholders. Therefore, the cash holdings in a pure - inverting firm will decrease after an inversion. However, diff erent from pure inversions, the affect ed by the cash holdings of the foreign acquirer. Ther efore, it is unclear how the cash holdings will change after an M&A inversion. I empirically test the cash holdings change for pure and M&A inversions separately in Table 11. Table 11: Effect of Inversion on Cash Holdings. This table reports the changes in cash holdings following an inversion. The control sample used in this table is CS1. I run regressions separately using all inversions (column (1)), pure inversions (column (2)), and M&A inversions (column (3)) as treated samples. The dependent variable, Cash Holdings, is defined to be the ratio of cash held within the firm to total assets. The explanatory variables include Reincorporation Dummy, log(total assets), Leverage, ROA, Capital Expenditure, R&D Expenditure, and Tax Rate. Firm and time fixed effe cts are both controlled. Standard errors are robust to heteroskedasticity and serial correlation, and are reported in parentheses. Note that I significance levels, r espectively. Variables All Pure Inversion M&A Inversion (1) (2) (3) D - 0.0488 - 0.1798 - 0.1593 (0.0514) (0.0753)*** (0.1003) Log(total assets) - 0.0299 - 0.0550 - 0.0071 (0.0160)** (0.0512) (0.0202) Leverage - 0.1629 - 0.0928 - 0.1778 (0.0577)*** (0.0 842) (0.0527)*** ROA 0.0080 0.8527 - 0.1180 (0.0771) (0.0941)*** (0.0649)** Capital Expenditure - 0.4239 - 0.0434 - 0.7438 (0.1649)** (0.3683) (0.1728)*** R&D Expenditure - 0.3946 - 0.8872 - 0.3740 (0.1068)*** (0.3657) (0.0883)*** Tax Rate - 0.0434 0.086 6 0.0522 (0.0282) (0.0757) (0.0312)* Constant 0.3944 0.5230 0.2373 (0.0757)*** (0.3213) (0.0659)*** 46 Variables All Pure Inversion M&A Inversion (1) (2) (3) Adj. R - squared 0.15 0.46 0.17 Observations 1502 357 1081 Firms in sampl e 58 24 29 Firm FE Yes Yes Yes Time FE Yes Yes Yes Regression results reported in Table 11 are consistent with our hypothesis proposed above. On the one hand, the ratio of cash to total assets decreases by about 18 percentage points and is statisticall y significant after a pure inversion (column (2)), which indicates that a pure inversion enables the U.S. subsidiary to utilize the cash held overseas, and subsequently the total cash holdings in the firm decrease. Additionally, for a firm with better prof itability (higher ROA), the cash holdings tend to be higher. On the other hand, though the impact of an M&A inversion on cash holdings is negative, it is statistically insignificant (column inversion, which indicates that the purpose of pure inversion is n ot to repatriate cash existing shareholders more dividends. Again, lacking dividend payment policy, we could not see the impact of inversion on dividend payments in the M&A cas e. The effects of corporate inversions on capital expenditures are reported in Table 13. A pure - the Cayman Islands. Considering that the cash holdings of the pu re - inverting firm decrease while their dividend payments do not change significantly, this indicates that pure - inverting firms repatriate their cash held in foreign subsidiaries to invest in potentially profitable projects in the United States. However, si milar to Table 11 and Table 12, the change in capital expenditures after an M&A inversion is statistically insignificant because the investment policy of the foreign 47 acquirer also affects the change in capital expenditures after an M&A inversion. Table 12 : Effect of Inversion on Dividend Payments. This table reports the changes in dividend payments following an inversion. The control sample used in this table is CS1. I run regressions separately using all inversions (column (1)), pure inversions (column ( 2)), and M&A inversions (column (3)) as treated samples. The dependent variable, Dividend , is defined to be the ratio of dividends paid to total assets. The explanatory variables include Reincorporation Dummy , log(total assets) , Leverage, ROA , Cash Holding s t - 1 , Capital Expenditure , R&D Expenditure , Tax Rate , and SG&A . Firm and time fixed effects are both controlled. Standard errors are robust to heteroskedasticity and serial correlation, and are reported to indicate that the estimates are significant at 1%, 5%, and 10% significance levels, respectively. Variables All Pure Inversion M&A Inversion (1) (2) (3) D - 0.0015 - 0.0018 - 0.0058 (0.0025) (0.0074) (0.0045) Log(total assets) - 0.0005 0.0015 - 0.0007 (0.0017) (0.0043) (0.0023) Leverage - 0.0066 0.0102 - 0.0089 (0.0035)* (0.0130) (0.0038)** ROA 0.0309 - 0.0369 0.0346 (0.0160)* (0.0688) (0.0168)** Cash Holdings t - 1 0.0001 0.0015 0.0024 (0.0060) (0.0155) (0.0089) Capital Expenditure 0.0472 0.1301 0.0339 (0.0279) (0.0962) (0.0369) R&D Expenditure 0.0976 0.0401 0.0809 (0.0548) (0.1003) (0.0670) Tax Rate - 0.0028 - 0.0014 - 0.0041 (0.0015) (0.0052) (0.0017)** SG&A - 0.0002 0.0126 - 0.0003 (0.0005) (0.0291) (0.0006) Constant 0.0071 - 0.0163 0.00 03 (0.0094) (0.0310) (0.0097) Adj. R - squared 0.02 0.16 0.02 Observations 1502 357 1081 Firms in sample 58 24 29 Firm FE Yes Yes Yes Time FE Yes Yes Yes 48 Table 13: Effect of Inversion on Capital Expenditures. This table reports the changes in capita l expenditures following an inversion. The control firms used in this table are CS1. We run regressions separately using all inversions (column (1)), pure inversions (column (2)), and M&A inversions (column (3)) as treated samples. The dependent variable, Capital Expenditures, is defined to be the ratio of capital expenditures to book value of total assets. The explanatory variables include Reincorporation Dummy , log(total assets) , Leverage , ROA , R&D Expenditure , and Tax Rate . Firm and time fixed effects ar e both controlled. Standard errors are robust to heteroskedasticity and serial correlation, and are reported in 1%, 5%, and 10% significance levels, respect ively. Variables All Pure Inversion M&A Inversion (1) (2) (3) D 0.0135 0.0258 0.0011 (0.0079)* (0.0113)** (0.0053) Log(total assets) 0.0024 0.0042 0.0035 (0.0021) (0.0100) (0.0025) Leverage 0.0165 - 0.0060 0.0204 (0.0067)** (0.0090) (0.0084)** ROA 0.0241 0.0521 0.0155 (0.0111)** (0.0527) (0.0106) R&D Expenditure 0.1433 0.3476 0.1117 (0.0374)*** (0.0700)*** (0.0528)** Tax Rate - 0.0012 0.0199 - 0.0046 (0.0053) (0.0113) (0.0060) Constant 0.0167 - 0.0122 0.0005 (0.0070)** (0.0576) (0.0070) Adj. R - squared 0.18 0.16 0.20 Observations 1502 357 1081 Firms in sample 58 24 29 Firm FE Yes Yes Yes Time FE Yes Yes Yes 49 7. CORPORATE INVERSION STRATEGIES AND EXISTING SHAREHOLDER VALUE 7.1. The Advantage of M&A Inversions over Pure Inversions In the former section, I have demonstrated that inverting by merging with a foreign entity is more favorable than inverting through a pure inversion strategy in terms of reducing financing costs. However, the tax savings through an M&A inversion strategy are usually less than those through a pure inversion strategy because the corporate tax rates in the destination jurisdictions of M&A inversions are usually higher than the tax rates in pure inversion destinations. Specifically, corporate tax rates in offshore tax - haven jurisdictions where pure inversions commonly reincorporate are usually zero, whereas the corporate tax rates in M&A inversion destination jurisdictions are usually higher than zero. For example, for the three commonly used M&A inversion destinat ions, Ireland, the United Kingdom, and Canada, their marginal corporate tax rates are 12.5%, 20%, and 26.5%, respectively, in 2016. Therefore, compared to the pure inversion strategy, taking an M&A inversion strategy lowers tax savings by around 20% depend ing on where the foreign acquirer is located. This begs the question , which strategy is more beneficial overall to existing shareholders? In this section, I look at the impact of the 2004 tax reform, which restricts the tax benefits in a pure inversion and induces firms, perhaps unintentionally, to use an M&A inversion strategy. Technically, I compare the existing shareholder value increase after a pure inversion to the increase after an M&A inversion using the following diff - in - diff model: (8) where is the market value of firm i t (in billions of dollars). equals 1 in the post - inversion stage and 0 in the pre - inversion stage. equals 1 for M&A - inverting firms and 0 for pure - inverting firms. The error term includes all the other 50 time - varying unobservable shocks to existing sharehol der value. It is easy to show that (9) which means that the co efficient of the interaction term, , measures the diff erence between the existing shareholder value increase in M&A inversions and the increase in pure inversions. The data used to estimate equation (8) are obt ained from CRSP. Before inversion, the pre - of outstanding shares. After inversion, the existing shareholder value is calculated as follows: (10) where is the pre - de notes the stock price and denotes the numb er of shares outstanding. denotes the U.S. inverting firm i pure - inverting fi rms, equals 100% because the pre - inversion shareholders still hold 100% of the firm after a pure i nversion. In contrast, is less than 100% for M&A inversions because when a U.S. multinational firm merges with a foreign company, only the surviving/combined firm is traded on the NYSE or NAS DAQ. The actual U.S. invert surviving company is obtained from Talley (2015), Appendix B, and from SEC filings (forms S - 4, 14D9, 14D6, and 8 - K). 22 The regression results are reported in Panel A of Table 14. Because the dependent variable is highly serially correlated, I report the Newey - West standard errors in parentheses. In column (1), I use the monthly data after 1978 for estimation.23 The estimate of (in boldface) is positive 22 Since the first inverting firm in our inversion sample, Helen of Troy, announced its pl ans to invert in 1993, I choose 1978 as the starting year to make sure there is a 15 - year window prior to its announcement. 