Early Evidence on the Use of Foreign Cash Following the Tax Cuts and Jobs Act of 2017

Brooke Beyer
Kansas State University
Department of Accounting
Manhattan, KS USA
ORCID 0000-0003-3424-0179

Jimmy F. Downes*
University of Nebraska-Lincoln
School of Accountancy
Lincoln, NE USA
ORCID 0000-0002-7776-9593

Mollie E. Mathis
Auburn University
School of Accountancy
Auburn, AL USA
ORCID 0000-0001-5492-5541

Eric T. Rapley
Colorado State University
Department of Accounting
Fort Collins, CO USA
ORCID 0000-0003-2374-6326

September 2020

*
Corresponding author: 445 H College of Business, P.O. Box 880488, Lincoln, NE 68588-0488;
+1 (402) 472-5152; downes@unl.edu.

Acknowledgments: We are grateful for helpful comments and suggestions from Harald Amberger, T.J. Atwood,
James Brushwood, Michelle Draeger, Jeffrey Gramlich, Robert Gutsche, Amy Hageman, Kim Key, Sandeep Nabar,
Tom Omer, Bill Schwartz, Terry Shevlin, Jim Stekelberg, Wayne Thomas and Arndt Weinrich. We appreciate the
feedback from workshop participants at Auburn University, Colorado State University, Oklahoma State University,
the 2019 AAA Annual Meeting and the 2020 Hawaiian Accounting Research Conference. We would also like to
thank a manager in the national office of a Big 4 accounting firm for their helpful consultation as well as Madison
May for her research assistance. Brooke Beyer is thankful for the summer research support grant provide by the
Center for Principled Business Faculty Fellows Program at Kansas State University.
 Early Evidence on the Use of Foreign Cash Following the Tax Cuts and Jobs Act of 2017

ABSTRACT: The Tax Cuts and Jobs Act of 2017 (TCJA) reduces multinational firms’ internal
capital market frictions related to repatriation costs. This change to the U.S. corporate tax system
provides an opportunity to examine how multinational firms’ access to lower cost internal capital
(i.e., foreign cash) affects spending and investment behavior. This investigation is pertinent
because two of the TCJA’s goals are repatriation of overseas income and economic growth. We
provide evidence that suggests easing access to foreign cash may not result in domestic economic
growth. Specifically, we find firms with more pre-TCJA foreign cash increase their post-TCJA
repurchases but do not change spending behavior related to shareholder dividends, capital
expenditures, or outstanding debt, on average. The increase in share repurchases is driven by
multinational firms with greater financial constraints and higher pre-TCJA repatriation costs.
Finally, we investigate whether the global intangible low-taxed income (GILTI) inclusion, a new
provision included in the TCJA, creates internal capital market frictions. For multinational firms
with more foreign cash who are likely subject to the GILTI inclusion regime, we find an increase
in foreign but not domestic capital expenditures, highlighting a potential unintended consequence
and new internal capital market friction created by the TCJA.

Keywords: Tax Cuts and Jobs Act; payout policy; capital investment; internal capital market

JEL Classifications: F23, G31, G38, H25, M40, M48.

Data Availability: Data are available from the public sources cited in the text.

i
 Early Evidence on the Use of Foreign Cash Following the Tax Cuts and Jobs Act of 2017

1. INTRODUCTION

The Tax Cuts and Jobs Act of 2017 (TCJA) dramatically changed the United States (U.S.)

corporate tax landscape. Before the TCJA, if multinational firms repatriated earnings from a

foreign subsidiary to the U.S. parent, they were required to pay taxes on the difference between

the U.S. statutory tax rate and the effective foreign tax rate when the U.S. tax rate was higher.

This repatriation tax created internal capital market frictions between foreign subsidiaries and the

U.S. parent (De Simone and Lester, 2018); many firms indefinitely deferred payment of

repatriation taxes and consequently held large amounts of cash overseas (Foley, Hartzell,

Titman, and Twite 2007). Among several sweeping provisions, the TCJA required a deemed

repatriation of unremitted foreign earnings (UFE) and eliminated future federal repatriation taxes

for U.S. multinational firms. 1 The intent of this change is to decrease internal capital market

frictions and allow foreign cash to be more easily repatriated to the domestic parent (Garber,

2017). 2 The recent passing of the TCJA provides a salient setting to investigate the effect of

internal capital market frictions on firm spending and investment behavior.

