The Effect of Foreign Cash Holdings on Internal Capital ... · PDF fileThe Effect of Foreign...
-
Upload
nguyenhuong -
Category
Documents
-
view
218 -
download
1
Transcript of The Effect of Foreign Cash Holdings on Internal Capital ... · PDF fileThe Effect of Foreign...
The Effect of Foreign Cash Holdings on Internal Capital Markets and Firm Financing
Lisa De Simone [email protected]
Rebecca Lester
July 2017
Abstract
Prior literature demonstrates that firms should first use internal capital before accessing costly external finance. However, the U.S. tax system imposes an internal capital market friction by taxing active foreign earnings repatriated from a U.S. multinational corporation’s (MNC) foreign operations. This repatriation tax motivates firms to retain cash offshore (Foley et al, 2007) and suggests that U.S. MNCs with large amounts of tax-induced foreign cash may be unable to inexpensively access their foreign capital. We test whether and to what extent U.S. MNCs use domestic financing rather than incur the repatriation tax to meet domestic cash needs. Consistent with prior literature on internal capital markets and the pecking order, we first show an overall negative relation between a firm’s foreign cash holdings and domestic liabilities – U.S. MNCs with more foreign cash on average borrow less than other MNCs. We then show that this negative relation is attenuated for U.S. MNCs that require domestic cash for the specific purpose of financing share repurchases. We also observe this result for MNCs with mobile income, as measured based on R&D activities. Additional tests suggest that these results are not driven by a differential cost of debt and that increased domestic liabilities are incremental worldwide liabilities, not offset by lower foreign borrowing. This paper adds to the literature on internal capital markets by showing how U.S. tax frictions impede the domestic use of internal capital held by foreign subsidiaries, thereby inducing firms to access domestic financing sources.
______________________
We thank Joshua Anderson, Cristi Gleason, James Fetzer, Michelle Hanlon, Shane Heitzman, Jim Hines, Zawadi Lemayian, Jon Mandrano, Kevin Markle, Peter Merrill, Patricia Naranjo, Joe Piotroski, Nemit Shroff, Rodrigo Verdi, and Bill Zeile, as well as seminar participants at the 2017 George Washington University Cherry Blossoms Conference, MIT, and the University of Iowa for helpful comments on this paper. We gratefully acknowledge financial support from the Stanford Graduate School of Business and the International Tax Policy Forum. The statistical analysis of firm-level data on U.S. multinational companies was conducted at the Bureau of Economic Analysis (BEA), Department of Commerce under arrangements that maintain legal confidentiality requirements. The views expressed in this study are those of the authors and do not reflect official positions of the U.S. Department of Commerce.
1
1. Introduction
U.S. multinational corporations (MNCs) hold over $2.5 trillion in foreign earnings
overseas to avoid U.S. taxation (Barthold, 2016), approximately one half of which is estimated to
be held in cash and other financial assets (Blouin, Krull, and Robinson, 2016). One explanation
for the large foreign cash holdings is that the U.S. tax system of deferral, which defers the U.S.
taxation of active foreign subsidiary earnings until these earnings are repatriated to the U.S.,
motivates firms to retain cash in foreign subsidiaries (Foley, Hartzell, Titman, and Twite, 2007;
De Simone, Piotroski, and Tomy 2017). Prior literature shows that the large amounts of locked-
out foreign cash affect firm investment (Faulkender and Petersen, 2012; Hanlon, Lester, and Verdi,
2015; Edwards, Kravet, and Wilson, 2016) and dividend policies (Nessa, 2017). This paper
examines how this U.S. tax policy affects firm financing. Specifically, firms with significant
offshore cash attributable to repatriation tax liabilities (tax-induced foreign cash) may be
constrained domestically if the amount of cash generated in the U.S. is insufficient to meet
domestic operational, investment, and payout needs. We study whether and to what extent U.S.
MNCs with significant tax-induced foreign cash use U.S. debt financing to meet these domestic
cash needs.
Theory and prior empirical work show that using internal capital to fund operations within
a firm is generally less costly than external capital (Myers, 1984; Myers and Majluf, 1984; Gertner,
Scharfstein, and Stein, 1994; Stein, 1997; Shyam-Sunders and Myers, 1999; Leary and Roberts,
2010), such that value-maximizing managers will select internal capital first to fund domestic cash
needs. This literature shows a negative association between firm cash holdings and external debt
financing. Conditional on accessing external financing, prior literature on firm capital structure
and the tax benefits of debt further motivates a negative relation between external debt financing
2
and the high tax-induced foreign cash held by firms with relatively low global effective tax rates:
because these firms have lower overall worldwide tax burdens, the value of interest deductions is
lower (Modigliani and Miller, 1958, 1963; Miller, 1977; DeAngelo and Masulis, 1980; Graham,
1996a). Despite this expected negative relation as motivated by these prior studies, firms with a
significant amount of cash on their balance sheets are tapping the external debt market rather than
using internal cash holdings (Linebaugh, 2012). For example, eBay (with 90 percent of cash
offshore) issued $3 billion in debt in July 2012 (Mead and Kucera, 2012). Apple (with 79 percent
of cash offshore in 2013) satisfied investor demands for return of capital by borrowing $17 billion
in 2013 (Lattman and Eavis, 2013). Microsoft (with 85 percent of cash offshore for its fiscal year
ending June 2011) sold $4.75 billion in bonds in September 2010 to finance dividends and share
repurchases and announced a new borrowing of $17 billion as recently as January 2017
(Abramowicz, 2017).
Frictions created by repatriation taxes and other U.S. tax rules limiting the use of foreign
assets in domestic operations may explain the use of external debt financing by Microsoft, eBay,
Apple, and other U.S. MNCs.1 In addition to imposing a repatriation tax liability on foreign
dividend payments to the U.S. parent, U.S. tax rules also limit a firm’s ability to loan funds from
foreign operations to the U.S. by recasting many of these loans as deemed dividends. For U.S.
MNCs with trapped foreign cash that require domestic liquidity to fund domestic operations, the
internal cost of capital will reflect repatriation taxes and therefore could be higher than the firm’s
external cost of capital. For these firms, the negative relation between tax-induced foreign cash
1Statements by corporate management validate that all three of these borrowings were motivated by U.S. tax policies and repatriation tax costs. For example, eBay’s CFO stated that the domestic debt issuances were intended to “get the wonderful benefit of an extremely low tax rate, but also get our cash geographically where we would like it to be to enable us to acquire and redistribute cash effectively” (Mead and Kucera, 2012).
3
and external domestic financing may be attenuated if it is more efficient for them to use external
debt financing rather than incur the U.S. repatriation tax cost.
Indeed, a sample of firms surveyed state that they avoid the repatriation tax liability by
raising capital in the U.S. debt market (Graham, Hanlon, and Shevlin, 2010). Furthermore, prior
work finds that the probability of a U.S. MNC issuing public debt is positively related to the
amount of foreign subsidiary earnings designated as permanently reinvested, a financial
accounting assertion that reduces deferred taxes and increases reported net income (Albring, 2006;
Petzel and Salbador, 2016).2 We directly test if the magnitude of tax-induced foreign cash is
positively related to total (public and private, internal and external) domestic liabilities using
jurisdiction-specific data from the Bureau of Economic Analysis (BEA). These tests allow us to i)
focus on cash tax incentives rather than financial reporting decisions that can be influenced by
earnings management, ii) examine domestic (not total) liabilities, including external bank debt,
publicly-held debt, and trade financing,3 iii) study the characteristics of firms most likely to engage
in this strategy (i.e., those with domestic demand for cash), and iv) quantify the extent to which
firms use this strategy to minimize U.S. tax liabilities.
However, prior literature suggests that we may not observe an attenuated association
between tax-induced foreign cash and external domestic financing for U.S. MNCs that need
domestic cash. First, although public press releases and news articles discuss that several U.S.
MNCs with large foreign cash balances have borrowed domestically to fund domestic cash needs,
2Albring (2006) and Petzel and Salbador (2016) estimate the probability of a firm issuing public debt as a function of permanently reinvested foreign earnings disclosed in the financial statements, finding a positive relation between permanently reinvested earnings and the likelihood of a public debt issuance. Our study is complementary in that we directly test the relation between tax-induced foreign cash (as opposed to a financial accounting assertion about foreign earnings) and total domestic liabilities to measure the magnitude of this activity for a large sample of MNCs. 3 Our current measure of domestic liabilities includes external financing and trade credit as well as intercompany loans. In future tests, we intend to examine each component separately using data from the BEA survey years for which these data are available (1999, 2004, and 2009).
4
it is unknown if this empirical relation holds for a large set of firms. Second, Desai, Foley, and
Hines (2007) show that – despite repatriation taxes – approximately 30 percent of U.S. MNCs’
foreign affiliates continue to pay dividends to the U.S. parent for U.S. investment and repayment
purposes. Third, firms have alternative methods for accessing the capital of foreign subsidiaries.
For example, Altshuler and Grubert (2003) find that firms circumvent the repatriation tax via
complex ownership, borrowing, and investment strategies with foreign affiliates that result in
relatively low or no additional repatriation tax. Fourth, firms can avoid deemed dividend treatment
on intercompany loans from foreign subsidiaries by structuring a short-term revolving loan
program in which the U.S. parent effectively borrows from a number of different subsidiaries.4
Finally, firms can trim domestic operations and investment to save domestic cash rather than pay
for external capital.5 Therefore, the relation between repatriation tax costs and firm financing is an
empirical question.
To test our hypothesis, we use data for a sample of U.S. multinational C corporations from
Compustat and the BEA over the period 1999 through 2012. The BEA data allow us to observe
detailed financial statement information of U.S. parent companies and their foreign affiliates,
which is otherwise not publicly available on an unconsolidated basis. We first follow Hanlon et al.
(2015) and use the BEA data on foreign cash holdings to estimate tax-induced foreign cash, which
4Sec. 956 of the Internal Revenue Code limits U.S. entities from borrowing from foreign subsidiaries by recasting these loans as deemed dividend payments. However, tax rules permit an exception to this rule by allowing foreign subsidiaries to make short-term loans to the U.S. parent. Companies can exploit the tax rules so that these loans are effectively long-term borrowings if the U.S. entity settles the loan with one foreign subsidiary and takes out a loan with another foreign subsidiary on the same day. HP employed such a “revolving loan program” between subsidiaries in Belgium and the Cayman Island (Permanent Subcommittee, 2013). See further discussion in Sec. 2. 5 Two concurrent working papers study the relation between repatriation taxes and the cost of debt. Blaylock, Downes, Mathis, and White (2016) find a positive association between the cost of debt and an interaction of a repatriation tax indicator and foreign earnings. Ma, Stice, and Wang (2016) find a positive association between interest spreads and the repatriation tax cost. Both papers interpret the results as evidence that banks will charge firms with high repatriation tax liabilities higher fees to borrow. Neither of these papers tests if firms choose to actually borrow more despite this higher cost.
5
is the amount of cash held offshore attributable to the firm’s repatriation tax liability. We then
regress measures of total domestic liabilities (also from the BEA) on a firm’s estimated tax-
induced foreign cash, controlling for other determinants of corporate financing decisions such as
size, investment spending, fixed assets, growth opportunities, the cost of debt capital, and firm
financial performance. Consistent with prior literature, we observe that tax-induced foreign cash
is negatively related to the level of firms’ total and domestic liabilities – firms with more foreign
cash borrow less.
Next, we test whether this negative relation between tax-induced foreign cash and domestic
borrowing is attenuated based on a firm’s demand for domestic cash to fund shareholder payouts,
domestic operations, and domestic investments. We find that the negative association between
domestic liabilities and tax-induced foreign cash is significantly weaker for firms with payout
obligations, which must be remitted out of the U.S. parent entity.6 Specifically, firms engaging in
shareholder repurchases have a less negative relation between domestic liabilities and tax-induced
foreign cash than other firms. These firms have 2.6 to 3.2 percent higher domestic liabilities
relative to other firms with tax-induced foreign cash. We observe little attenuation within
dividend-paying firms. Further, we find no differential association between domestic liabilities
and tax-induced foreign cash for firms with a high proportion of domestic employees or
compensation expense, and we find only limited evidence of a differential association for firms
under-invested domestically (measured following Biddle, Hilary, and Verdi (2009) and extended
in Harford, Wang, and Zhang (2017)). Collectively, these results suggest that some firms with tax-
6 Dividends are distributed (and repurchases are made) by the entity in which shareholders have a direct ownership interest. Although it is possible for a U.S. MNC to have a separately listed foreign subsidiary in which some shareholders own a direct interest, in our study we focus on the behavior of U.S.-incorporated firms with a publicly-traded U.S. parent.
6
induced foreign cash balances use domestic financing to fund shareholder repurchases, but not to
meet domestic capital needs for operational or investment purposes.
We conduct several additional tests to validate our results and provide additional insights.
First, we find that the cost of debt financing does not affect the relation between tax-induced
foreign cash and domestic liabilities, which provides evidence that our main results are not driven
by firms that simply have a lower cost of debt. Second, we examine whether the higher levels of
domestic liabilities exhibited by firms distributing cash to shareholders constitute either
incremental total liabilities or substituted foreign liabilities. For firms with the attenuated relation
(i.e., firms repurchasing shares), we observe higher total worldwide (consolidated) liabilities but
no difference in the level of foreign liabilities, providing evidence consistent with these effects
resulting in incremental total financing. Finally, building off of Pinkowitz, Stulz, and Williamson
(2016), we show that the negative relation between tax-induced foreign cash and domestic
liabilities is also attenuated among firms engaging in higher levels of R&D activities. Thus, it
appears that the decision by firms to accrue resources to engage in R&D activities offshore as
documented in Pinkowitz et al. (2016) induces internal capital market constraints that are
associated with additional external domestic financing.
Our findings contribute to the literature by documenting how the U.S tax policy of
worldwide tax with deferral affects firms’ internal capital markets and debt financing decisions.
Specifically, we show the extent to which U.S. tax frictions impede the domestic use of internal
foreign capital, thereby inducing firms to access external financing. We also provide empirical
evidence on the characteristics of firms that engage in this type of tax planning by showing that
the negative relation between tax-induced foreign cash and domestic liabilities is attenuated among
firms that repurchase shares, as well as high-intangibles firms. Finally, this paper contributes to
7
the literature that documents how the repatriation tax affects MNCs (Foley et al., 2007; Hanlon et
al., 2015; Edwards et al., 2016; De Simone et al., 2017c; Nessa, 2017). It is important to
understand the effects on firm financing because, in addition to circumventing repatriation taxes,
U.S. MNCs with external debt also benefit from U.S. interest deductions, which further reduce
U.S. taxable income and cash taxes due. If repatriation taxes and trapped foreign cash are
associated with increased U.S. debt financing, the U.S. system of deferral may have more
significant tax revenue implications for policy makers than originally estimated.
2. Background and Hypothesis Development
2.1. Overview of Tax Rules for U.S. Multinationals
The United States taxes the worldwide income of companies incorporated in the U.S.