51 and statistically significant at the 1% significance level. That means, all else equal, an M&A inversion delivers around $2.07 billion additional to existing shareholders than a pure inversion on average. Table 14: Existing Shareholder Value Change. This table reports the regression results of existing shareholder value change based on the following diff - in - diff model: , where is existing shareholder value of firm i at time t (in billions of dollars). equals 1 in the post - inversion stage and 0 in the pre - inversion stage. equals 1 for M&A - inverting firms and 0 for pure - invertin g firms. The error term includes all other time - varying unobservable shocks to existing shareholder value. I use all the pure - inverting and M&A - inverting firms in Panel A, and I drop the firms that pure - inverted after 2004 in Panel B. In column ( 1) of each panel, I use all the available monthly data from 1978 to 2016. Then I separately estimate the regression in a 10 - year, 5 - year, and 3 - year window, and report the results in columns (2), (3), and (4). Moreover, as a robustness check, to control fo r the potential price reaction to inversion before the announcement due to information leakage and insider trading, I drop the year prior to the inversion announcement month and redo the regressions in columns (2) to (4). Results for these censored tests a re reported in columns (5) to (7). Newey - West standard errors are reported in parentheses. Variables All 10 - Year Window 5 - Year Window 3 - Year Window Censor. 10 - Year Censor. 5 - Year Censor. 3 - Year (1) (2) (3) (4) (5) (6) (7) Panel A: All pure - inverted and M&A - inverted firms After 4.10 3.55 2.46 1.27 3.72 2.63 1.42 (0.23)** * (0.26)** * (0.31)** * (0.35)** * (0.26)** * (0.31)** * (0.36)** * MA 5.29 5.06 4.57 4.73 5.01 4.32 4.34 (0.39)** * (0.47)** * (0.59)** * (0.77)** * (0.49)** * (0.65)** * (0.93)** * After*M & A 2.07 2.88 4.60 4.78 2.93 4.86 5.16 (0.87)** * (0.99)** * (1.16)** * (1.34)** * (1.01)** * (1.20)** * (1.44)** * Constant 2.44 2.91 3.40 3.75 2.74 3.23 3.61 (0.08)** * (0.10)** * (0.14)** * (0.18)** * (0.10)** * (0.15)** * (0.20)** * Adj. R - Squared 0.08 0.09 0.1 0 0.09 0.09 0.10 0.10 Obs. 9143 7127 4506 2956 6584 3963 2413 Firms in Sample 53 53 53 53 53 53 53 52 Variables All 10 - Year Window 5 - Year Window 3 - Year Window Censor. 10 - Year Censor. 5 - Year Censor. 3 - Year (1) (2) (3) (4) (5) (6) (7) Panel B: Dropping the firms pure - inverted after 2004 After 4.08 3.47 1.97 0.98 3.66 2.15 1.10 (0.24)** * (0.28)** * (0.32)** * (0.37)** * (0.28)** * (0.32)** * (0.38)** * MA 5.61 5.40 4.81 4.98 5.37 4.58 4.63 (0.39)** * (0.47)** * (0.60)** * (0.77)** * (0.50)** * (0.66)** * (0.93)** * After*M& A 2.08 2.97 5.09 5.14 3.00 5.33 5.48 (0.58)** * (0.99)** * (1.16)** * (1.35)** * (1.01)** * (1.20)** * (1.45)** * Constant 2.13 2.56 3.16 3.50 2.37 2.98 3.32 (0.09)** * (0.11)** * (0.15)** * (0.19)** * (0.11)** * (0.16)** * (0.21)** * Adj. R - Squared 0.09 0.09 0.10 0.10 0.09 0.11 0.10 Obs. 8638 6641 4189 2749 6139 3687 2247 Firms in Sample 46 46 46 46 46 46 46 affect the ex isting shareholder value, I rerun the regression with shor ter windows. In the second spec ification, I use a 10 - year window (the data within 120 months before the announcement month and the data within 120 months after the completion month of inversions) for the regression. The regr ession results reported in column (2) of Table 14 are close to the results in column (1). Similarly, I run regressions with a 5 - year window and a 3 - year window. With these shorter windows, the estimate of is still positive and statistically signific ant, but the magnitude is larger. t before the announcement of in verting decisions. This reaction could be caused by information leakage and insider trading. As a robustness check, I delete the data within the 12 month s before the inversion announce ment and 53 redo the regressions in columns (2), (3), and (4), and I report the results with windows in columns (5), (6), and (7). I find no significant diff erence with and without censoring. Additionally, it can be shown that (11) which means that measures the increase in existing shareholder value if the firm inverts th rough a pure inversion strategy. The results in columns (1) to (7) in Table 14 reveal that although a pure shareholder value, and the increase is statistically significant. Inc luding firms that pure - inverted after the 2004 tax reform in the sample tends to lower the average existing shareholder value increase in pure inversions, and therefore overestimates the advantage of an M&A inversion strategy over a pure inversion strategy . That is because pure - 7874 will not be able to save as much in repatriation taxes. As a robustness check, in Panel B of Table 14, I drop all the firms that pure - inverte d after 2004 and redo the regression reported in Panel A. 23 All the estimates of reported in Panel B are positive and statistically significant at the 1% level and are very close in magnitude to the results reported in Panel A, which is consistent with our earlier results that the existing shareholder value tends to increase more if the firm inverts by merging with a foreign entity. 