The literature provides evidence that repatriation taxes lead to an increase in cash held by

foreign subsidiaries (Foley et al., 2007), resulting in a higher likelihood of managers using

foreign cash for value-destroying foreign investments (Hanlon, Lester, and Verdi, 2015;

Edwards, Kravet, and Wilson, 2016), less efficient investment (Amberger, Markle, and Samuel,

2020), fewer shareholder payouts and higher levels of abnormal debt (Beyer, Downes, and

1
See Section 2 for additional details.
2
Foreign cash is related to, but not equivalent to permanently reinvested earnings (PRE) or UFE. Laplante and
Nesbit (2017) discuss the differences between foreign cash, PRE and UFE; Blaylock, Downes, Mathis, and White
(2019) document the high correlation (>90%) between PRE and estimated UFE. We choose to examine foreign cash
because it is often held overseas for tax reasons (Foley et al., 2007) and it is the most liquid asset that is likely to be
used for investment purposes in the post-TCJA setting. Foreign cash disclosures are also the most reliable, publicly
available estimates of firms’ foreign cash balances.

1
Rapley, 2017; Nessa, 2017; De Simone and Lester, 2018). Based on internal capital market

theory, if firms have domestic capital investment opportunities but face financial constraints (i.e.,

limited internal funds and costly access to external funds), a decrease in internal capital market

frictions from reduced repatriation costs should increase their domestic capital investment. 3

Without capital investment opportunities, a lower cost of internal capital should increase firms’

ability and/or willingness to payout cash to shareholders and/or reduce outstanding debt.

Opponents of the TCJA argue that the repatriated earnings would have a significant effect on

shareholder payouts, but not domestic capital investment (Bloomberg 2017). For example,

economist Kyle Pomerleau stated: “A company could get an extra couple of billion from

overseas, but that doesn’t change their willingness to invest the cash. Most analysts, myself

included, assumed that repatriation would provide no boost in investments” (Davis and Chandra

2018). Relatedly, S&P 500 dividends and stock repurchases were the highest they have ever been

during the fourth quarter of 2018; these firms’ shareholder payouts totaled a record-high $1.26

trillion during 2018 (PR Newswire 2019). De Simone and Lester (2018) provide evidence that

firms with tax-induced foreign cash may substitute debt issuance for repatriation, indicating that

U.S. firms with tax-induced foreign cash do not forgo domestic investment opportunities if they

are not financially constrained. Consistent with these expectations, prior research documents that

the American Jobs Creation Act (AJCA) tax holiday resulted in repatriations by firms with

limited investment opportunities, and these firms increased payouts to shareholders, but only

marginally increased domestic investment (Blouin and Krull, 2009; Dharmapala, Foley, and

3
Increases in domestic investment as a result of reductions in repatriation taxes first requires firms to repatriate
foreign cash to the domestic parent. Bureau of Economic and Analysis data suggest that post-TCJA foreign cash
repatriations significantly increased compared to the pre-TCJA period for multinational firms
(https://www.bea.gov/news/2019/us-international-transactions-first-quarter-2019-and-annual-update, last accessed
September 1, 2020). Specifically, in the first quarter of 2018 reinvested earnings in foreign affiliates decreased
$151.6 billion compared to increasing $76.5 billion in the first quarter of 2017. Similarly, dividends from foreign
subsidiaries to the domestic parent increased by $247.7 billion for the same time periods compared.

2
Forbes, 2011). According to the AJCA tax holiday literature, if firms do not have domestic

investment opportunities and/or are not financially constrained, we would not expect a change to

domestic investment, and instead expect an increase in shareholder payouts following the TCJA.