However, U.S. tax law provides an exception, which permits deferral of U.S. taxation of a U.S.
parent’s foreign subsidiaries’ earnings until the earnings are repatriated back to the U.S. parent as
a dividend.7 These foreign operating earnings are subject to foreign taxation, if any, in the
jurisdiction where they are earned or sourced, and foreign income taxes paid can be credited
against U.S. taxes upon repatriation. As a result, a U.S. MNC will effectively pay worldwide tax
at the higher of the U.S. or local country rate. Since 1986, most other countries have lowered their
corporate tax rates while the U.S. rate has remained relatively constant; given that the U.S. now
has one of the highest statutory corporate tax rates in the world, most U.S. firms remit additional
U.S. tax upon repatriation.
Related U.S. tax rules prevent firms from circumventing the repatriation tax by restricting
long-term lending to domestic entities by foreign subsidiaries. Specifically, Internal Revenue
Code Section 956 prohibits the use of intercompany loans to deploy foreign cash in domestic
7 Material passive income is generally taxed immediately under Subpart F of the U.S. Internal Revenue Code.
8
operations by recasting long-term borrowings as deemed dividends, thereby subjecting these loans
to repatriation taxes.8 However, short term loans made by a foreign subsidiary to a related U.S.
entity are not subject to deemed dividend treatment if they are repaid within 30 days and if all
loans made by the foreign subsidiary throughout the year are outstanding for less than 60 days
total.9
2.2 Prior Literature and Hypothesis
The prior literature on internal capital markets suggests that firms with a large amount of
cash should use this cash as the primary source to finance firm operations, investment, and payout
policies. Ignoring taxes, the cost of using internal capital is likely lower than the external cost of
capital due to strong internal monitoring incentives that decrease internal information asymmetry
(Williamson, 1975; Alchian, 1969; Gertner et al., 1994; Stein, 1997).10 Managers should therefore
prefer internal capital when available because internal capital is predicted to maximize investment
returns and, in turn, maximize firm value. This literature would suggest that large firm cash
holdings (whether domestic or foreign) should be negatively related to external financing.
8 Specifically, IRC Section 956 recasts certain uses of foreign subsidiary cash as deemed dividend payments to the U.S. parent, which in turn triggers repatriation taxes. These uses include i) purchase of tangible property in the U.S.; ii) purchase of a material share of a domestic corporation; iii) an obligation of a U.S. person; or iv) the right to the use of U.S. intangibles developed by a foreign subsidiary. Treasury Regulations pertaining to IRC Section 956 elaborate that foreign assets used as collateral or foreign guarantees on U.S. obligations are also subject to repatriation taxes. 9 The rules apply to foreign subsidiaries that are Controlled Foreign Corporations (“CFCs”). Briefly, a CFC is any foreign corporation more than 50 percent owned by U.S. shareholders who at least own ten percent each. In response to the recent financial crisis, the 30/60 day periods were extended to 60/180 days for fiscal years ending after October 3, 2008 and before January 1, 2011 to help firms suffering from domestic liquidity problems. 10 Prior literature predicts that internal capital markets benefit from higher quality – and likely greater – information, reducing information asymmetry within the firm relative to information asymmetry between the firm and external capital providers. This relatively lower level of information asymmetry should translate into less costly internal capital because external capital providers will price protect. Indeed, models of internal and external capital generally include costs of external financing but minimal or no costs of internal financing, and less internal “wealth” (retained earnings) translates into greater deadweight costs of external financing (Stein, 2002). Later theoretical and empirical work examines the “dark side” of internal capital markets (Rajan, Servaes, and Zingales, 2000; Scharfstein and Stein, 2000; Lewellen, 1971) in that underperforming divisions may be inefficiently subsidized through an internal capital market mechanism. Importantly, however, these studies do not challenge the assertion that internal capital markets benefit from better monitoring, which in turn reduces the relative amount of information asymmetry within the firm as compared to between the firm and its external capital providers.
9
Furthermore, the literature on firm capital structure and the tax benefits of debt also
motivates a negative association between tax-induced foreign cash holdings and debt financing.
Specifically, external debt can add value to the firm by generating tax-reducing, and therefore
income-increasing, interest deductions (Modigliani and Miller, 1958, 1963; Miller, 1977;
DeAngelo and Masulis, 1980). To empirically estimate firm-specific tax benefits of debt, Shevlin
(1987, 1990) and Graham (1996a, 1996b) develop a firm-specific marginal tax rate (MTR),
defined as the present value of current and future taxes owed on an extra dollar of income earned
today (Scholes and Wolfson, 1992).11
The calculation of a firm-specific MTR incorporates U.S. tax rules, such as the U.S.
statutory tax rate and tax loss offset periods, but generally does not reflect foreign tax rules
applicable to MNCs. Consequently, a U.S. multinational’s MTR is often measured with error, due
to researchers’ inability to observe the proportion of firm income that is earned in other countries
and taxed under local country tax rules. To address this measurement error, Faulkender and Smith
(2016) use BEA data to construct a worldwide weighted-average firm-specific tax rate, calculated
as the sum of the proportion of firm income earned in each foreign affiliate’s country, multiplied
by the corresponding country’s statutory tax rate. They show that firm debt is increasing in the
estimated worldwide weighted-average tax rate.12
We focus on the relation between tax-induced foreign cash and firm borrowing. Following
Hanlon et al. (2015), we first estimate tax-induced foreign cash using a measure of a firm’s
11 Consistent with predictions, Graham (1996a, 2000) and Graham, Lemmon, and Schallheim (1998) find a positive association between simulated MTRs and corporate debt, and Mackie-Mason (1990) and Dhaliwal, Trezevant, and Wang (1992) find that firms expected to have lower MTRs use less debt in the presence of other non-debt tax shields. For a review of this literature, see Graham (2003). Because empirical research in this area generally documents that firms on average use less leverage than the expected value-maximizing level, recent studies focus on improving the measurement of ex ante distress costs (Molina, 2005), overlooked non-debt tax shields such as deductions from stock options (Graham, Lang and Shackelford, 2004), and MTRs (Graham and Kim, 2009; Blouin, Core, and Guay, 2010). 12 This result also holds when measuring debt using only domestic borrowings. However, the paper does not test the relation between domestic debt and tax-induced foreign cash, which is the focus of our paper.
10
repatriation tax costs. By construction, a firm’s repatriation tax cost is negatively correlated with
a firm’s MTR: firms with high estimated repatriation taxes enjoy low foreign tax rates, defer more
U.S. tax payments on foreign earnings, and therefore have relatively low global effective tax rates
relative to firms with low estimated repatriation taxes. Consequently, prior literature on MTRs also
motivates a negative relation between tax-induced foreign cash holdings (which is positively
correlated with a firm’s repatriation tax cost liability) and a firm’s external debt.
Therefore, as validation of the prior relation documented in the MTR empirical literature
and consistent with well-functioning internal capital markets, we predict the following:
H1: Tax-induced foreign cash is negatively related to the level of firm liabilities.
This prediction assumes that firms with tax-induced foreign cash have an efficiently
operating internal capital market in which capital can move freely within the firm. However,
incorporating repatriation taxes into an internal capital market framework suggests that some firms
may be geographically constrained, thereby leading them to consider external financing. The U.S.
system of deferral and the associated repatriation tax cost motivates firms to shift income to foreign
jurisdictions to benefit from lower foreign tax rates (Grubert, 1998; Klassen and Laplante, 2012a;
Klassen and Laplante, 2012b; Dyreng and Markle, 2016; De Simone, Huang, and Krull 2017) and
retain the associated cash or other assets offshore (Hines and Hubbard 1990; Desai, Foley, and
Hines, 2001; Foley et al., 2007; Laplante and Nesbitt, 2016; De Simone et al., 2017c). Repatriation
taxes therefore create an internal capital market friction that constrains cash mobility within the
firm and raises the cost of using foreign cash for domestic purposes.13 Instead, a firm’s external
13Several related papers focus on firm behavior following the repatriation tax holiday enacted by the American Jobs Creation Act of 2004, which intended to reduce the internal capital market tax friction by temporarily decreasing repatriation taxes. Blouin and Krull (2009) and Dharmapala, Foley, and Forbes (2011) find that the majority of funds (50 percent and 90 percent, respectively) were paid out to shareholders. Faulkender and Petersen (2012) challenge these results, finding that only a small portion was paid out to shareholders and that financially constrained firms used this cash for investment. Laplante and Nesbitt (2016) use the AJCA to estimate the likelihood and determinants of having trapped foreign cash. De Simone et al. (2017c) examine excess cash holding behavior following the AJCA and
11
cost of capital could actually be lower than the cost of internal capital, inclusive of these
repatriation taxes. As a result, we may observe an attenuated negative relation between the level
of external domestic financing and tax-induced foreign cash holdings for firms with a high demand
for domestic cash.14
Three studies provide evidence consistent with this prediction, although they do not
empirically test this relation. In a survey of approximately 400 tax executives, approximately 44
percent state that they issue external debt capital to avoid the repatriation tax (Graham et al. 2010).
Albring (2006) and Petzel and Salbador (2016) estimate the probability that a firm will issue
external debt as a function of the firm’s permanently reinvested foreign earnings, a financial
reporting decision that results in the firm not accruing a deferred tax liability for estimated U.S.
repatriation taxes on foreign earnings. Both papers find that foreign earnings designated as
permanently reinvested are positively associated with the likelihood of the U.S. MNC issuing
public debt.15
We expect that firms with large amounts of tax-induced foreign cash and the greatest
demand for domestic cash, such as those making shareholder payouts (which must be paid by the
proposals in Congress for another (temporary or permanent) repatriation tax reduction. In other work, Chen (2014) examines the effect of repatriation tax costs on investors’ valuation of cash holdings and finds a negative relation. In subsequent tests, she tests and finds that this relation is pronounced in firms with limited domestic borrowing capacity. Harford et al. (2017) also investigate the valuation of foreign cash and discuss how the tax costs constrain the internal capital market.14 Prior literature provides an ordering for the use of internal versus external capital, exclusive of tax costs. By adding in tax costs, the cost to access internal capital should increase. However, this does not necessarily mean that information asymmetry with external capital providers will also increase. In fact, external capital providers may know and understand that the firm seeks external capital due to the repatriation tax. Therefore, our hypotheses assume that the external cost of capital is constant (assuming no incremental information asymmetry between the capital providers and the firm due to the repatriation tax and therefore no additional price protection). Empirically, Ma et al. (2016) and Blaylock et al. (2016) interpret their results as evidence of a positive relation between repatriation tax costs and the cost of external debt financing. 15 Albring (2016) studies a relatively small sample of 156 manufacturing firms from 1993 to 2002. Notably, Albring’s (2006) results are not robust to measuring the change in the dollar value of public debt issuances, which is attributed to an even smaller sample of firms with available data. Petzel and Salbador (2016) study a more recent sample of 222 S&P 500 firms and focus on testing whether the likelihood of debt issuances was different in the pre-AJCA (1998-2002) and post-AJCA (2006-2010) period. They find a positive association in the post-AJCA period but are unable to replicate the Albring (2006) result in the pre-AJCA period.
12
U.S. domestic parent company), requiring funding for domestic operations, and needing capital
for domestic investment, will turn to external debt financing rather than incur the repatriation tax
cost to access their costly internal foreign capital. Thus, we predict the following:
H2: The negative relation between tax-induced foreign cash and domestic firm liabilities
is attenuated for firms with a greater demand for domestic cash.
However, it is not clear ex ante that we would observe an attenuated negative relation
between tax-induced foreign cash and domestic liabilities for several reasons. First, Pinkowitz et
al. (2016) show that the high levels of foreign cash holdings are concentrated in R&D-intensive
firms, suggesting that we may not observe a relation between offshore cash, repatriation tax
liabilities, and domestic borrowing in a broad sample of MNCs. Second, Desai et al. (2007) show
that approximately 30 percent of U.S. MNCs’ foreign affiliates continue to pay dividends to the
U.S. parent for U.S. investment and payout needs, despite the repatriation tax liability. Thus, this
paper would suggest that some firms do not need to access external financing because they are
able to use internal funds domestically. Third, Altshuler and Grubert (2003) show that firms
employ different foreign organizational tax planning strategies, which permit firms to effectively
repatriate foreign cash while minimizing the repatriation tax.16 Fourth, firms may avoid deemed
dividend treatment on intercompany loans from foreign subsidiaries by borrowing (on a short-term
basis) from foreign subsidiaries. Finally, we may not observe a result if firms opt to trim domestic
operations and investment to save domestic cash rather than pay for external capital, which
concurrent papers suggest is higher for firms with greater repatriation tax costs (Blaylock et al.,
16 Discussions with practitioners confirm that companies with high expected repatriation costs have developed a number of strategies to tax-efficiently repatriate foreign earnings including separating earnings and profits from foreign taxes, so-called “Killer B” and “Deadly D” reorganizations, basis shifting, and others. An ideal cross-sectional test would identify firms engaging in these strategies to examine whether they have lower domestic debt relative to firms unable to capitalize on these strategies; however, we are unable to identify these firms because such strategies are not disclosed.
13
(2016); Ma et al. (2016). For these reasons, we may observe no attenuation in the negative relation
between tax-induced foreign cash and external firm financing.
3. Research Design, Sample Selection, and Empirical Constructs 3.1.Research Design
To test the relation between tax-induced foreign cash and domestic firm financing (H1),
we use the following regression specification:
!"#_%&'(),+ = . +012345),+ + 5"678"9: + ; (1)
Dom_Liabi,t is a scaled measure of the dollar amount of domestic liabilities in year t and is
described in more detail in Section 3.2.1. TIFCi,t is the firm’s estimated tax-induced foreign cash
in year t (following Hanlon et al., 2015 and Foley et al., 2007) and is described in Section 3.2.2.
We define the controls in Section 3.2.4. As validation and as a test of our first hypothesis, we
predict 01 < 0.
To test whether the negative relation between TIFCi,t and Dom_Liabi,t is attenuated for firms
with the greatest domestic demand for cash (H2), we use the following regression specification:
!"#_%&'(),+ = . +012345),+ + 0?!"#_!@#'6A),+ + 0B2345),+ ∗ !"#_!@#'6A),+
+5"678"9: + ; (2)
where Dom_Demandi,t is an indicator equal to one if the firm has a high demand for domestic cash
in year t (as described in Section 4.2.3.) relative to other firms in the sample, or zero otherwise.
The other variables are the same as in Equation (1) and are described in Section 3.2.4. We predict
that the coefficient on the interaction of TIFCi,t and Dom_Demandi,t (0B) will be positive as a test
of H2.
All regressions include industry and year fixed effects to control for underlying time trends
and industry-specific characteristics that may affect firm financing decisions unrelated to tax-
14
induced foreign cash. All variables are winsorized at the 1st and 99th percentiles. Finally, we report
standard errors that are clustered by firm and by year.