7.2. A Moral Hazard Implication In Section 6 and Section 7.1, I have established that compared to the pure inversion 23 Seven firms are dropped in this panel: Lazard, Freescale Semiconductor, Western Goldfields, Tim Hortons, ENSCO International, Styr on, and Delphi Automotive. 54 quity but also adds more value to existing shareholders. Therefore, theoretically, to maximize existing shareholder value, firm managers should choose an M&A inversion as the main inversion strategy. However, in reality, a pure inversion strategy was more preferable prior to the passing of Section 7874 of the 2004 American Jobs Creation Act, which effect ively restricts the tax savings from pure inversions. We observe 28 pure inversions from 1990 to 2004 but only three M&A inversions in the same period. The disagreement between theory and reality discussed above has two implications. First, inter est. Since the main objective of inversion is to save repatriation taxes by chang ing the incorporation country, reincorporating overseas without changing the headquarters location and management team taking a pure inversion strategy, managers still have the same level of control in the company. In taking a pure inversion strategy, though it adds less to existing shareholder value. Second, while the main objective of passing Section 78 74 is to curb corporate pure in versions and maintain the tax revenue paid to the U.S. government, the tax reform has two unintended consequences. First, managers are induced to take an M&A inversion strategy after the tax reform, which might not be expected by the policy makers. The trend of in verting to foreign countries has not been stopped by the 2004 reform, and there are even more M&A inversions from 2004 to 2014 than pure inver sions from 1990 to 2004 (38 ver sus 28). Therefore, the tax reform appears to hav e failed in stopping lost tax revenue and switches corporate inversions from one mode to another, echoing the current mainstream view that another tax reform is needed. 55 Second, and more important, the tax reform has an unintended consequence of reducing t he managerial agency problem because managers are induced to take an M&A inversion strategy, which adds m ore value to existing sharehold ers. Contrary to the common belief that economic interventions generally are detrimental to social welfare, the tax refo rm tends to increase shareholder wealth in the inversion case. Therefore, the 2004 tax reform appears to be successful. If the government policy aims at increasing shareholder wealth instead of maximizing tax revenue, no additional tax reform is needed. 56 8. CONCLUSION The past two decades have witnessed more than 80 U.S. multinational companies leaving to reincorporate overseas. The exodus of U.S. corporations for tax - haven countries has led to heated discussions among policy makers regarding possible tax and other regulatory reforms. For example, in his weekly presidential address on July 26, 2014, President Obama said, Even as corporate profits are as high as ever, a small but growing group of big corporations are fl eeing the country to get out of payin g taxes. They are keeping paying their fair share. 24 In this paper, I analyzed the changes in the cost of equity caused by a corporate inversion contingent on two major inversion strategies: pure inversions and M&A inversions. The linear and nonlinear regression results indicate that the impacts of inversions on the cost of equity between these tw o inversion strategies are diff erent. On average, a pure inversion increases the cost of equity by around 10%, whereas an M&A inversion decreases the cost of equity by around 13%. The diff erence arises from two sources: the diff erential political and econo mic risks associated with the new jurisdictions and the synergy created through the merger. My findings are robust to the control sample size and the missing variable issue. Moreover, I documented that while pure inversions do increase existing shareholder value, these shareholders benefit more from M&A inversions, and the results are economically and statistically significant and robust to various testing windows. My findings imply that the managers 24 Barack Obama, President, United States of America, Weekly Presidential Address: Closing Corporate Tax Loopholes (July 26, 2014) (the transcript is available at: https://obamawhitehouse.archives.gov/the - press - office/2014/07/26/w eekly - address - closing - corporate - tax - loopholes). 57 appear not to maximize existing shareholder value. By loo king at the impact of the 2004 tax reform that restricted tax savings under pure inversions, I am able to document that whereas before 2004 most inversions were pure, the ones after 2004 were done mostly through mergers. This finding suggests that the refo rm had an unintended consequence of reducing the managerial agency problem by making their more preferred mode unattractive and increasing existing shareholder value. There are two potential directions that future research on inversions could follow. One d impact on the cost of equity from the synergy created in an M&A transaction. This research would help us to better understand the pure effect and quantify the synergy created by merging with a foreign entity. The other direction is to investigate the consequences of the higher (lower) post - inversion cost of equity on corporate financing and investment behavior. This paper has provided some evide nce on the changes in cash holdings and capital expenditures after an inversion, but a more detailed analysis should be a valuable path to take in the future. 58 APPENDICES 59 APPENDIX A . List of Corporate Inversions Announced between 1993 and 2015 . This ap pendix lists 88 firms that announced plans to reincorporate overseas between 1993 and 2015. The first column of Table A1 lists the company name at the time of the inversion announcement (some firms changed their names after the inversion was completed). Th e second column lists the industry in which the firm is operating. Industry classification follows the 48 - industry classification system in Fama and French (1997). The third column lists the foreign in this column means that the firm announced plans to pure - invert abroad. The fourth column lists the inversion strategy: pure inversion, M&A inversion, or spin - of f inversion. The fifth column lists the destination jurisdiction. Columns (6) and (7) report the time manage ment announced their inversion decision and the time when the in column (8) indicates this company was/is listed in the S&P 500 portfolio. T able A1: List of Corporate Inversions Announced between 1993 and 2015 U.S. Target N ame Industry Foreign Acquirer Name Type Dest. Ann . F inish S & P Helen of Troy Consumer Goods N/A Pure Bermuda 1993 - 12 1994 Core Laboratories Petroleum and Natural Gas N/A Pure Netherla nds 1993 - 12 1994 Loral Communica tion N/A Pu re Bermuda 1996 - 01 1996 Triton Energy Petroleum and Natural Gas N/A Pure Cayman Islands 1996 - 02 1996 Chicago Bridge & Iron Constructio n N/A Pure Netherla nds 1997 - 03 1997 Tyco International Electronic Equipment ADT M&A Bermuda 1997 - 03 1997 Y Flirty G irl International Business Services N/A Pure Antigua 1998 - 05 1998 Xoma Pharmaceuti cal Products N/A Pure Bermuda 1998 - 11 1999 Gold Reserve Precious Metals N/A Pure Canada 1999 - 02 1999 Fruit of the Loom Textiles N/A Pure Cayman Islands 1999 - 03 1999 Transocean Offshore Petroleum and Natural Gas N/A Pure Cayman Islands 1999 - 05 1999 Y 60 Table A1 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann . F inish S& P Everest Reinsurance Holdings Insurance N/A Pure Bermuda 199 9 - 09 2000 PXRE Insurance N/A Pure Bermuda 1999 - 10 1999 White Mountains Insurance Group Insurance N/A Pure Bermuda 1999 - 10 1999 Trenwick Group Insurance LaSalle Re Holdings M&A Bermuda 1999 - 12 1999 Tycom Electronic Equipment N/A Spin - off Bermuda 200 0 - 01 2000 Applied Power Machinery N/A Pure Bermuda 2000 - 07 2000 Seagate Technology Computers N/A Pure Cayman Islands 2000 - 08 2000 Y Arch Capital Group Insurance N/A Pure Bermuda 2000 - 09 2000 Foster Wheeler Constructio n N/A Pure Bermuda 2000 - 12 2001 Cooper Industries Electrical Equipment N/A Pure Bermuda 2001 - 06 2002 Y Global Marine Petroleum and Natural Gas Santa Fe Int'l M&A Cayman Islands 2001 - 09 2001 Ingersoll - Rand Machinery N/A Pure Bermuda 2001 - 10 2001 Y Noble Drilling Petroleum and Natura l Gas N/A Pure Cayman Islands 2002 - 01 2002 Y Stanley Works Constructio n Materials N/A Pure Bermuda 2002 - 2 Failed Y Herbalife Wholesale N/A Pure Cayman Islands 2002 - 02 2002 Nabors Industries Petroleum and Natural Gas N/A Pure Bermuda 2002 - 06 2002 Weatherford Int'l Petroleum and Natural Gas N/A Pure Bermuda 2002 - 06 2002 Luna Gold Precious Metals N/A Pure Canada 2005 - 10 2005 Lazard Finance Trading N/A Pure Bermuda 2005 - 12 2005 61 Table A1 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann . F inish S& P Covidien Medical Equipment N/A Spin - off Ireland 2006 - 01 2007 Y TE Connectivity Electronic Equipment N/A Spin - off Switzerl and 2006 - 01 2007 Y Freescale Semiconductor Electronic Equipment N/A Pure Bermuda 2006 - 09 2006 Patch International