3.2.Empirical Constructs
3.2.1. Measures of Financing
Firms generally disclose in their financial statements the total amount of liabilities after
consolidating on a worldwide basis and eliminating intercompany obligations. Therefore, to
observe domestic liabilities, we use data from the BEA. These data permit observation of
jurisdiction-specific (foreign versus domestic) liabilities and include intercompany loans, which
we intend to further analyze in future tests. To construct Dom_Liabi,t, we use the total amount of
liabilities reported by the U.S. reporter (i.e., the U.S. parent company) on the BEA annual and
benchmark surveys from 1999 through 2012, which includes short-term payables and trade credit,
bank debt, publicly traded debt, and other long-term liabilities.17 Following Faulkender and Smith
(2016), we construct two measures of domestic liabilities, including Dom_Mkt_Liabi,t, calculated
as total domestic liabilities divided by the sum of domestic liabilities and the market value of equity
(CSHO*PRCC_F), and Dom_Book_Liabi,t, calculated as total domestic liabilities divided by the
sum of domestic liabilities and the book value of equity (SEQ). In Section 4, we also discuss
results using two additional measures of net domestic market leverage and net domestic book
leverage, which are calculated by subtracting domestic cash holdings from the total amount of
domestic liabilities prior to scaling.18
3.2.2. Tax-Induced Foreign Cash Holdings
17 The BEA “annual” surveys conducted in 2000-2003, 2005-2008, and 2010-2013 request the amount of “Total Liabilities” for the U.S. parent but do not include additional details as to the type of liability (i.e., trade credit versus long-term debt). The “benchmark” surveys conducted in 1999, 2004, and 2009 include separate line-items for Trade accounts/notes payable and Other liabilities. In future work, we will perform tests using only the benchmark survey data for these three years to examine if the relation with tax-induced foreign cash varies based on the type of liability. 18 Results are also robust to scaling by the sum of worldwide debt and either the market value of equity or the book value of equity.
15
To estimate the amount of foreign cash that is held due to repatriation taxes (TIFCi,t), we
use Foley et al.’s (2007) firm-level measure of repatriation taxes. We first calculate the total U.S.
tax that would be due on foreign earnings (before a foreign tax credit) by multiplying a firm’s
foreign income (PIFO) by the U.S. statutory rate (35 percent).19 From this amount, we subtract the
amount of foreign taxes paid (TXFO) as an estimate of the foreign tax credit. The higher of the net
tax due, or zero, is the repatriation tax cost, or REPATi,t. As in prior papers, we acknowledge that
this measure reflects three assumptions: i) foreign reported earnings are an approximation of
unobservable foreign taxable income, ii) use of annual foreign earnings is proportional to the total
stock of foreign earnings that have not yet been repatriated, and iii) foreign tax rates applicable at
the time foreign taxes are paid will be similar to actual foreign rates at the time of repatriation.
To then estimate TIFCi,t, we follow Hanlon et al. (2015) and regress a measure of total
foreign cash in year t on REPAT and control variables (all measured in year t-1). Because foreign
cash data are unavailable from publicly available sources for a large sample of firms, we also
obtain data on foreign cash holdings from the BEA. We present the results of this estimation in
Appendix B and, for comparison purposes, also present Col. (1) from Appendix B in Hanlon et al.
(2015). The similarity in the size and economic significance of the coefficients confirms that the
relation between foreign cash and repatriation tax costs has persisted through our more recent
sample period. We then use these coefficients to predict out-of-sample values for the amount of
foreign cash attributable to the repatriation tax, or tax-induced foreign cash (TIFCi,t), for purposes
of estimating Equations (1) and (2).20
19 Foley et al. (2007) use both the statutory tax rate and a firm’s marginal tax rate and arrive at similar empirical results. 20 Beginning in 2009, the BEA surveys did not include a separate line for foreign affiliates’ cash holdings. Thus, we estimate tax-induced foreign cash for the period 1998 through 2008 using observed foreign cash data and then estimate out-of-sample values for the remaining sample period. Results are robust to excluding 2009 through 2012 (the years for which foreign cash data are not directly observable) as tabulated in Table 4.
16
3.2.3. Measures of Domestic Demand for Cash
We measure the demand for domestic cash (Dom_Demandi,t) several ways. First, we
construct an indicator variable Payouti,t, which is equal to one if the firm repurchases shares or
pays a dividend in year t, and zero otherwise. Second, we use the proportion of domestic employees
(Dom_Empi,t) as a proxy for domestic capital needed to fund domestic operations. Specifically, we
construct an indicator equal to one if the firm is in the highest quartile based on Dom_Empi,t in a
firm-year, and zero otherwise. We also construct a similar indicator using the proportion of
domestic compensation (Dom_Compi,t). Third, we construct an indicator Underinvesti,t-1, which
is equal to one if the firm is under-invested domestically in year t-1, and zero otherwise. We
identify firms that are domestically underinvested using the worldwide firm-level investment
specification from Biddle et al. (2009), adapted for domestic operations by Harford et al. (2017).
Specifically, we regress domestic capital expenditures scaled by domestic assets in year t-1 on
foreign and domestic measures of sales growth, return on sales, and size, all of which are measured
in year t-2 using BEA data. If the firm has a negative residual from this regression, we identify the
firm as domestically underinvested in year t-1. In subsequent tests, we also set Underinvesti,t-1
equal to one if the residual from this regression is in the bottom quartile of the sample in that year,
or zero otherwise.
3.2.4 Control Variables
The control variables include constructs that prior literature has shown to be related to firm
financing, including firm size, financial performance, investment spending, and shareholder
payout policies (Rajan and Zingales, 1995; Frank and Goyal, 2003; Leary and Roberts, 2010; Rauh
and Sufi, 2010; Naranjo, Saavedra, and Verdi, 2017). We follow Faulkender & Smith (2016) to
create the empirical proxies for these constructs. We first include Sizei,t, measured as the logarithm
17
of a firm’s total sales, to control for the firm’s demand for external financing. We control for the
cost of debt by including the indicator Rating_Indi,t, equal to one if the firm has a bond rating, and
zero otherwise. We control for Tangibilityi,t, defined as the sum of the firm’s fixed assets (PPENT),
scaled by the firm’s total assets, and ROAi,t, which is pre-tax income (PI) scaled by total assets.
We include Div_Indicatori,t, which is an indicator equal to one if the firm pays a dividend, and
zero otherwise. We also control for the firm’s investment spending and growth opportunities by
including R&Di,t, the total amount of research and development expense (XRD) divided by the
firm’s assets, and Advertisingi,t, the total amount of advertising expense (XAD) divided by the
firm’s assets. Due to a significant number of missing values, we set R&Di,t and Advertisingi,t equal
to zero if missing. MTBi,t is the ratio of market value of equity (CSHO*PRCC_F) plus total firm
debt (DLTT+DLC) to total assets. To control for the variability in firm performance and cash flow
affecting the firm’s demand for external financing, we calculate σ(CF)i,t, which is the standard
deviation of a firm’s operating cash flows (OANCF) over the five preceding years. Finally, we
control for Depreciationi,t, measured as total depreciation (DP) scaled by total assets.
3.3. Sample Selection and Descriptive Statistics
We select all U.S. C Corporations in the intersection of Compustat and the BEA for the
period 1999 through 2012 (n=12,284 firm-years). We eliminate observations for firms not
incorporated in the U.S. (n=664 firm-years) and in the financial or utilities industries (n=1,264)
because they are subject to different regulatory provisions that may affect their tax profile. We
require data to estimate TIFCi,t, as well as the measures of domestic liabilities and control
variables, which results in a loss of an additional 907 observations. Specifically, we require
positive, non-missing data for total assets (AT), cash (CHE), pre-tax income (PI), and sales
(SALE). We also require non-missing data to construct measures of the book value and market
18
value of equity (CEQ, CSHO, PRCC_F), capital expenditures (CAPX), consolidated debt (DLTT
and DLC), domestic liabilities from the BEA, ordinary income (OIBDP), and fixed assets
(PPENT). This results in a sample of 9,449 firm-year observations. Table 1, Panel A outlines
these steps. Table 1, Panel B shows that the sample of firms is evenly distributed across the period.
Furthermore, Panel B shows that the 9,449 firm-year observations reflect the activity of almost
136,000 foreign affiliates.21
Table 2, Panel A presents descriptive statistics. Due to limitations on data disclosure, we
present only mean values and the standard deviation for variables constructed from BEA data. On
average, sample firms have Dom_Mkt_Liabi,t (Dom_Book_Liabi,t) of 0.34 (0.50). Note that these
amounts are higher than the total worldwide leverage Mkt_Levi,t (Book_Levi,t) of 0.24 (0.37),
calculated using consolidated amounts of long-term debt from the publicly reported financial
statements in Compustat. The higher values of domestic liabilities relative to worldwide debt is
attributable to two important differences in the construction of these variables. First, we can only
observe total liabilities from BEA, as opposed to the domestic equivalent of long-term debt used
to measure total worldwide market (book) leverage. However, we also calculate worldwide
Mkt_Liabi,t (Book_Liabi,t) using total liabilities from Compustat (LT) instead of long-term debt
(DLC+DLTT) to compare to the BEA measures that we use and also present descriptive statistics
for these variables in Table 2. The worldwide averages increase to 0.40 (for Mkt_Liabi,t) and 0.57
(for Book_Liabi,t) and are (as expected) higher than the analogous domestic measures using BEA
data given that they reflect worldwide (not just domestic-only) liabilities. Second, the measures
21 The BEA benchmark surveys conducted in 1999, 2004, and 2009 are more comprehensive in that they cover more MNCs and obtain more detailed data on affiliate-level activity in these years. Because our sample is composed of large, publicly-traded MNCs that meet the filing thresholds in most years (including the non-benchmark survey years), the number of firms does not vary significantly across the period. However, because the benchmark surveys in 1999, 2004, and 2009 request more detailed data about MNCs’ foreign affiliate activity, we observe larger numbers of foreign affiliates in these three years.
19
constructed from BEA data include intercompany debt between the U.S. operations and foreign
affiliates, whereas the consolidated Compustat data eliminate these intercompany loans. In sum,
the BEA measures capture a firm’s total domestic liabilities and allow for subsequent tests of how
tax-induced foreign cash affects the level of both external and internal financing.
The average amount of tax-induced foreign cash (TIFCi,t) is 0.17. This amount is greater
than the 0.10 average reported in Hanlon et al. (2015), likely due to the more recent sample period
during which firms have been earning more amounts of income offshore and retaining (rather than
repatriating) larger amounts of corresponding cash in foreign jurisdictions. The average predicted
amount of cash holdings attributable to other non-tax factors (PredForCash-Controlsi,t) is -4.82,
consistent with the value of -4.87 in Hanlon et al. (2015). The mean values of the control variables
are very similar to those in Faulkender and Smith (2016), including Mkt_Levi,t of 0.24 (0.21 in
F&S), Book_Levi,t of 0.37 (0.34 in Faulkender and Smith 2016), Tangibilityi,t of 0.26 (0.26 in
Faulkender and Smith 2016), and Div_Indicatori,t of 0.56 (0.59 Faulkender and Smith 2016).
Table 2, Panel B presents a correlation matrix.
4. Results
4.1 Tests of H1
Table 3 presents results from estimating Equation (1), in which we regress measures of
domestic liabilities on TIFCi,t and control variables.22 The first three columns use Dom_Mkt_Liabi,t
as the dependent variable, and the last three columns use Dom_Book_Liabi,t. Columns (1) and (4)
estimate Equation (1) on the sample years for which we observe foreign cash (1999 through 2008),
22Although this paper focuses on the effect of repatriation taxes on internal capital markets and domestic financing levels, we also examine the relation between estimated repatriation taxes and total worldwide liabilities for validation and present these results in Appendix C. Because a firm’s repatriation tax cost is negatively correlated with a firm’s MTR (firms with high estimated repatriation taxes have relatively low global effective tax rates relative to firms with low estimated repatriation taxes), prior literature motivates a negative relation between estimated repatriation taxes or tax-induced foreign cash holdings (which is positively correlated with a firm’s repatriation tax cost liability) and corporate debt financing. Consistent with this prediction, we observe a negative relation between total worldwide market and book liabilities and our proxies for repatriation taxes and tax-induced foreign cash.
20
such that we can also include the residual from the first-stage estimation of TIFCi,t (presented in
Appendix B), while the remaining columns use out-of-sample predicted values for all years from
1999 through 2012. In Column (3) and (6), we present results using the net amounts of
Dom_Mkt_Liabi,t and Dom_Book_Liabi,t, which are calculated by subtracting domestic cash from
total domestic liabilities in the numerator.23
Across all six columns, we find that domestic borrowing is negatively and significantly
associated with a firm’s estimated tax-induced foreign cash. This result is consistent with the prior
literature on internal capital markets and the pecking order theory, which states that firms should
first use internal cash before accessing external financing. It is also consistent with the literature
on marginal tax rates, which finds that the lower the marginal tax rate (and, by extension, the
higher the repatriation tax cost and the higher a firm’s tax-induced foreign cash), the lower the
amount of external financing.
To estimate the economic magnitude of these results, we standardize TIFCi,t and the control
variables to have a mean of zero and a standard deviation of one (untabulated). The coefficient of
-0.028 in Column (1) means that a one-standard-deviation increase in a firm’s tax-induced foreign
cash is associated with an approximately 0.9 percentage point decrease in domestic borrowing.
Given the average liabilities ratio of 0.34 (Table 2, Panel A), this is equivalent to an approximately
2.6 percent decrease in the amount of domestic liabilities. Using the book value of domestic
liabilities in Columns (4) to (6), we observe the same finding: domestic borrowing is negatively
and significantly associated with TIFCi,t. A one-standard-deviation increase in a firm’s tax-induced
foreign cash is associated with an approximately 1.6 percentage point percent decrease in domestic
23 Domestic cash is only observable in the same years in which we can observe foreign cash (1999-2008). Therefore, we can only construct these net leverage measures for a sub-sample of observations.
21
borrowing. Given an average liabilities ratio of 0.5 in this sample, this is equivalent to a 3.2 percent
decrease in the amount of domestic liabilities.24 We find results of similar size and significance in
Columns (3) and (6) when using the net amounts of domestic market and book leverage,
confirming that our results are not sensitive to the type or measurement of domestic liabilities.
Finally, we note that the relation holds for both samples presented in Table 3, suggesting that our
predicted values of tax-induced foreign cash for firm-years that do not report foreign cash are
reasonable. We therefore use this sample in all further tests.
4.2 Tests of H2
Table 4 presents results of estimating Equation (2), which tests whether the negative
relation between domestic financing and tax-induced foreign cash is attenuated for firms that need
domestic cash. In Panel A, we examine firms requiring domestic cash for shareholder payouts
because distributions to shareholders of U.S. incorporated firms must be made through the U.S.
parent entity. To test whether these payout obligations affect the relation between TIFCi,t and
domestic borrowing, we replace the control variable Div_Indicatori,t with the variable Payouti,t and
interact this variable with TIFCi,t. In Columns (1) and (2), Payouti,t is an indicator equal to one if
the firm-year observation either pays a dividend or repurchases shares, and zero otherwise. We
further test the potentially different effects of dividends and repurchases by setting Payouti,t equal
to one if the firm-year observation pays a dividend (repurchases shares) in Columns (3)-(4)
(Columns (5)-(6)), or zero otherwise.