Petroleum and Natural Gas Damascus Energy M&A Canada 2006 - 12 2006 Argonaut Group Insurance PXRE Group M&A Bermuda 2007 - 03 2007 Star Maritime Acq. Transportati on Star Bulk Carriers M&A Marshall Islands 20 07 - 03 2007 Western Goldfields Precious Metals N/A Pure Canada 2007 - 05 2007 Ascend Acquisition Business Services e.Pak Res. (S) Pte. M&A Bermuda 2007 - 07 2007 Vantage Energy Services Petroleum and Natural Gas Offshore Group Inv. M&A Cayman Islands 2007 - 08 2007 Lincoln Gold Non - metallic and Industrial Metal Mining N/A Pure Canada 2007 - 09 2007 Energy Infrastructure Acquisition Transportati on N/A Pure Marshall Is. 2008 - 06 Failed InterAmerican Acquisition Finance Trading N/A Pure BVI 2008 - 07 2008 Arcade Acquisition Transportati on Conbulk M&A Marshall Is. 2008 - 09 2008 Hungarian Tel & Cable Communica tion Invitel Ho ldings M&A Austria 2008 - 11 2007 Alpha Security Finance Trading Soya China Pte. M&A Bermuda 2008 - 12 Failed Invitel Holdings A/S Communica tion N/A Pure Denmar k 2009 - 02 2009 Ideation Acquisition Finance Trading SearchMe M&A Cayman Islands 2009 - 04 2009 62 Table A1 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann . F inish S& P 2020 ChinaCap Acquirco Consumer Goods Windrace Int'l M&A British Virgin Islands 2009 - 05 2009 Altisource Portfolio Solutions Business Services N/A Spin - off Luxemb ourg 2009 - 07 2009 Tim Hortons restaurants, hotels, motels N/A Pure Canada 2009 - 0 9 2009 Delphi Automotive Automobile and Trucks N/A Pure Jersey 2009 - 10 2011 Y ENSCO Int'l Petroleum and Natural Gas N/A Pure UK 2009 - 11 2009 Y Plastinum Polymer Technologies Rubber and Plastic Products N/A Pure Netherla nds 2010 - 06 Failed Styron Chemi cals N/A Pure Luxemb ourg 2010 - 06 2010 Valeant Pharmaceuticals Pharmaceuti cal Products Biovail M&A Canada 2010 - 06 2010 Alkermes Pharmaceuti cal Products Elan M&A Ireland 2011 - 05 2011 Jazz Pharmaceuticals Pharmaceuti cal Products Azur Pharma M&A Ireland 2011 - 09 2011 Tronox Chemicals Exxaro Res. M&A Australia 2011 - 09 2011 AON Insurance N/A Pure UK 2012 - 01 2012 Y Rowan Companies Petroleum and Natural Gas N/A Pure UK 2012 - 02 2012 Y Pentair Machinery Tyco Internatio nal M&A Switzerl and 2012 - 03 2012 Y Stratasys Computers Objet M&A Israel 2012 - 04 2012 Tower Group Insurance Canopius Holdings Bermuda M&A Bermuda 2 012 - 04 2012 Eaton Electrical Equipment Cooper Industries M&A Ireland 2012 - 05 2012 63 Table A1 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann . F inish S& P Axalta Coating Systems Chemicals The Carlyle Group M&A Bermuda 2013 - 02 2013 Liberty Global Communica tion Virgin Media M&A UK 2013 - 02 2013 Actavis Pharmaceuti cal Prod ucts Warner Chilcott M&A Ireland 2013 - 05 2013 Y Omnicom Group Business Services Publicis Groupe UK M&A UK 2013 - 07 Failed Y Perrigo Pharmaceuti cal Products Elan, Blisfont M&A Ireland 2013 - 07 2013 Y Applied Materials Electronic Equipment Tokyo Electron Pu re Netherla nds 2013 - 09 Failed Y Allegion Constructio n Materials N/A Spin - off Ireland 2013 - 11 2013 Y Endo Health Solu tions Pharmaceuti cal Products Paladin Labs M&A Ireland 2013 - 11 2013 Y Multi Packaging Solutions Int'l Shipping Containers AGI - Shorewoo d Group M&A Bermuda 2013 - 11 2014 Horizon Pharma Pharmaceuti cal Products Vidara Therapeut ics Int'l M&A Ireland 2014 - 03 2014 Chiquita Brands Int'l Agriculture Fyffes Plc M&A Ireland 2014 - 04 2014 Pfizer Pharmaceuti cal Products Astrozeni ca M&A UK 2014 - 04 Failed Y Questcor Pharmaceuticals Pharmaceuti cal Products Mallinckr odt M&A Ireland 2014 - 04 2014 Theravance Pharmaceu ti cal Products N/A Pure Cayman Islands 2014 - 04 2014 Mondelez International Food Products D.E. Master Blenders M&A Netherla nds 2014 - 05 2014 C&J Energy Services Petroleum and Natural Gas Nabors Industries M&A Bermuda 2014 - 06 2014 Medtronic Medical Equi pment Covidien M&A Ireland 2014 - 06 2014 Y AbbVie Pharmaceuti cal Products Shire M&A UK 2014 - 07 Failed Y 64 Table A1 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann . F inish S& P Mylan Pharmaceuti cal Products Abbott Laboratori es Non - US Assets M&A Netherla nds 2014 - 07 2014 Y Burger King Worldwide restaurants, hotels, motels Tim Hortons M&A Canada 2014 - 08 2014 Auxilium Pharmaceuticals Pharmaceuti cal Products Endo Internatio nal M&A Ireland 2014 - 1 0 2014 Steris Medical Equipment Synergy Health M&A UK 2014 - 10 2014 Wright Medical Group Medical Equipment Tornier M&A Netherla nds 2014 - 10 2014 Paragon Offshore Petroleum and Natural Gas N/A spin - off UK 2014 - 11 2015 65 APPENDIX B . Corporate Inversion s Used for Empirical Analysis . This appendix lists the inverting firms used for empirical analysis in this paper. The firms listed in Appendix A are further screened by comparing each firm with their SEC Edgar filings based on the following criteria and u sing the ones that qualify: (ii) the inverting company was a publicly traded U.S. company before inversion; (iii) stocks of the surviving company are still publicly traded in the United States; (iv) the transa ction was closed/completed by December 31, 2015. We are left with 63 inversions after screening. Table A2: Corporate Inversions Used for Empirical Analysis U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann. F inish S&P Helen of Troy Consumer Goods N/A Pure Bermuda 1993 - 12 1994 Loral Communic ation N/A Pure Bermuda 1996 - 01 1996 Triton Energy Petroleum and Natural Gas N/A Pure Cayman Islands 1996 - 02 1996 Chicago Bridge & Iron Constructi on N/A Pure Netherla nds 1997 - 03 1997 Tyco International Electronic Equipment ADT M&A Bermuda 1997 - 03 1997 Y Xoma Pharmaceu tical Products N/A Pure Bermuda 1998 - 11 1999 Gold Reserve Precious Metals N/A Pure Canada 1999 - 02 1999 Fruit of the Loom Textiles N/A Pure Cayman Islands 1999 - 03 1999 Transocean Offshore Petroleum and Natural Gas N/A Pure Cayman Islands 1999 - 05 1999 Y PXRE Insurance N/A Pure Bermuda 1999 - 10 1999 White Mountains Insurance Group Insurance N/A Pure Bermuda 1999 - 10 1999 Trenwick Group Insurance LaSalle Re Ho ldings M&A Bermuda 1999 - 12 1999 66 Table A2 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann. F inish S&P Everest Reinsurance Holdings Insurance N/A Pure Bermuda 1999 - 09 2000 Tycom Electronic Equipmen t N/A Spin - off Bermuda 2000 - 01 2000 Applied Power Machinery N/A Pure Bermuda 2000 - 07 2000 Seagate Technology Computers N/A Pure Cayman Islands 2000 - 08 2000 Y Arch Capital Group Insurance N/A Pure Bermuda 2000 - 09 2000 Foster Wheeler Constructi on N/A Pure Bermuda 2000 - 12 2001 Global Marine Petroleum and Natural Gas Santa Fe Int'l M&A Cayman Islands 2001 - 09 2001 Ingersoll - Rand Machinery N/A Pure Bermuda 2001 - 10 2001 Y Noble Drilling Petroleum and Natural Gas N/A Pure Cayman Islands 2002 - 01 2002 Y Herbalife Wholesale N/A Pure Cayman Islands 2002 - 02 2002 Nabors Industries Petroleum and Natural Gas N/A Pure Bermuda 2002 - 06 2002 Weatherford Int'l Petroleum and Natural Gas N/A Pure Bermuda 2002 - 06 2002 Lazard Finance Trading N/A Pure Bermuda 2005 - 12 2005 Freescale Semiconductor Electronic Equipment N/A Pure Bermuda 2006 - 09 2006 Covidien Medical Equipment N/A Spin - off Ireland 2006 - 01 2007 Y TE Connectivity Electronic Equipment N/A Spin - off Switzerl and 2006 - 01 2007 Y Argonaut Group Insurance PX RE Group M&A Bermuda 2007 - 03 2007 67 Table A2 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann. F inish S&P Star Maritime Acq. Transporta tion Star Bulk Carriers M&A Marshall Islands 2007 - 03 2007 Western Goldfields Precious Metals N/A Pure Canada 2007 - 05 2007 Altisource Portfolio Solutions Business Services N/A Spin - off Luxemb ourg 2009 - 07 2009 Tim Hortons restaurants , hotels, motels N/A Pure Canada 2009 - 09 2009 ENSCO Int'l Petroleum an d Natural Gas N/A Pure UK 2009 - 11 2009 Y Styron Chemicals N/A Pure Luxemb ourg 2010 - 06 2010 Valeant Pharmaceutica Pharmaceu tical Products Biovail M&A Canada 2010 - 06 2010 Delphi Automotive Automobil e and Trucks N/A Pure Jersey 2009 - 10 2011 Y Al kermes Pharmaceu tical Products Elan M&A Ireland 2011 - 05 2011 Jazz Pharmaceutica ls Pharmaceu tical Products Azur Pharma M&A Ireland 2011 - 09 2011 Tronox Chemicals Exxaro Res. M&A Australia 2011 - 09 2011 AON Insurance N/A Pure UK 2012 - 01 2012 Y Rowan Com panies Petroleum and Natural Gas N/A Pure UK 2012 - 02 2012 Y Pentair Machinery Tyco Internatio nal M&A Switzerl and 2012 - 03 2012 Y Stratasys Computers Objet M&A Israel 2012 - 04 2012 Tower Group Insurance Canopius Holdings Bermuda M&A Bermuda 2012 - 04 2012 68 Table A2 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann. F inish S&P Eaton Electrical Equipment Cooper Industries M&A Ireland 2012 - 05 2012 Axalta Coating Systems Chemicals The Carlyle Gro up M&A Bermuda 2013 - 02 2013 Liberty Global Communic ation Virgin Media M&A UK 2013 - 02 2013 Actavis Pharmaceu tical Products Warner Chilcott M&A Ireland 2013 - 05 2013 Y Perrigo Pharmaceu tical Products Elan, Blisfont M&A Ireland 2013 - 07 2013 Y Allegion Co nstructi on Materials N/A Spin - off Ireland 2013 - 11 2013 Y Endo Health Solutions Pharmaceu tical Products Paladin Labs M&A Ireland 2013 - 11 2013 Y Horizon Pharma Pharmaceu tical Products Vidara Therapeut ics Int'l M&A Ireland 2014 - 03 2014 Questcor Pharmaceut ica ls Pharmaceu tical Products Mallinckr odt M&A Ireland 2014 - 04 2014 Theravance Pharmaceu tical Products N/A Pure Cayman Islands 2014 - 04 2014 Mondelez International Food Products D.E. Master Blenders M&A Netherla nds 2014 - 05 2014 C&J Energy Services Pet roleum and Natural Gas Nabors Industries M&A Bermuda 2014 - 06 2014 Medtronic Medical Equipment Covidien M&A Ireland 2014 - 06 2014 Y Mylan Pharmaceu tical Products Abbott Laboratori es Non - US Assets M&A Netherla nds 2014 - 07 2014 Y 69 Table A2 U.S. Target