We estimate a negative and significant coefficient on TIFCi,t in all six columns of Panel A,
consistent with results of H1. We again standardize the coefficients for purposes of interpreting
24 The standardized coefficients on TIFCi,t, are -0.009 and -0.016 using the BEA domestic market and book leverage ratios, respectively.
22
economic magnitude (untabulated). The coefficient of -0.062 in Col. (1) means that a one-
standard-deviation increase in a firm’s tax-induced foreign cash is associated with a 2.0 percentage
point decrease in domestic borrowing, or 5.8 percent decrease in the amount of domestic liabilities
for the average firm. As predicted in our second hypothesis, we find a positive coefficient on the
interaction of Payouti,t and TIFCi,t across five of the six columns, consistent with an attenuation of
the negative relation between tax-induced foreign cash and domestic borrowing for firms that need
domestic cash to distribute funds to shareholders. The positive coefficient of 0.039 in Col. (1)
means that a one-standard-deviation increase in a payout firm’s tax-induced foreign cash is
associated with a 0.8 percentage point decrease in domestic borrowing. This translates into a 2.4
percent decrease in domestic borrowing, which is approximately 40 percent of the estimated effect
calculated above for non-payout firms. This finding suggests that the cost of external domestic
finance for firms that must meet shareholder demands for return of capital is less expensive than
incurring the repatriation tax cost to access the firm’s own foreign capital.
We observe that the coefficients on the interaction term in Columns (5) and (6) for
repurchasing firms appear larger than the coefficients in Columns (3) and (4). Statistical tests
confirm that the coefficient in Col .(5) is significantly larger than the interaction coefficient in
Column (3) (p-value = 0.086), although the coefficients in Col. (4) and (6) are not significantly
different. The relatively weak results for dividend-paying firms suggest that the attenuated
negative relation between tax-induced foreign cash and domestic liabilities is primarily driven by
firms who repurchase shares.
In Panel B, we test if the relation between domestic borrowing and TIFCi,t is attenuated for
firms requiring domestic cash to fund operations in the form of domestic employees and
compensation expense (Dom_Employmenti,t). In Columns (1) and (2), Dom_Employmenti,t is an
23
indicator variable defined based on whether the firm is in the highest quartile based on the
proportion of domestic employees to the firm’s total employees; in Columns (3) and (4), it is
defined based on whether the firm is in the highest quartile based on the proportion of domestic
compensation expense to total compensation expense. Interestingly, we find no evidence of an
attenuated negative relation between TIFCi,t and domestic liabilities for firms with a relatively high
proportion of domestic employees or domestic compensation expense when using the market value
of domestic liabilities as the dependent variable (Columns (1) and (3)). Further, we find some
evidence of a more negative relation when using the book value of domestic leverage as the
dependent variable (Columns (2) and (4)). We therefore conclude that U.S. MNCs with tax-
induced foreign cash do not appear to access domestic financing to fund domestic operations,
measured using the proportion of domestic employment levels or compensation expense.
We next consider firms requiring domestic cash to fund domestic investment opportunities,
as prior literature shows that domestic investment opportunities for firms with a significant amount
of permanently reinvested earnings (the financial accounting assertion related to offshore cash) is
more sensitive to domestic cash flow (Blouin et al., 2016). We first identify firms that are
domestically underinvested by estimating domestic capital expenditures as a function of foreign
and domestic sales growth, return on sales, and size (all measured using BEA data), following
Biddle et al. (2009) and adapted by Harford et al. (2017). A negative residual suggests that the
firm is underinvested domestically in that year and thus has some need for domestic cash to fund
these projects. Panel C presents results of estimating domestic investment in year t-1 as a function
of determinants of investment in year t-2 for our sample of firms. Our results show that domestic
24
investment is positively associated with a firm’s domestic assets and is negatively associated with
foreign assets.25
In Panel D, we present results of estimating Equation (2) when including Underinvesti,t-1
as the measure of domestic demand for cash. We predict a positive coefficient on the interaction
term; that is, we expect that the negative relation between tax-induced foreign cash and domestic
liabilities is attenuated for firms that need to finance domestic investment. In Columns (1) and (2),
Underinvesti,t-1 is an indicator equal to one if the firm-year residual from estimating the domestic
investment model in year t-1 is negative. In Columns (3) and (4), Underinvesti,t-1 is an indicator
equal to one if the residual is in the bottom quartile of the sample. We find limited evidence of a
differential relation for firms that appear to underinvest domestically; specifically, we only observe
a positive and significant coefficient in Column (2). In the remaining columns, the coefficients are
insignificant. Thus, it does not appear that firms with TIFCi,t borrow domestically to address
domestic underinvestment.
The evidence in Table 4 is consistent with our second hypothesis: firms with a greater
domestic demand for cash exhibit an attenuated negative relation between domestic borrowing and
TIFCi,t. However, this result is driven by firms with shareholder payout obligations, most notably
firms that engage in share repurchases, rather than by the need for domestic cash to fund operations
or investment projects. Collectively, these results suggest that some firms with tax-induced foreign
cash borrow domestically to subsequently distribute cash to shareholders.
5. Additional Analyses
25 These results reflect estimating investment in year t-1 (relative to year t when the amount of domestic liabilities are measured) using determinants in year t-2. In additional untabulated tests, we also examine investment in year t using determinant measures from year t-1, which yields a larger sample. In these results, we observe that domestic investment is positively and significantly associated with both domestic sales growth and domestic assets, and is negatively associated with foreign sales growth and foreign assets, consistent with the predicted effects from Harford et al. (2017).
25
5.1 Falsification Tests
It is possible that our results could reflect that some firms have a lower cost of borrowing
rather than a high domestic demand for cash relative to other firms. To mitigate this concern, our
main tests control for the cost of debt (Rating_Indi,t). Nonetheless, we perform additional tests to
address this alternative explanation for our results. Specifically, we perform a falsification test by
constructing an indicator LowDebtCosti,t, equal to one for firms that we expect to have a lower
cost of debt and zero otherwise, and then re-estimating Equation (2) by replacing this indicator for
Dom_Demandi,t. Table 5 presents these results. We use three measures of LowDebtCosti,t. In
Columns (1) and (2), LowDebtCosti,t is Rating_Indi,t, an indicator equal to one if the firm-year has
a bond rating, and zero otherwise.26 In Columns (3) and (4), LowDebtCosti,t is Rating_Cati,t, a
categorical variable equal to zero if the firm has no rating, or equal to one, two, or three if the
firm’s bond rating is in the C range, B range, or A range, respectively, in year t. In Columns (5)
and (6), LowDebtCosti,t is an indicator if the firm is in the bottom tercile based on the firm’s
Whited-Wu Index score. In Col (1) through (4), we find a positive association between the main
effect of LowDebtCosti,t and domestic financing, suggesting that firms that can access external
capital at a lower cost do so. However, in five out of the six columns we find an insignificant
coefficient on the interaction between LowDebtCosti,t and TIFCi,t, meaning that there is no
statistical difference in the relation between tax-induced foreign cash and domestic liabilities for
firms with a lower cost of debt. These results show that our main results are not driven by
borrowing behavior of firms that can more inexpensively access the capital markets.
5.2 Total Worldwide and Foreign Borrowing
26 This variable is the control variable that we included in the specifications in Tables 3 and 4; in this test, we now consider this a variable of interest and interact it with TIFCi,t.
26
Our main results show that the negative relation between TIFCi,t and domestic liabilities is
attenuated for firms with shareholder payout obligations. A natural question is whether the higher
levels of domestic liabilities in these firms reflect either a substitution of foreign liabilities for
domestic liabilities or additional financing that increases the total worldwide liabilities.27
To test the potential substitution of domestic-for-foreign liabilities, we re-estimate
Equation (2) but use measures of total (consolidated) worldwide liabilities as the dependent
variable. Consistent with the tax benefits of debt literature and our results for H1, we expect a
negative relation between tax-induced foreign cash and total liabilities. We make no prediction
about the main effect of shareholder payouts. Our variable of interest is the interaction between
Payouti,t and TIFCi,t. We may find no effect of this interaction term on total liabilities if the higher
domestic liabilities of these payout firms substitute for foreign liabilities (i.e., the amount of
foreign financing goes down as domestic financing increases). In contrast, the interaction may be
positive if the higher domestic liabilities reflect incremental financing.
Table 6, Panel A presents these results. In Columns (1) and (2), Payouti,t is an indicator
equal to one if the firm-year observation pays either a dividend or repurchases shares, and zero
otherwise. In Columns (3)-(4) (Columns (5)-(6)), Payouti,t is an indicator equal to one if the firm-
year observation pays a dividend (repurchases shares), or zero otherwise. Across all measures of
shareholder payouts and total worldwide liabilities, we estimate a positive and significant
coefficient on the interaction term. These results indicate an attenuation of the negative relation
27 Several papers study how tax rates affect debt placement by examining the level of debt within the foreign subsidiaries of an MNC. Desai, Foley, and Hines (2004) show that a ten percent increase in higher local country rates is associated with 2.7 percent more subsidiary debt because the value of the interest deduction is greater in higher-tax countries. Similarly, Huizinga, Laeven, and Nicodeme (2008) find that firms strategically locate internal intracompany debt in relatively high-tax subsidiaries. Blouin, Huizinga, Laeven, and Nicodeme (2014) examine the extent to which this behavior is curbed by local country thin capitalization regulations and enforcement. In our tests, we study the placement of liabilities in the U.S. parent versus in foreign affiliates.
27
between tax-induced foreign cash and total liabilities for firms making shareholder payouts,
suggesting that these firms have higher total liabilities relative to other firms. The coefficients in
Col. (5) suggest that there is no overall decrease in total borrowing associated with an increase in
tax-induced foreign cash (coefficient of -0.054 on the main effect of TIFCi,t, plus the coefficient
of 0.054 on the interaction term).
To shed further light on whether higher domestic financing is incremental, we next test the
relation between tax-induced foreign cash and foreign liabilities (from BEA) for firms making
shareholder payouts. We present these results in Panel B. If higher domestic liabilities are
additional liabilities that increase the worldwide liabilities, we expect no change to foreign
liabilities, and therefore an insignificant coefficient on the interaction. In contrast, if higher
domestic liabilities replace foreign liabilities, we expect a negative coefficient on the interaction.
In four out of the six columns, we estimate a positive but insignificant coefficient on the
interaction, suggesting little to no effect on the relation between tax-induced foreign cash and
foreign liabilities for the subsample of firms with shareholder payout obligations. Collectively,
the results in Table 6 provide evidence supporting that the relatively higher levels of domestic
liabilities by the subset of payout firms are incremental worldwide liabilities, not offset by lower
foreign borrowing. Further, because total worldwide liabilities eliminate intercompany debt on
consolidation, these findings also support that our main results are not driven by intercompany
loans that circumvent the deemed dividend rules of U.S. tax law. Given evidence in prior literature
that corporations appear to be under-leveraged (e.g., Graham 2000), future work could explore
whether tax-induced internal capital market constraints push firms with domestic demand for cash
towards a more optimal debt ratio.
5.3 High R&D Firms
28
We next investigate whether the negative relation between tax-induced foreign cash and
domestic liabilities is attenuated for the subset of R&D-intensive firms. We focus on these firms
for several reasons. First, Graham et al. (2010) find that repatriation and investment decisions of
high-intangibles firms are most sensitive to U.S. repatriation tax accounting and cash tax policies.
Furthermore, Pinkowitz et al. (2016) show that the high levels of foreign cash holdings are
concentrated in R&D-intensive firms, which are firms with highly mobile income (Harris, Morck,
Slemrod, and Yeung, 1993; Grubert and Slemrod, 1998; De Simone, Mills, and Stomberg, 2017).
We measure R&D-intensity based on industry affiliation and construct an indicator, HighR&Di,t,
equal to one for firms likely to have high level of R&D and intangibles based on Compustat
reporting of a pharmaceutical or computer software industry affiliation, or zero otherwise.
In Table 7, we present results of testing whether the negative relation between domestic
liabilities and tax-induced foreign cash is attenuated for these high-R&D firms. We observe a
positive and significant coefficient on the interaction term TIFCi,t*HighR&Di,t across both
measures of domestic liabilities, which we interpret as evidence that these firms also borrow more
domestically relative to other firms in the sample. To estimate the economic magnitude of these
results, we again standardize TIFCi,t and the control variables to have a mean of zero and a standard
deviation of one (untabulated). The coefficient of -0.045 in Column (1) means that a one-standard-
deviation increase in a firm’s tax-induced foreign cash is associated with an approximately 1.5
percentage point (or 4.4 percent) decrease in the amount of domestic liabilities for the average firm
in the sample. To validate that these tests are not driven by a differential cost of debt for high-tech
firms, in untabulated tests we examine bond ratings and the Whited-Wu cost of debt index for
these high-R&D firms relative to other firms. If our results are attributable to a differential cost of
debt, we would expect to see high-R&D firms exhibiting a lower cost of debt. In contrast, we find
29
that on average high-R&D firms are less likely to have a bond rating (t=12.76), have lower bond
ratings (t=12.92), and have a higher Whited-Wu Index score (t=11.91).
5.4 Future Work
BEA data allow us to observe different components of liabilities in the benchmark years,
including short-term liabilities (such as trade payables) and long-term liabilities. In future work,
we intend to examine what form of domestic liabilities is preferred by MNCs with tax-induced
foreign cash. In addition, we intend to explore whether firms that do not appear to require
additional domestic liabilities to meet domestic cash needs use intercompany loans instead.
6. Conclusion
In summary, this paper examines how U.S. repatriation taxes and tax-induced foreign cash
affect external financing decisions. The academic literature has begun to document the economic
effects of the U.S. repatriation tax, but there is still much to know about how the large amount of
foreign cash holdings, partially induced by repatriation tax liabilities, affect firms and the U.S.
economy. We extend this line of literature by examining the important decision of why cash-rich
U.S. MNCs access external financing. We revisit the internal capital market and pecking order
predictions that firms will generally use internal capital first to fund operations, investment, and
payout policies. In our setting, this prediction does not seem to be valid for a subset of firms with
domestic cash needs, given that the repatriation tax cost makes internal capital expensive for these
firms. We empirically test and show that the negative relation between domestic borrowing and
tax-induced foreign cash is attenuated for firms that repurchase shares and engage in relatively
high levels of R&D activity. Additional tests suggest that these results are not driven by a
differential cost of debt and that increased domestic liabilities are incremental worldwide
30
liabilities, not offset by lower foreign borrowing. Our study adds to both the capital markets
literature and policy discussions about the implications of the U.S. worldwide tax system.
31
References
Abramowicz, L. 2017. Tech companies to Trump: Show us the tax break. Bloomberg, January 30.
Albring, S. 2006. The Effects of the cost of foreign internal funds on the probability that a firm issues domestic debt. Journal of the American Taxation Association 28: 25-41.
Alchian, A.A. 1969. Corporate management and property rights in H. Manne, ed., Economic Policy and the Regulation of Corporate Securities. American Enterprise Institute 337-360.