Name Industry Foreign Acquirer Name Type Dest . Ann. F inish S&P Burger King Worldwide restaurants , hotels, motels Tim Hortons M&A Canada 2014 - 08 2014 Steris Medical Equipment Synergy Health M&A UK 2014 - 10 2014 Wright Medical Group Medical Equipment Tornier M&A Netherla nds 2014 - 10 2014 Paragon Offshore Petroleum and Natural Gas N/A spin - off UK 2014 - 11 2015 70 APPENDIX C . Variable Definitions Cash Holdings : The ratio of cash and short - term investments to the book value of total assets. Capital Expenditure : The ratio of capital expenditures to the book value of total assets. D it : A dummy variable that identifies the incorporation status of company i at time t . If company i is incorporated in the United States at time t , D it equals 0. If the company reincorporates overseas and thus becomes foreign - incorporated at time t , D it equals 1. Dividend : The ratio of total dividends paid to the book value of total assets. Leverage : Leverage ratio, defined to be the ratio of the b ook value of total debt to the book value of total assets. R&D Expenditure : The ratio of R&D expenditures to the book value of total assets. ROA : Return on assets, defined to be the ratio of earnings to the book value of total assets. SG&A : The ratio of se lling, general, and administrative expenses to the book value of total assets. Tax Rate : The ratio of cash tax paid to pre - tax income before special items. Extreme values are truncated at zero and one. Total Assets : The book value of total assets (in milli ons of dollars). 71 BIBLIOGRAPHY 72 BIBLIOGRAPHY Abadie, A., D. Drukker, J. L. Herr, and G. W. Imbens. 2004. Implementing Matching Estimators for Average Treatment Effect s in Stata. The Stata Journal 4:290 - 311. Atlshuler, R., T. S. Newlon, and W. C. Rando lph. 1995. Do Repatriation Taxes Matter? Evidence From the Tax Returns of U.S. Multinationals. In M. Feldstein, J. R. Hines, Jr., and R. G. Hubbard (eds.), The Effect s of Taxation on Multinational Corporations , pp. 253 - 272. Chicago, IL: University of Chica go Press. Babkin, A., B. Glover, and O. Levine. 2016. Are Corporate Inversions Good for Shareholders? Working Paper . Cloyd, C. B., L. F. Mills, and C. D.Weaver. 2003. Firm Valuation Effect s of the Expatriation of U.S. Corporations to TaxHaven Countries. The Journal of the American Taxation Association: Supplement 2003 25:87 - 109. Cortes, F., A. Gomes, and R. Gopalan. 2015. Corporate Inversions: A Case of Having the Cake and Eating I t Too? Working Paper . Daines, R. 2001. Does Delaware law Improve Firm Val ue? Journal of Financial Economics 62:525 - 558. Desai, M. A., C. F. Foley, and J. R. Hines, Jr. 2001. Repatriation Taxes and Dividend Distortions. National Tax Journal 54:829 - 851. Desai, M. A., and J. R. Hines, Jr. 2002. Expectations and Expatriations: Tr acing the Causes and Consequences of Corporate Inversions. National Tax Journal 55:409 - 440. Dharmapala, D., and J. R. Hines, Jr. 2009. Which countries become tax havens? Journal of Public Economics 93:1058 - 1068. Doidge, C., G. A. Karolyi, and R. M. Stulz . 2004. Why Are Foreign Firms Listed in the U.S. Worth More? Journal of Financial Economics 71:205 - 238. Doidge, C., G. A. Karolyi, and R. M. Stulz. 2007. Why Do Countries Matter So Much for Corporate Governance? Journal of Financial Economics 86:1 - 39. Fa ma, E. F., and K. R. French. 1997. Industry Costs of Equity. Journal of Financial Economics 43:153 - 193. Gerbardt, W. R., C. M. C. Lee, and B. Swaminathan. 2001. Toward an Implied Cost of Capital. Journal of Accounting Research 39:135 - 176. 73 Hines, J. R., J r., and R. G. Hubbard. 1990. Coming Ho me to America: Dividend Repatri ations by U.S. Multinationals. In A. Razin and J. Slemrod (eds.), Taxation in the Global Economy , pp. 161 - 200. Chicago, IL: Univeristy of Chicago Press. Hines, J. R., Jr., and E. M. Rice . 1994. Fiscal Paradise: Foreign Tax Havens and American Business. The Quarterly Journal of Economics 109:149 - 182. Imbens, G., and J. Wooldridge. 2007. What s New in Econometrics? NBER Lecture Notes . Licht, A. N. 2003. Cross - Listing and Corporate Governa nce: Bonding or Avoiding? Chicago Journal of International Law 4:141 - 163. Pastor, L., M. Sinha, and B. Swaminathan. 2008. Estimating the Intertemporal RiskReturn Tradeo ff Using the Implied Cost of Capital. The Journal of Finance 63:2859 - 2897. Seida, J. A Invers ion Transactions. Working Paper . Inversions. Tax Notes pp. 1146 - 1150. Shnitser, N. 20 10. A Free Pass for Foreign Firms? An Assessment of SEC and Private Enforcement Against Foreign Issuers. Yale Law Journal 119:1638 - 1701. Siegel, J. 2005. Can Foreign Firms Bond Themselves Effect ively by Renting U.S. Securities Laws? Journal of Financial E conomics 75:319 - 359. Talley, E. L. 2015. Corporate Inversions and the Unbundling o f Regulatory Competition. Virginia Law Review 101:1649 - 1751.