Altshuler, R. and Grubert, H. 2003. Repatriation taxes, repatriation strategies and multinational financial policy. Journal of Public Economics 87: 73-107.
Barthold, T. 2016. Joint Committee on Taxation letter to Chairman Brady and Mr. Neal dated August 3, 2016.
Biddle, G., Hilary, G., and Verdi, R. 2009. How does financial reporting quality relate to investment efficiency? Journal of Accounting and Economics 48: 112-131.
Blaylock, B. S., Downes, J. F., Mathis, M. E., and White, S. D. 2016. Measuring the cost of trapped foreign earnings: Evidence from new bond issuances. Working paper.
Blouin, J. L., Core, J., and Guay, W. 2010. Have the tax benefits of debt been overestimated? Journal of Financial Economics 98: 195-213.
Blouin, J. L., Huizinga, H., Laeven, L., and Nicodeme, G. 2014. Thin capitalization rules and multinational firm capital structure. Working paper.
Blouin, J. L., and Krull, L. K. 2009. Bringing it home: A study of the incentives surrounding the repatriation of foreign earnings under the American Jobs Creation Act of 2004. Journal of Accounting Research 47: 1027–59.
Blouin, J., L., Krull, L. K., and Robinson, L. 2016. The location, composition, and investment implications of permanently reinvested earnings. Working paper.
Chen, N. 2014. Repatriation taxes and the value of cash holdings. Working paper.
De Simone, L., Huang, J., and Krull, L. 2017a. R&D and the rising foreign profitability of U.S. multinational corporations. Working paper.
De Simone, L., Mills, L., and Stomberg, B. 2017b. Using IRS data to examine income shifting firms. Working paper.
De Simone, L., Piotroski, J., and Tomy, R. 2017c. Repatriation taxes and foreign cash holdings: The impact of anticipated tax policy. Working paper.
32
DeAngelo, H., and Masulis, R. 1980. Optimal capital structure under corporate and personal taxation. Journal of Financial Economics 8: 3-29.
Desai, M., Foley, F. and Hines, J. 2001. Repatriation taxes and dividend distortions. National Tax Journal 54: 829-851.
Desai, M., Foley, F. and Hines, J. 2004. A multinational perspective on capital structure choice and internal capital markets. Journal of Finance 59: 2451-2487.
Desai, M., Foley, F. and Hines, J. 2007. Dividend policy inside the multinational firm. Financial Management 36: 5-26.
Dhaliwal, D., Trezevant, R., and Wang, S. 1992. Taxes, investment-related tax shields and capital structure. Journal of the American Taxation Association 14: 1-21.
Dharmapala, D., Foley, C. F., and Forbes, K. 2011. Watch what I do, not what I say: The unintended consequences of the Homeland Investment Act. Journal of Finance 66: 753–87.
Dyreng, S. and Markle, K. 2016. The effect of financial constraints on tax-motivated income shifting by U.S. multinationals. The Accounting Review 91: 1601-1627.
Edwards, A., Kravet, T., and Wilson, R. 2016. Trapped cash and the profitability of foreign cash acquisitions. Contemporary Accounting Research 33: 44-77.
Faulkender, M. and Petersen, M. 2012. Investment and capital constraints: Repatriations under the American Jobs Creation Act. Review of Financial Studies 25: 3351-3388.
Faulkender, M. and Smith, J. 2016. Taxes and leverage at multinational corporations. Journal of Financial Economics 122: 1-20.
Foley, F., Hartzell, J., Titman, S., Twite, G. 2007. Why do firms hold so much cash? A tax-based
explanation. Journal of Financial Economics 86: 579-607. Frank, M.Z., and Goyal, V.K. 2003. Testing the pecking order theory of capital structure. Journal
of Financial Economics 67: 217-248.
Gertner, R., Scharfstein, D., and Stein, J. 1994. Internal versus external capital markets. Quarterly Journal of Economics 109: 1211-1230.
Graham, J. 1996a. Debt and the marginal tax rate. Journal of Financial Economics 41: 41-74. Graham, J. 1996b. Proxies for the marginal tax rate. Journal of Financial Economics 42: 187-221.
Graham, J. 2000. How big are the tax benefits of debt? Journal of Finance 55: 1901-1941. Graham, J. 2003. Tax and corporate finance: A review. Review of Financial Studies 16: 1075-
1129.
33
Graham, J. and Kim, H. 2009. The effects of the length of the tax-loss carryback period on tax receipts and corporate marginal tax rates. National Tax Journal LXII: 413-427.
Graham, J., Hanlon, M., and Shevlin, T. 2010. Barriers to mobility: The lockout effect of U.S. taxation of worldwide corporate profits. National Tax Journal 63: 1111–44.
Graham, J., Lang, M., and Shackelford, D. 2004. Employee stock options, corporate taxes and debt policy. Journal of Finance 59: 1585-1618.
Graham, J., Lemmon, M., and Schallheim, J. 1998. Debt, leases, taxes, and the endogeneity of corporate tax status. Journal of Finance 53: 131-162.
Grubert, H. 1998. Taxes and the division of foreign operating income among royalties, interest, dividends and retained earnings. Journal of Public Economics 68: 269-290.
Grubert, H. and Slemrod, J. 1998. The effect of taxes on investment and income shifting to Puerto Rico. Review of Economics and Statistics 80: 365-373.
Hanlon, M., Lester, R., and Verdi, R. 2015. The effect of repatriation tax costs on firm investment. Journal of Financial Economics 116: 179-196.
Harford, J., Wang, C., and Zhang, K. 2017. Foreign cash: Taxes, internal capital markets, and agency problems. Review of Financial Studies, forthcoming.
Harris, D., Morck, R., Slemrod, J., and Yeung, B. 1993. Income shifting in U.S. multinational corporations. In Studies in International Taxation, University of Chicago Press.
Hines, J.R. and Hubbard, R.G. 1990. Coming home to America: Dividend repatriations by US multinationals. In: Razin, A., Slemrod, J. (Eds.), Taxation in the Global Economy. University of Chicago Press, Chicago, IL: 161-200.
Huizinga, H., Laeven, L., and Nicodeme, G. 2008. Capital structure and international debt shifting. Jounral of Financial Economics 88: 80-118.
Klassen, K. and Laplante, S. 2012a. The effect of foreign reinvestment and financial reporting incentives on cross-jurisdictional income shifting. Contemporary Accounting Research 29: 928-955.
Klassen, K. and Laplante, S. 2012b. Are U.S. multinational corporations becoming more aggressive shifters? Journal of Accounting Research 50: 1245-1286.
Laplante, S. and Nesbitt, W. 2017. Estimating the likelihood that a U.S. multinational firm has cash trapped in foreign subsidiaries. Working paper.
Lattman, P. and Eavis, P. 2013. To satisfy its investors, cash-rich Apple borrows money. New York Times, April 30.
Leary, M.T. and Roberts, M.R. 2010. The pecking order, debt capacity, and information asymmetry. Journal of Financial Economics 95: 332-355.
34
Lewellen, W.G. 1971. A pure financial rationale for the conglomerate merger. Journal of Finance 26: 521-537.
Linebaugh, K. 2012. Top U.S. firms are cash-rich abroad, cash-poor at home. Wall Street Journal, December 4.
Ma, Z., Stice, D., and Wang, D. 2016. Earnings repatriation and bank loan contracting. Working paper.
Mackie-Mason, J. K. 1990. Do taxes affect corporate financing decisions?” Journal of Finance 45: 1471-1493.
Mead, C. and Kucera, D. 2012. EBay skirts cash repatriation with bond issue: Corporate finance. Bloomberg.
Miller, M. H. 1977. Debt and taxes. Journal of Finance 32: 261-275. Modigliani, F., and Miller, M. H. 1958. The cost of capital, corporation finance and the theory of
investment. American Economic Review 48: 261-297. Modigliani, F., and Miller, M. H. 1963. Corporate income taxes and the cost of capital: A
correction (in communications). American Economic Review 53: 433-443. Molina, C. 2005. Are firms underleveraged? An examination of the effect of leverage on default
probabilities. Journal of Finance 60: 1427-1459.
Myers, S. 1984. The capital structure puzzle. Journal of Finance 39: 574-592.
Myers, S. and Majluf, N. 1984. Corporate financing decisions when firms have investment information that investors do not. Journal of Financial Economics, 13: 187-221.
Naranjo, P., Saavedra, D. and Verdi, R. 2017. Financial reporting regulation and financing decisions. Working paper.
Nessa, M. 2017. Repatriation tax costs and U.S. multinational companies’ shareholder payouts. The Accounting Review 92: 217-241.
Petzel, T., and Salbador, H. 2016. Do tax holidays increase expectations of future tax holidays: An examination of debt issuances pre- and post-AJCA. Working paper, Virginia Polytechnic Institute & State University.
Pinkowitz, L., Stulz, R. M., and Williamson, R. 2016. Do U.S. firms hold more cash than foreign firms do? Review of Financial Studies 29: 309-348.
Rajan, R.G., and Zingales, L. 1995. What do we know about capital structure? Some evidence from international data. Journal of Finance 50: 1421-1460.
Rajan, R.G., Servaes, H., and Zingales, L. 2000. The cost of diversity: the diversification discount and inefficient investment. Journal of Finance, 55: 35-80.
35
Rauh, J. and Sufi, A. 2010. Capital structure and debt structure. Review of Financial Studies 23: 4242-4280.
Scharfstein, D. and Stein, J.C. 2000. The dark side of internal capital markets: divisional rent-seeking and inefficient investment. Journal of Finance 55: 2537-2564.
Scholes, M., and Wolfson, M. 1992. Taxes and business strategy: A planning approach. Prentice-Hall, Englewood Cliffs, NJ.
Shevlin, T. 1987. Taxes and off-balance sheet financing: Research and development limited partnerships. The Accounting Review 62: 480-509.
Shevlin, T. 1990. Estimating corporate marginal tax rates with asymmetric tax treatment of gains and losses. Journal of the American Taxation Association 12: 51-67.
Shyam-Sunders, L. and Myers, S. 1999. Testing static tradeoff against pecking order models of capital structure. Journal of Financial Economics 51: 219-244.
Stein, J.C. 1997. Internal capital markets and the competition for corporate resources. Journal of Finance 52: 111-133.
U.S. Senate Permanent Subcommittee on Investigations – Committee on Homeland Security and Governmental Affairs. 2011. Repatriating offshore funds: 2004 tax windfall for select multinationals.
Whited, T., and Wu, G. 2006. Financial constraints risk. Review of Financial Studies 19: 531-
559.
Williamson, O.E. 1975. Markets and hierarchies: Analysis and antitrust implications. Collier Macmillan Publishers, Inc.
Appendix A Variable Definitions
Variables are constructed from Compustat and BEA data. The data codes from Compustat are provided for each variable constructed from this public data source.
Advertising Total advertising expense (XAD), scaled by total sales (SALE). This
variable is set equal to zero if missing. Asset_Size The natural logarithm of total assets (AT).
Book_Lev The ratio of total worldwide debt (DLTT+DLC) to the sum of total debt and the book value of equity (SEQ).
Book_Liab The ratio of total worldwide liabilities (LT) to the sum of total worldwide liabilities and the book value of equity (SEQ).
BTM The ratio of the book value of equity (SEQ) to the market value of equity (PRCC_F*CSHO).
Capex Total capital expenditures (CAPX), scaled by total assets (AT).
Depreciation Depreciation expense (DP), scaled by total assets (AT).
Div_Indicator Indicator variable equal to one if the firm pays a dividend (DV), or zero otherwise.
Dom_Book_Liab Total domestic liabilities (measured using BEA data), scaled by the sum of total domestic liabilities and the book value of equity (SEQ).
Dom_Investment The ratio of domestic capital expenditures (measured using BEA data) to total domestic assets.
Dom_Employment Indicator equal to one if the firm is in the top quartile based on the proportion of domestic employees to total employees (both measured using BEA data) or zero otherwise. Alternatively constructed as an indicator equal to one if the firm is in the top quartile based on the proportion of domestic compensation to total compensation, or zero otherwise.
Dom_Income Pre-tax domestic income (PIDOM), scaled by total assets (AT). Missing values are set equal to zero.
Dom_Mkt_Liab Total domestic liabilities (measured using BEA data), scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F).
Dom_Net_Book_Liab
Total domestic liabilities less domestic cash (both measured using BEA data); this amount is scaled by the sum of total domestic liabilities and the book value of equity (SEQ).
Dom_Net_Mkt_Liab Total domestic liabilities less domestic cash (both measured using BEA data); this amount is scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F).
37
Dom_ROS Total domestic net income scaled by domestic sales (both measured using BEA data).
Dom_Sales_Growth The difference in domestic sales from year t-1 to t, scaled by domestic sales in year t-1 (both measured using BEA data).
Dom_Size The natural logarithm of total domestic assets (measured using BEA data).
Foreign_Cash The natural logarithm of the ratio of foreign cash (measured using BEA data) to net assets (AT minus CHE).
For_ROS Total foreign net income scaled by foreign sales (both measured using BEA data).
For_Sales_Growth The difference in foreign sales from year t-1 to t, scaled by foreign sales in year t-1 (both measured using BEA data).
For_Size The natural logarithm of total foreign assets (measured using BEA data).
HighR&D Indicator equal to one if a firm is in a computer or pharmaceutical industry, or zero otherwise.
LowDebtCost This variable is defined in three ways: first, as an indicator equal to one if the firm has a bond rating; second, as a categorical variable equal to 1, 2, or 3 if the firm’s bond rating is in the C range, B range, or A range, respectively; and third, as an indicator equal to one based on the Whited-Wu index.
Mkt_Lev The ratio of total debt (DLTT+DLC) to the sum of total debt and the market value of equity (CHSO*PRCC_F).
Mkt_Liab The ratio of total worldwide liabilities (LT) to the sum of total worldwide liabilities and the market value of equity (CHSO*PRCC_F).
MTB The sum of the market value of equity and total long-term debt (DLTT+DLC) scaled by total assets (AT).
Payout Indicator equal to one if the firm either pays dividends or repurchases shares in year t, or zero otherwise.
PredForCash-Controls
Estimated amount of foreign cash attributable to non-tax factors, estimated following Foley et al. (2007) and Hanlon et al. (2015). This estimation is reported in Appendix B.
PredForCash- Residual Residual from estimating the Foley et al. (2007) and Hanlon et al. (2015) tax-induced foreign cash model; the residual captures the unexplained portion of foreign cash holdings. This estimation is reported in Appendix B.
Rating_Cat A categorical variable equal to 1, 2, or 3 if the firm’s bond rating is in the C range, B range, or A range, respectively in year t.
Rating_Ind Indicator equal to one if the company has a bond rating, or zero otherwise.
REPAT Measures the incremental U.S. tax due upon repatriation of cash from foreign subsidiaries. This measure is calculated by multiplying foreign earnings (PIFO) by the statutory U.S. tax rate of 35%. From this, foreign taxes (TXFO) are subtracted as an estimate of the allowable foreign tax credit. The remaining liability is the estimated
38
U.S. tax due upon repatriation. The maximum of this difference or zero is scaled by total assets (AT).
R&D Total R&D expenses (XRD), scaled by total sales (SALE). This variable is set equal to zero if missing.
RD_Ratio Total R&D expenses (XRD), scaled by total assets (AT). This variable is set equal to zero if missing.
ROA Pre-tax income (PI) scaled by total assets (AT). σ(CF) Standard deviation, over the preceding five years, of the firm’s
operating cash flow (OANCF) to total assets (AT). σ(OpInc) Standard deviation, over the sample period, of the ratio of the firm’s
EBITDA (OIBDP) to total assets (AT). Size The natural logarithm of total sales (SALE). Tangibility Total plant, property, and equipment (PPENT) scaled by total assets
(AT). TIFC Estimated tax-induced foreign cash estimated by regressing total
foreign cash on REPAT and control variables following Foley et al. (2007) and Hanlon et al. (2015). This estimation is reported in Appendix B.
Underinvest Indicator equal to one if the firm is in the bottom quartile or has a negative residual from the domestic investment model following Biddle et al. (2009) and Harford et al. (2017), zero otherwise.
Whited_Wu An index of firm financing constraints from Whited and Wu (2006) equal to (-0.091*Cash Flow) - (0.062*Dividend_Indicator) + (0.021*Long Term Debt) - (0.044*Size) + (0.102*Industry Sales Growth) - (0.035*Sales Growth).
39
Appendix B First-Stage Estimation of Tax-Induced Foreign Cash
1998 through 2008 This table presents estimated coefficients from a first-stage regression in which we regress a firm’s foreign cash holdings (observed from BEA data) on REPATi,t-1, the estimated amount of a firm’s repatriation tax liability, and control variables. This specification allows us to estimate TIFCi,t, the amount of tax-induced foreign cash, for the period 1998 through 2008. The columns present estimated coefficients and t-statistics from the determinants of foreign cash model following Foley et al. (2007) and Hanlon et al. (2015): Foreign_Cashi,t = a + β1*REPATi,t-1 + β2*Sizei,t-1 + β3*Dom_Incomei,t-1 + β4*Div_Indicatori,t-1 +
β5*BTMi,t-1 + β6*σ(OpInc)i,t-1 + β7*R&Di,t-1 + β8*Capexi,t-1 + β9*Mkt_Levi,t-1 + IndustryFE + YearFE The dependent variable Foreign_Cashi,t is the natural logarithm of the ratio of foreign cash to net assets (total assets minus cash). The variable of interest, REPATi,t-1, is the maximum of foreign earnings (PIFO) times the U.S. statutory tax rate of 35 percent less foreign taxes (TXFO), and zero, scaled by total assets. We use the estimated coefficient on REPATi,t-1 to obtain TIFCi,t as follows: TIFCi,t = 0 1*REPATi,t-1. All variables are defined in Appendix A. The regression specification includes year and industry fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively. Dependent Var: Foreign_Cashi,t Hanlon et al. (2015)
Appendix B, Col. (1) 1989-2003
1998-2008
REPATt-1 45.29*** 43.32*** (5.29) (8.17) DomIncomet-1 -0.70 -0.70* (-0.97) (-1.68) AssetSizet-1 -0.14*** -0.02 (-2.99) (-0.70) Div_Indicatort-1 -0.19 -0.05 (-1.32) (-0.60) BTMt-1 -0.04 -0.01 (-0.67) (-1.35) σ(OpInc)t-1 2.80 2.81** (1.50) (2.05) RD_Ratiot-1 2.04 6.24*** (1.39) (6.38) Capext-1 1.77 -1.47 (1.30) (-1.45) Mkt_Levt-1 -0.66** -1.06*** (-2.08) (-5.37) Industry and Year FE? Y Y Observations 3,012 5,853 R-squared 0.112 0.185
40
Appendix C Relation between Total Liabilities, Repatriation Tax Costs, and Tax-Induced Foreign Cash
This table presents estimated coefficients and t-statistics from the following pooled total worldwide leverage model:
Worldwide_Liabi,t = a + β1*TaxIncentivei,t + β2*Sizei,t + β3*Rating_Indi,t + β4*Tangibilityi,t + β5*ROAi,t + β6*Div_Indicatori,t + β7*R&Di,t + β8*Advertisingi,t + β9*MTBi,t + β10*σ(CF)i,t+ β11*Depreciationi,t +
IndustryFE + YearFE The dependent variable in Col. (1) and (3) is total market leverage Mkt_Liabi,t, measured as total worldwide liabilities (using Compustat data) scaled by the sum of total worldwide liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2) and (4) is total book leverage Book_Liabi,t, measured as total worldwide liabilities (using Compustat data) scaled by the sum of total worldwide liabilities and the book value of equity (SEQ). REPATi,t is the maximum of foreign earnings (PIFO) times the U.S. statutory tax rate of 35 percent less foreign taxes (TXFO), and zero, scaled by total assets. TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), reported in Appendix B. All variables are defined in Appendix A. The specification includes year and industry fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and by year. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.
Dependent Variable: Mkt_Liabi,t (1)
Book_Liabi,t (2)
Mkt_Liabi,t (3)
Book_Liabi,t (4)
REPATi,t -1.139*** -1.239** (-2.822) (-2.562) TIFCi,t -0.023** -0.028*** (-2.464) (-2.601) PredForCash-Controlsi,t -0.097*** -0.082*** (-7.607) (-5.878) Sizei,t 0.010*** 0.023*** 0.009** 0.022*** (2.627) (4.978) (2.317) (4.664) Rating_Indi,t 0.073*** 0.119*** 0.059*** 0.107*** (8.768) (10.097) (7.467) (9.378) Tangibilityi,t 0.127*** 0.032 0.087*** -0.002 (4.701) (0.982) (3.629) (-0.077) ROAi,t -0.504*** -0.573*** -0.452*** -0.527*** (-7.269) (-7.439) (-6.900) (-7.145) Div_Indicatori,t -0.059*** -0.032*** -0.061*** -0.034*** (-7.845) (-3.305) (-8.606) (-3.508) R&Di,t -0.762*** -0.970*** -0.384*** -0.647*** (-12.227) (-10.916) (-4.806) (-6.429) Advertisingi,t -0.240** -0.098 -0.264*** -0.117 (-2.319) (-0.530) (-2.785) (-0.641) MTBi,t -0.098*** -0.011 -0.091*** -0.005 (-9.520) (-1.448) (-8.458) (-0.541) σ(CF)i,t 0.147 0.189 0.351*** 0.363** (1.297) (1.362) (3.089) (2.468) Depreciationi,t -0.444** 0.364 -0.312 0.475* (-2.008) (1.406) (-1.371) (1.746) Industry and Year FE? Y Y Y Y Observations 9,449 9,449 9,449 9,449 R-squared 0.587 0.321 0.620 0.343
41
Table 1 Sample Selection
This table summarizes the sample selection procedures. Panel A lists the data requirements imposed, and Panel B provides the number of firm and foreign affiliate observations by year. In Panel A, we retain observations for all non-financial, non-utility, U.S.-incorporated firms at the intersection of BEA and Compustat data with data to calculate the necessary variables. Specifically, we require positive, non-missing data on assets (AT), cash (CHE), pre-tax income (PI), and sales (SALE). We also require non-missing data for equity (CEQ, CSHO, PRCC_F), capex (CAPEX), debt (DLTT, DLC), domestic debt (from BEA), ordinary income (OIBDP), and fixed assets (PPENT) to construct variables used in the empirical tests. All variables are defined in Appendix A.
Panel A: Sample Selection
No. of obs. dropped
No. of obs. remaining
BEA firm-years joined with Compustat data, 1999-2012 12,284 Less: non-U.S. incorporated firm-years (664) 11,620 Less: financial/utility firms (1,264) 10,356 Less: firm-years missing data to calculate variables (907) 9,449
Panel B: Observations by Year
Year Total
Firm-Years No. of Foreign
Affiliates 1999 722 11,957 2000 747 7,969 2001 722 7,628 2002 715 8,760 2003 697 8,941 2004 638 11,037 2005 669 8,769 2006 690 9,867 2007 625 9,389 2008 603 8,040 2009 757 14,125 2010 675 10,482 2011 612 9,762 2012 577 9,094 Total 9,449 135,820
Table 2 Descriptive Statistics
This table presents descriptive statistics (Panel A) and correlations (Panel B) for key variables. All variables are defined in Appendix A. Due to disclosure requirements, we are restricted from reporting firm-specific median, 5%, and 95% values for the variables constructed from BEA data. Bolded values in Panel B indicate statistical significance at the 10% level.
Panel A: Descriptive Statistics
Variable # Obs. Mean Median Std. Dev. 5% 95% Variables from BEA Data Dom_Mkt_Liabi,t 9,449 0.34 - 0.23 - - Dom_Book_Liabi,t 9,449 0.50 - 0.26 - - Dom_Net_Mkt_Liabi,t 6,191 0.30 - 0.25 - - Dom_Net_Book_Liabi,t 6,191 0.43 - 0.31 - - TIFCi,t 9,449 0.17 - 0.33 - - PredForCash-Controlsi,t 9,449 -4.83 - 0.71 - - PredForCash-Residuali,t 5,853 0.00 - 1.58 - - Other Variables Mkt_Levi,t 9,449 0.24 0.18 0.22 0.00 0.71 Book_Levi,t 9,449 0.37 0.35 0.29 0.00 0.88
Mkt_Liabi,t 9,449 0.40 0.38 0.22 0.09 0.82 Book_Liabi,t 9,449 0.57 0.56 0.23 0.22 0.96 Salesi,t 9,449 7,582 2,017 17,061 211 34,588 Sizei,t 9,449 7.72 7.61 1.53 5.35 10.45 Rating_Indi,t 9,449 0.62 1.00 0.49 0.00 1.00 Tangibilityi,t 9,449 0.26 0.20 0.19 0.04 0.67 ROAi,t 9,449 0.06 0.07 0.11 -0.12 0.22 Div_Indicatori,t 9,449 0.56 1.00 0.50 0.00 1.00 R&Di,t 9,449 0.04 0.01 0.07 0.00 0.19 Advertisingi,t 9,449 0.01 0.00 0.03 0.00 0.06 MTBi,t 9,449 1.42 1.12 1.04 0.50 3.40 Depreciationi,t 9,449 0.04 0.04 0.02 0.01 0.09 σ(CF)i,t 9,449 0.04 0.04 0.03 0.01 0.11
43
Table 2 Descriptive Statistics (cont’d.)
Panel B: Correlation Matrix
(1) (2) (3)
(4) (5) (6)
(7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
(17) (1) Dom_Mtk_Liabi,t 1.00 0.97 0.77 0.76 -0.25 -0.43 -0.05 0.06 0.24 0.22 -0.47 -0.04 -0.32 -0.08 -0.48 0.11 -0.05
(2) Dom_Book_Liabi,t 0.96 1.00 0.78 0.83 -0.21 -0.47 -0.06 0.07 0.25 0.26 -0.41 0.01 -0.34 -0.09 -0.46 0.10 -0.12
(3) Dom_Net_Mkt_Liabi,t 0.79 0.81 1.00 0.94 -0.19 -0.36 -0.04 0.20 0.34 0.18 -0.27 0.04 -0.27 0.01 -0.20 0.13 -0.07
(4) Dom_Net_Book_Liabi,t 0.78 0.86 0.95 1.00 -0.18 -0.42 -0.04 0.23 0.37 0.25 -0.24 0.11 -0.31 -0.01 -0.26 0.11 -0.17
(5) TIFCi,t -0.25 -0.22 -0.19 -0.17 1.00 0.19 0.00 0.00 -0.05 -0.08 0.16 -0.06 0.23 0.05 0.23 -0.07 0.07
(6) PredForCash-Controlsi,t -0.43 -0.46 -0.39 -0.42 0.21 1.00 0.01 -0.14 -0.25 -0.37 0.06 -0.15 0.54 -0.04 0.24 -0.10 0.20
(7) PredForCash-Residuali,t -0.08 -0.06 -0.06 -0.04 0.03 -0.01 1.00 0.03 -0.02 -0.07 0.03 -0.01 0.01 0.05 0.04 -0.01 0.02
(8) Sizei,t 0.03 0.10 0.22 0.25 0.04 -0.12 0.07 1.00 0.55 0.07 0.21 0.36 -0.15 0.12 0.06 -0.10 -0.28
(9) Rating_Indi,t 0.23 0.28 0.37 0.39 0.00 -0.23 0.01 0.54 1.00 0.20 0.02 0.24 -0.17 0.08 -0.06 -0.01 -0.23
(10) Tangibilityi,t 0.22 0.27 0.22 0.28 -0.09 -0.36 -0.04 0.08 0.19 1.00 -0.05 0.14 -0.30 -0.11 -0.12 0.51 -0.11
(11) ROAi,t -0.58 -0.50 -0.30 -0.28 0.20 0.16 0.04 0.23 0.01 -0.03 1.00 0.22 -0.12 0.08 0.33 -0.27 -0.11
(12) Div_Indicatori,t -0.01 0.05 0.10 0.14 0.03 -0.12 -0.01 0.36 0.24 0.20 0.25 1.00 -0.25 0.06 0.01 -0.09 -0.25
(13) R&Di,t -0.30 -0.31 -0.20 -0.23 0.18 0.58 0.03 -0.07 -0.10 -0.28 0.01 -0.09 1.00 -0.03 0.33 0.05 0.18
(14) Advertisingi,t -0.11 -0.10 -0.01 -0.02 0.05 0.03 0.05 0.20 0.03 -0.14 0.12 0.02 0.03 1.00 0.17 -0.05 -0.02
(15) MTBi,t -0.65 -0.60 -0.31 -0.31 0.21 0.32 0.01 0.10 -0.02 -0.14 0.53 0.08 0.32 0.19 1.00 0.01 0.07
(16) Depreciationi,t 0.09 0.10 0.09 0.10 -0.10 -0.15 0.00 -0.08 0.01 0.61 -0.13 0.00 0.00 -0.06 0.00 1.00 0.07
(17) σ(CF)i,t -0.08 -0.13 -0.13 -0.18 0.00 0.16 0.01 -0.28 -0.22 -0.16 -0.09 -0.28 0.09 0.00 -0.02 0.03 1.00
Table 3 H1: Domestic Liabilities and Tax-Induced Foreign Cash
This table presents estimated coefficients and t-statistics from the following pooled domestic leverage model:
Dom_Liabi,t = a + β1*TIFCi,t + β2*PredForCash-Controlsi,t + β3*PredForCash-Residuali,t + β4*Sizei,t + β5*Rating_Indi,t + β6*Tangibilityi,t + β7*ROAi,t + β8*Div_Indicatori,t + β9*R&Di,t + β10*Advertisingi,t +
β11*MTBi,t + β12*σ(CF)i,t+ β13*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1)-(2) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (4)-(5) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the book value of equity (SEQ). The dependent variables in Column (3) and (6) are the net amounts of Dom_Mkt_Liab and Dom_Book_Liab, which are calculated by subtracting domestic cash from total domestic liabilities in the numerator. TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. Columns (1) and (4) present results for sample years for which we observe foreign cash (1999 through 2008). The remaining columns use out-of-sample predicted values for all years from 1999 through 2012, such that we are unable to include the residual from the first-stage estimation of TIFCi,t. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Dependent Variable: Dom_Mkt_Liabi,t Dom_Book_Liabi,t (1) (2) (3) (4) (5) (6) TIFCi,t -0.028*** -0.035*** -0.020** -0.050*** -0.057*** -0.048*** (-2.754) (-4.746) (-2.101) (-3.702) (-5.363) (-3.699) PredForCash-Controlsi,t -0.117*** -0.106*** -0.115*** -0.094*** -0.095*** -0.105*** (-8.377) (-9.212) (-7.780) (-4.789) (-5.762) (-5.287) PredForCash-Residuali,t -0.003 -0.005* (-1.518) (-1.676) Sizei,t 0.010* 0.010* 0.013* 0.025*** 0.021*** 0.031*** (1.860) (1.949) (1.922) (3.922) (4.020) (4.368) Rating_Indi,t 0.053*** 0.060*** 0.065*** 0.108*** 0.120*** 0.128*** (4.098) (6.266) (4.399) (6.684) (9.168) (7.213) Tangibilityi,t 0.113*** 0.098*** 0.168*** 0.022 0.004 0.118** (3.340) (2.957) (4.387) (0.533) (0.111) (2.521) ROAi,t -0.711*** -0.735*** -0.684*** -0.536*** -0.567*** -0.607*** (-13.171) (-13.672) (-10.358) (-6.229) (-7.195) (-5.987) Div_Indicatori,t -0.050*** -0.045*** -0.040*** -0.038*** -0.032*** -0.027* (-4.939) (-5.432) (-3.891) (-2.807) (-2.737) (-1.900) R&Di,t -0.343*** -0.417*** -0.413*** -0.523*** -0.590*** -0.702*** (-4.403) (-5.718) (-4.643) (-4.087) (-5.670) (-5.205) Advertisingi,t -0.257** -0.173 -0.306** -0.105 -0.057 -0.080 (-2.117) (-1.616) (-2.258) (-0.459) (-0.277) (-0.329) MTBi,t -0.055*** -0.053*** -0.045*** -0.010 0.000 -0.013 (-7.263) (-7.360) (-6.078) (-0.930) (0.031) (-1.469) σ(CF)i,t 0.338*** 0.385*** 0.141 0.344** 0.467*** -0.046 (2.672) (2.999) (1.027) (1.980) (2.965) (-0.242) Depreciationi,t -0.480* -0.319 -0.654** 0.311 0.641** 0.217 (-1.690) (-1.167) (-2.166) (0.932) (1.963) (0.608) Industry and Year FE? Y Y Y Y Y Y Observations 5,853 9,449 6,191 5,853 9,449 6,191 R-squared 0.540 0.534 0.503 0.300 0.308 0.365
Table 4 H2: Demand for Domestic Cash
Panel A: Payout Policy – Test of Domestic Liabilities and Tax-Induced Foreign Cash This table presents results of tests of H2. Panel A presents estimated coefficients and t-statistics from the following pooled domestic leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Dom_Liabi,t = a + β1*TIFCi,t + β2*Payouti,t + β3*TIFCi,t*Payouti,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9* R&Di,t + β10*Advertisingi,t + β11*MTBi,t + β12*σ(CF)i,t+
β13*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1), (3), and (5) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2), (4), and (6) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), Payouti,t is an indicator variable equal to one if the firm either pays dividends or repurchases shares in year t, and zero otherwise. In Col. (3)-(4), Payouti,t is an indicator variable equal to one if the firm pays dividends in year t, and zero otherwise. In Col. (5)-(6), Payouti,t is an indicator variable equal to one if the firm repurchases shares in year t, and zero otherwise. We exclude Div_Indicatori,t due to collinearity with Payouti,t. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Payouti,t = 1 if: Dividend or Repurchase Dividend Repurchase Dependent Variable: Dom_Mkt_Liabi,t
(1) Dom_Book_Liabi,t
(2) Dom_Mkt_Liabi,t
(3) Dom_Book_Liabi,t
(4) Dom_Mkt_Liabi,t
(5) Dom_Book_Liabi,t
(6) TIFCi,t -0.062*** -0.093*** -0.044*** -0.075*** -0.064*** -0.086*** (-4.659) (-4.767) (-4.996) (-5.823) (-4.930) (-5.050) Payouti,t -0.067*** -0.075*** -0.044*** -0.075*** -0.044*** -0.039*** (-5.838) (-5.560) (-4.996) (-5.823) (-5.157) (-3.889) TIFCi,t*Payouti,t 0.039*** 0.049** 0.022 0.042** 0.050*** 0.047** (2.909) (2.474) (1.614) (1.983) (3.386) (2.327) PredForCash-Controlsi,t -0.106*** -0.095*** -0.107*** -0.095*** -0.104*** -0.093*** (-9.179) (-5.864) (-9.205) (-5.747) (-8.742) (-5.624) Sizei,t 0.009* 0.022*** 0.010** 0.022*** 0.007 0.020*** (1.786) (4.245) (1.978) (4.084) (1.437) (3.799) Rating_Indi,t 0.059*** 0.118*** 0.060*** 0.119*** 0.059*** 0.119*** (6.363) (9.082) (6.251) (9.171) (6.264) (9.036) Tangibilityi,t 0.089*** -0.002 0.098*** 0.005 0.083** -0.007 (2.690) (-0.040) (2.966) (0.113) (2.507) (-0.185) ROAi,t -0.725*** -0.547*** -0.735*** -0.566*** -0.734*** -0.562*** (-13.306) (-7.035) (-13.635) (-7.181) (-13.108) (-6.921) R&Di,t -0.385*** -0.576*** -0.419*** -0.593*** -0.370*** -0.558*** (-5.587) (-5.624) (-5.712) (-5.686) (-5.294) (-5.336) Advertisingi,t -0.182* -0.064 -0.182* -0.075 -0.169 -0.052 (-1.709) (-0.314) (-1.685) (-0.359) (-1.569) (-0.254) MTBi,t -0.053*** 0.000 -0.053*** -0.000 -0.053*** 0.000 (-7.530) (0.039) (-7.355) (-0.002) (-7.477) (0.049) σ(CF)i,t 0.398*** 0.453*** 0.388*** 0.473*** 0.425*** 0.489*** (3.286) (2.991) (3.009) (2.982) (3.467) (3.220) Depreciationi,t -0.318 0.621* -0.320 0.639* -0.267 0.678** (-1.177) (1.933) (-1.173) (1.956) (-0.972) (2.084) Ind. & Year FE? Y Y Y Y Y Y Observations 9,449 9,449 9,449 9,449 9,449 9,449 R-squared 0.537 0.315 0.534 0.309 0.533 0.309
46
Table 4 H2: Demand for Domestic Cash (cont’d.)
Panel B: Domestic Employment and Compensation – Test of Domestic Liabilities and Tax-Induced Foreign Cash This table presents results of tests of H2. Panel B presents estimated coefficients and t-statistics from the following pooled domestic leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Dom_Liabi,t = a + β1*TIFCi,t + β2*Dom_Employmenti,t + β3*TIFCi,t*Dom_Employmenti,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9*Div_Indicatori,t + β10*R&Di,t +
β11*Advertisingi,t + β12*MTBi,t + β13*σ(CF)i,t+ β14*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1) and (3) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2) and (4) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), Dom_Employeesi,t is an indicator variable equal to one if the firm’s ratio of domestic employees to total worldwide employees in year t is in the top quartile of the sample, and zero otherwise. In Col. (3)-(4), Dom_Compensationi,t is an indicator variable equal to one if the firm’s ratio of domestic compensation to total worldwide compensation in year t is in the top quartile of the sample, and zero otherwise. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Dom_Employmenti,t Dom_Employeesi,t Dom_Compensationi,t Dependent Variable: Dom_Mkt_Liabi,t
(1) Dom_Book_Liabi,t
(2) Dom_Mkt_Liabi,t
(3) Dom_Book_Liabi,t
(4) TIFCi,t -0.026*** -0.044*** -0.028*** -0.045*** (-3.775) (-4.483) (-3.594) (-4.234) Dom_Employmenti,t 0.038*** 0.038*** 0.030*** 0.029*** (4.895) (3.271) (4.045) (2.696) TIFCi,t*Dom_Employmenti,t -0.015 -0.057** -0.008 -0.060** (-0.732) (-2.060) (-0.472) (-2.564) PredForCash-Controlsi,t -0.106*** -0.094*** -0.106*** -0.094*** (-9.473) (-5.929) (-9.372) (-5.868) Sizei,t 0.010** 0.022*** 0.009* 0.021*** (1.974) (4.051) (1.931) (4.010) Rating_Indi,t 0.060*** 0.119*** 0.060*** 0.119*** (6.280) (9.210) (6.248) (9.158) Tangibilityi,t 0.092*** -0.001 0.093*** -0.000 (2.800) (-0.037) (2.787) (-0.007) ROAi,t -0.735*** -0.566*** -0.735*** -0.567*** (-13.432) (-7.110) (-13.539) (-7.149) Div_Indicatori,t -0.046*** -0.033*** -0.046*** -0.033*** (-5.698) (-2.904) (-5.677) (-2.840) R&Di,t -0.427*** -0.594*** -0.437*** -0.595*** (-5.886) (-5.713) (-6.028) (-5.709) Advertisingi,t -0.187* -0.074 -0.185* -0.074 (-1.760) (-0.360) (-1.740) (-0.360) MTBi,t -0.054*** -0.000 -0.054*** -0.000 (-7.273) (-0.040) (-7.322) (-0.003) σ(CF)i,t 0.374*** 0.457*** 0.374*** 0.454*** (2.944) (2.945) (2.971) (2.937) Depreciationi,t -0.274 0.682** -0.282 0.661** (-0.998) (2.105) (-1.023) (2.044) Industry and Year FE? Y Y Y Y Observations 9,449 9,449 9,449 9,449 R-squared 0.538 0.311 0.536 0.310
47
Table 4 H2: Demand for Domestic Cash (cont’d.)
Panel C: Domestic Underinvestment – First-Stage Analysis This table presents results of tests of H2. Panel C presents estimated coefficients and t-statistics from the first-stage pooled domestic investment model following Harford et al. (2017):
Dom_Investmenti,t-1 = a + β1*Dom_Sales_Growthi,t-2 + β2*For_Sales_Growthi,t-2 + β3*Dom_ROSi,t-2 + β4*Fom_ROSi,t-2 + β5*Dom_Sizei,t-2 + β6*For_Sizei,t-2 + IndustryFE + YearFE
The dependent variable is domestic capital expenditures scaled by total domestic assets. We define the indicator variable Underinvesti,t-1 based on the residual from this model. Specifically, Underinvesti,t-1 is an indicator variable equal to one if the firm has a negative residual or is in the bottom quartile of the sample, and zero otherwise. Due to missing data required to estimate the domestic investment model, Panel C estimates results for a sample of 4,811 U.S. MNC firm-year observations. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively. Dependent Var: Dom_Investmenti,t-1 Dom_Sales_Growthi,t-2 0.002 (1.127) For_Sales_Growthi,t-2 -0.000 (-0.643) Dom_ROSi,t-2 0.003 (1.183) For_ROSi,t-2 0.001 (0.263) Dom_Sizei,t-2 0.003*** (3.923) For_Sizei,t-2 -0.002*** (-3.721) Industry and Year FE? Y Observations 4,811 R-squared 0.147
Table 4 H2: Demand for Domestic Cash (cont’d.)
Panel D: Domestic Underinvestment – Test of Domestic Liabilities and Tax-Induced Foreign Cash This table presents results of tests of H2. Panel D presents estimated coefficients and t-statistics from the following pooled domestic leverage model:
Dom_Liabi,t = a + β1*TIFCi,t + β2*Underinvesti,t + β3*TIFCi,t*Underinvesti,t-1 + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9*Div_Indicatori,t + β10*R&Di,t + β11*Advertisingi,t +
β12*MTBi,t + β13*σ(CF)i,t+ β14*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1) and (3) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2) and (4) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities scaled by the sum of total domestic liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), Underinvesti,t-1 is an indicator variable equal to one if the residual from the domestic investment model presented in Panel C is negative, and zero otherwise. In Col. (3)-(4), Underinvesti,t-1 is an indicator variable equal to one if the residual from the domestic investment model is in the bottom quartile of the sample, and zero otherwise. Due to missing data required to estimate the domestic investment model, we estimate results for a sample of 4,811 U.S. MNC firm-year observations. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Underinvesti,t-1: Residual < 0 Residual < p25 Dependent Variable: Dom_Mkt_Liabi,t
(1) Dom_Book_Liabi,t
(2) Dom_Mkt_Liabi,t
(3) Dom_Book_Liabi,t
(4) TIFCi,t -0.043*** -0.082*** -0.027** -0.046*** (-2.717) (-4.370) (-2.177) (-2.657) Underinvesti,t-1 0.016* 0.007 0.017* 0.009 (1.664) (0.534) (1.698) (0.724) TIFCi,t*Underinvesti,t-1 0.019 0.046** -0.014 -0.023 (1.158) (2.208) (-0.791) (-0.707) PredForCash-Controlsi,t -0.129*** -0.110*** -0.128*** -0.110*** (-6.407) (-4.122) (-6.297) (-4.036) Sizei,t 0.013*** 0.026*** 0.013*** 0.025*** (2.670) (3.977) (2.666) (3.938) Rating_Indi,t 0.054*** 0.116*** 0.054*** 0.117*** (4.641) (7.424) (4.755) (7.421) Tangibilityi,t 0.082** -0.018 0.078** -0.023 (2.322) (-0.399) (2.156) (-0.486) ROAi,t -0.713*** -0.529*** -0.714*** -0.530*** (-10.430) (-5.889) (-10.678) (-5.988) Div_Indicatori,t -0.049*** -0.040*** -0.048*** -0.040*** (-4.897) (-2.808) (-4.826) (-2.781) R&Di,t -0.375*** -0.654*** -0.370*** -0.641*** (-4.756) (-4.666) (-4.706) (-4.557) Advertisingi,t -0.190 0.036 -0.192 0.028 (-1.376) (0.155) (-1.404) (0.122) MTBi,t -0.061*** 0.005 -0.061*** 0.005 (-8.076) (0.421) (-8.080) (0.367) σ(CF)i,t 0.518*** 0.562*** 0.506*** 0.556*** (3.664) (3.002) (3.575) (2.996) Depreciationi,t 0.015 0.849** -0.036 0.791** (0.051) (2.246) (-0.126) (2.158) Industry and Year FE? Y Y Y Y Observations 4,811 4,811 4,811 4,811 R-squared 0.541 0.288 0.540 0.287
Table 5 Falsification Test: Low Cost of Debt
This table presents estimated coefficients and t-statistics from the following pooled domestic leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Dom_Liabi,t = a + β1*TIFCi,t + β2*LowDebtCosti,t + β3*TIFCi,t*LowDebtCosti,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Tangibilityi,t + β7*ROAi,t + β8* R&Di,t +β9* R&Di,t + β10*Advertisingi,t +
β11*MTBi,t + β12*σ(CF)i,t+ β13*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1), (3), and (5) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2), (4), and (6) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities (using BEA data) scaled by the sum of total domestic liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), LowDebtCosti,t is Rating_Ind, an indicator variable equal to one if the company has a bond rating, and zero otherwise. In Col. (3)-(4), LowDebtCosti,t is Rating_Cat, a categorical variable equal to 1, 2, or 3 if the firm’s bond rating is in the C range, B range, or A range, respectively in year t. In Col. (5)-(6), LowDebtCosti,t is Whited_Wu, defined as an indicator equal to one if the firm is in the bottom quartile based on the Whited-Wu index. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Cost of Debt: Rating_Ind Rating_Cat Whited_Wu Dependent Variable: Dom_Mkt_Liabi,t
(1) Dom_Book_Liabi,t
(2) Dom_Mkt_Liabi,t
(3) Dom_Book_Liabi,t
(4) Dom_Mkt_Liabi,t
(5) Dom_Book_Liabi,t
(6)
TIFCi,t -0.017* -0.046*** -0.027*** -0.048*** -0.040*** -0.055*** (-1.792) (-3.224) (-2.754) (-3.298) (-4.907) (-4.375) LowDebtCosti,t 0.066*** 0.124*** 0.014*** 0.043*** -0.017* -0.014 (6.817) (9.295) (3.457) (7.276) (-1.908) (-1.025) TIFCi,t*LowDebtCosti,t -0.035*** -0.022 -0.007 -0.008 0.017 -0.013 (-2.752) (-1.138) (-1.376) (-1.052) (1.263) (-0.576) PredForCash-Controlsi,t -0.106*** -0.094*** -0.111*** -0.099*** -0.114*** -0.110*** (-9.257) (-5.775) (-9.190) (-5.823) (-9.022) (-5.980) Sizei,t 0.009* 0.021*** 0.014*** 0.023*** 0.023*** 0.045*** (1.920) (4.006) (2.829) (4.131) (4.969) (8.434) Tangibilityi,t 0.098*** 0.004 0.107*** 0.016 0.117*** 0.039 (2.933) (0.105) (3.220) (0.386) (3.561) (0.956) ROAi,t -0.735*** -0.567*** -0.752*** -0.592*** -0.761*** -0.617*** (-13.737) (-7.205) (-13.506) (-7.363) (-13.443) (-7.444) Div_Indicatori,t -0.045*** -0.032*** -0.047*** -0.037*** -0.041*** -0.028** (-5.377) (-2.721) (-5.522) (-3.202) (-4.536) (-2.172) R&Di,t -0.419*** -0.591*** -0.414*** -0.598*** -0.395*** -0.551*** (-5.781) (-5.703) (-5.470) (-5.602) (-5.188) (-4.969) Advertisingi,t -0.161 -0.050 -0.151 -0.045 -0.127 0.040 (-1.513) (-0.242) (-1.373) (-0.215) (-1.131) (0.191) MTBi,t -0.053*** 0.000 -0.054*** -0.002 -0.053*** 0.000 (-7.381) (0.028) (-7.243) (-0.183) (-7.135) (0.026) σ(CF)i,t 0.384*** 0.467*** 0.372*** 0.471*** 0.349*** 0.393** (3.007) (2.967) (2.909) (2.987) (2.761) (2.464) Depreciationi,t -0.320 0.641* -0.360 0.590* -0.397 0.500 (-1.161) (1.958) (-1.297) (1.775) (-1.432) (1.495) Industry and Year FE? Y Y Y Y Y Y Observations 9,449 9,449 9,449 9,449 9,449 9,449 R-squared 0.534 0.309 0.527 0.297 0.525 0.279
Table 6 Worldwide Liabilities & Foreign Liabilities
Panel A: Worldwide Liabilities and Tax-Induced Foreign Cash This table examines worldwide and foreign liabilities of firms that exhibit higher domestic liabilities. Panel A presents estimated coefficients and t-statistics from the following pooled worldwide leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Worldwide_Liabi,t = a + β1*TIFCi,t + β2*Payouti,t + β3*TIFCi,t*Payouti,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9* R&Di,t + β10*Advertisingi,t + β11*MTBi,t + β12*σ(CF)i,t+
β13*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1), (3), and (5) is worldwide market leverage Mkt_Liabi,t, measured as total worldwide liabilities (using Compustat data) scaled by the sum of total worldwide liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2), (4), and (6) is worldwide book leverage Book_Liabi,t, measured as total worldwide liabilities (using Compustat data) scaled by the sum of total worldwide liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), Payouti,t is an indicator variable equal to one if the firm either pays dividends or repurchases shares in year t, and zero otherwise. In Col. (3)-(4), Payouti,t is an indicator variable equal to one if the firm pays dividends in year t, and zero otherwise. In Col. (5)-(6), Payouti,t is an indicator variable equal to one if the firm repurchases shares in year t, and zero otherwise. We exclude Div_Indicatori,t due to collinearity with Payouti,t. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Payouti,t = 1 if: Dividend or Repurchase Dividend Repurchase Dependent Variable: Mkt_Liabi,t
(1) Book_Liabi,t
(2) Mkt_Liabi,t
(3) Book_Liabi,t
(4) Mkt_Liabi,t
(5) Book_Liabi,t
(6) TIFCi,t -0.057*** -0.070*** -0.039*** -0.053*** -0.054*** -0.053*** (-3.708) (-3.550) (-3.440) (-3.963) (-4.611) (-3.304) Payouti,t -0.082*** -0.068*** -0.067*** -0.044*** -0.058*** -0.037*** (-10.225) (-5.801) (-8.752) (-4.431) (-9.301) (-4.167) TIFCi,t*Payouti,t 0.050*** 0.058*** 0.039*** 0.060*** 0.054*** 0.042** (3.388) (2.886) (2.966) (3.116) (4.569) (2.375) PredForCash-Controlsi,t -0.097*** -0.083*** -0.097*** -0.083*** -0.095*** -0.081*** (-7.580) (-5.917) (-7.589) (-5.843) (-7.231) (-5.698) Sizei,t 0.008** 0.022*** 0.009** 0.022*** 0.006 0.020*** (1.971) (4.796) (2.358) (4.745) (1.477) (4.356) Rating_Indi,t 0.057*** 0.106*** 0.058*** 0.106*** 0.057*** 0.106*** (7.303) (9.297) (7.460) (9.378) (7.258) (9.233) Tangibilityi,t 0.074*** -0.009 0.086*** -0.003 0.065*** -0.015 (3.090) (-0.293) (3.631) (-0.082) (2.702) (-0.463) ROAi,t -0.442*** -0.512*** -0.450*** -0.525*** -0.449*** -0.524*** (-6.839) (-6.991) (-6.848) (-7.120) (-6.785) (-6.900) R&Di,t -0.333*** -0.627*** -0.382*** -0.648*** -0.313*** -0.609*** (-4.444) (-6.242) (-4.811) (-6.393) (-4.163) (-5.988) Advertisingi,t -0.276*** -0.126 -0.280*** -0.142 -0.256*** -0.112 (-2.864) (-0.700) (-2.929) (-0.770) (-2.615) (-0.619) MTBi,t -0.091*** -0.005 -0.091*** -0.005 -0.090*** -0.005 (-8.661) (-0.547) (-8.445) (-0.599) (-8.657) (-0.524) σ(CF)i,t 0.379*** 0.361** 0.358*** 0.373** 0.407*** 0.391*** (3.555) (2.515) (3.118) (2.509) (3.770) (2.732) Depreciationi,t -0.302 0.465* -0.313 0.472* -0.240 0.514* (-1.337) (1.722) (-1.373) (1.736) (-1.028) (1.889) Industry and Year FE? Y Y Y Y Y Y Observations 9,449 9,449 9,449 9,449 9,449 9,449 R-squared 0.623 0.350 0.621 0.345 0.618 0.344
51
Table 6 (cont’d). Worldwide Liabilities & Foreign Liabilities
Panel B: Foreign Liabilities and Tax-Induced Foreign Cash This table examines worldwide and foreign liabilities of firms that exhibit higher domestic liabilities. Panel B presents estimated coefficients and t-statistics from the following pooled foreign leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Foreign_Liabi,t = a + β1*TIFCi,t + β2*Payouti,t + β3*TIFCi,t*Payouti,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9* R&Di,t + β10*Advertisingi,t + β11*MTBi,t +
β12*σ(CF)i,t+ β13*Depreciationi,t + IndustryFE + YearFE The dependent variable in Col. (1), (3), and (5) is foreign market leverage For_Mkt_Liabi,t, measured as total foreign liabilities (using Compustat data) scaled by the sum of total foreign liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2), (4), and (6) is foreign book leverage Book_Liabi,t, measured as total foreign liabilities (using Compustat data) scaled by the sum of total foreign liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. In Col. (1)-(2), Payouti,t is an indicator variable equal to one if the firm either pays dividends or repurchases shares in year t, and zero otherwise. In Col. (3)-(4), Payouti,t is an indicator variable equal to one if the firm pays dividends in year t, and zero otherwise. In Col. (5)-(6), Payouti,t is an indicator variable equal to one if the firm repurchases shares in year t, and zero otherwise. We exclude Div_Indicatori,t due to collinearity with Payouti,t. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Payouti,t = 1 if: Dividend or Repurchase Dividend Repurchase Dependent Variable: For_Mkt_Liabi,t
(1) For_Bk_Liabi,t
(2) For_Mkt_Liabi,t
(3) For_Bk_Liabi,t
(4) For_Mkt_Liabi,t
(5) For_Bk_Liabi,t
(6) TIFCi,t -0.001 0.012 0.006 0.011 0.005 0.007 (-0.075) (0.640) (0.843) (1.050) (0.394) (0.473) Payouti,t -0.040*** -0.022** -0.036*** -0.024*** -0.015** -0.011* (-4.509) (-2.235) (-4.577) (-3.636) (-2.363) (-1.683) TIFCi,t*Payouti,t 0.029 0.009 0.030*** 0.016 0.025* 0.019 (1.531) (0.514) (2.863) (1.143) (1.707) (1.315) PredForCash-Controlsi,t -0.030*** -0.008 -0.030*** -0.008 -0.029*** -0.007 (-4.382) (-1.230) (-4.463) (-1.281) (-4.072) (-1.132) Sizei,t -0.022*** -0.027*** -0.021*** -0.026*** -0.024*** -0.028*** (-6.314) (-10.252) (-6.401) (-9.937) (-6.330) (-10.581) Rating_Indi,t -0.005 0.006 -0.004 0.006 -0.004 0.006 (-0.776) (0.981) (-0.710) (1.027) (-0.674) (1.049) Tangibilityi,t 0.045** 0.055*** 0.051** 0.060*** 0.042* 0.054*** (2.132) (2.738) (2.350) (2.975) (1.952) (2.628) ROAi,t -0.327*** -0.173*** -0.329*** -0.172*** -0.338*** -0.178*** (-8.745) (-4.402) (-8.688) (-4.165) (-8.428) (-4.347) R&Di,t -0.221*** -0.214*** -0.246*** -0.234*** -0.212*** -0.211*** (-4.598) (-3.733) (-4.631) (-3.852) (-4.366) (-3.696) Advertisingi,t -0.019 -0.060 -0.025 -0.062 -0.018 -0.059 (-0.291) (-0.665) (-0.380) (-0.693) (-0.273) (-0.676) MTBi,t -0.009*** -0.004 -0.009*** -0.004 -0.009*** -0.004 (-3.752) (-1.539) (-3.751) (-1.500) (-3.656) (-1.544) σ(CF)i,t 0.349*** 0.327*** 0.336*** 0.313*** 0.372*** 0.338*** (3.742) (2.868) (3.475) (2.664) (4.078) (2.977) Depreciationi,t 0.114 0.098 0.108 0.090 0.145 0.115 (0.737) (0.640) (0.691) (0.586) (0.922) (0.748) Industry and Year FE? Y Y Y Y Y Y Year fixed effects? Y Y Y Y Y Y Observations 7,633 7,633 7,633 7,633 7,633 7,633 R-squared 0.267 0.116 0.267 0.117 0.260 0.115
52
Table 7 High R&D Firms
This table presents results for testing the relation between tax-induced foreign cash and domestic borrowings for firms with high R&D. We present estimated coefficients and t-statistics from the following pooled domestic leverage model using the full sample of 9,449 U.S. MNC firm-year observations for 1999 through 2012:
Dom_Liabi,t = a + β1*TIFCi,t + β2*HighR&Di,t + β3*TIFCi,t*HighR&Di,t + β4*PredForCash-Controlsi,t + β5*Sizei,t + β6*Rating_Indi,t + β7*Tangibilityi,t + β8*ROAi,t + β9*Div_Indicatori,t + β10*R&Di,t +
β11*Advertisingi,t + β12*MTBi,t + β13*σ(CF)i,t+ β14*Depreciationi,t + YearFE The dependent variable in Col. (1) is domestic market leverage Dom_Mkt_Liabi,t, measured as total domestic liabilities scaled by the sum of total domestic liabilities and the market value of equity (CSHO*PRCC_F). The dependent variable in Col. (2) is domestic book leverage Dom_Book_Liabi,t, measured as total domestic liabilities scaled by the sum of total domestic liabilities and the book value of equity (SEQ). TIFCi,t is estimated tax-induced foreign cash following Foley et al. (2007) and Hanlon et al. (2015), presented in Appendix B. HighR&Di,t is an indicator variable equal to one if the firm is in a high technology industry, and zero otherwise. All variables are defined in Appendix A. The regression specification includes industry and year fixed effects. t-statistics are presented in parentheses. Standard errors are clustered by firm and year. *, **, and *** indicate two-tailed statistical significance at the 10%, 5%, and 1% levels, respectively.
Dependent Variable: Dom_Mkt_Liabi,t (1)
Dom_Book_Liabi,t (2)
TIFCi,t -0.045*** -0.068*** (-4.535) (-4.929) HighR&Di,t -0.003 -0.015 (-0.370) (-1.056) TIFCi,t*HighR&Di,t 0.041*** 0.044** (2.651) (2.047) PredForCash-Controlsi,t -0.073*** -0.074*** (-9.183) (-8.193) Sizei,t 0.009* 0.019*** (1.850) (3.746) Rating_Indi,t 0.065*** 0.124*** (6.499) (9.310) Tangibilityi,t 0.048 -0.048 (1.351) (-1.114) ROAi,t -0.758*** -0.584*** (-15.213) (-7.535) Div_Indicatori,t -0.035*** -0.023** (-4.452) (-2.047) R&Di,t -0.462*** -0.567*** (-7.074) (-5.870) Advertisingi,t -0.194* -0.033 (-1.806) (-0.160) MTBi,t -0.058*** -0.002 (-7.617) (-0.230) σ(CF)i,t 0.236** 0.309** (2.037) (2.082) Depreciationi,t -0.382 0.644** (-1.382) (1.997) Industry and Year FE? Y Y Observations 9,449 9,449 R-squared 0.513 0.296