However, it is unclear whether the TCJA would lead to the same outcome as the AJCA

because the TCJA differs from the AJCA in both timing and scope. The TCJA is permanent

whereas the AJCA was temporary; firms’ choice of payout method may be affected by whether

additional cash flow is perceived as permanent or temporary (Jagannathan, Stephens, and

Weisbach, 2000). Additionally, to convert the U.S. tax system from a modified worldwide tax

system to a quasi-territorial tax system, the TCJA required a deemed repatriation of all

unremitted foreign earnings. The deemed repatriation eliminated the permanent deferral option

while the AJCA did not; prior literature shows the permanent deferral option led some firms to

invest in negative net present value foreign financial assets (De Waegenaere and Sansing, 2008).

Therefore, the TCJA reduced the benefits of this investment strategy, and firms will likely

reevaluate their investment opportunities in light of this change.

Unlike the AJCA, the TCJA also includes several provisions which could result in the

unintended consequence of incentivizing multinational firms to invest in foreign assets. First, the

TCJA introduces a global intangible low-taxed income (GILTI) inclusion to discourage income

shifting from the U.S. to foreign jurisdictions and/or among foreign jurisdictions. In addition to

the GILTI inclusion, the TCJA also creates the foreign-derived intangible income (FDII)

deduction, which incentivizes U.S. firms to export products and services to foreign markets and

maintain ownership of intellectual property in the U.S. Importantly, both GILTI and FDII proxy

for intangible income using a return on foreign and domestic tangible assets, respectively. Thus,

3
to minimize overall taxes, firms can minimize their GILTI inclusion and maximize their FDII

deduction by increasing foreign and reducing domestic investment in tangible assets.

We examine our research question using the TCJA as a proxy for the permanent reduction in

repatriation costs causing an exogenous shock to firms’ internal capital markets. We examine

changes in firm spending and investment behavior by looking at the change in shareholder

payouts (i.e., dividends and repurchases), capital expenditures and debt reductions following

TCJA. 4 Two samples are used to test our research question. The first is a sample consisting of

both multinational and domestic firms. Given that the TCJA resulted in many broad changes to

the corporate tax landscape, using a sample of multinational and domestic firms allows us to use

a difference-in-differences (DiD) design that isolates the changes related to multinational firms’

foreign cash. Using this sample comprised of 4,940 firm-year observations for the period 2015 to

2019, we find no incremental change in multinational firms’ spending and investment behavior

compared to domestic firms, on average. 5 However, we also document that, compared to

domestic-only firms, financially constrained multinational firms increased repurchases rather

than dividends. This provides evidence that the permanent reduction in repatriation taxes allowed

financially constrained multinational firms to increase their shareholder payouts with the more

flexible form of payouts (i.e., repurchases). We also find that financially constrained

multinational firms did not change total capital expenditures in the post-TCJA period.

The second sample is limited to only multinational firms and is comprised of 1,744 firm-year

observations for the period 2015 to 2019. Using this sample, we are able to incorporate publicly

disclosed amounts of multinational firms’ foreign cash balances into the analysis, which is

4
In untabulated analysis we examine research and development expenditures, but do not find post-TCJA changes.
5
The year 2017 is omitted because it includes cash spending decisions from both the pre- and post-TCJA time
period. Inferences are similar if we include 2017 data.

4
advantageous because UFE held in cash does not face the same liquidation costs as UFE held in

operating assets (Slemrod 1992), making foreign cash more readily available for shareholder

payouts, investment purposes and/or debt reduction in the post-TCJA setting. We document an

increase in post-TCJA repurchases for firms with greater levels of pre-TCJA foreign cash. The

increase in post-TCJA repurchases is strongest among firms with greater pre-TCJA repatriation

costs, consistent with the notion that multinational firms are accessing previously trapped foreign

cash in order to increase shareholder repurchases.

On average, we find that the level of pre-TCJA foreign cash does not affect multinational

firms’ capital expenditure behavior. This finding suggests that multinational firms were not

forgoing domestic investment opportunities pre-TCJA. There are two potential explanations for

this finding. First, consistent with prior literature (Blouin and Krull, 2009; Dharmapala et al.,

2011), multinational firms’ investment opportunity set did not change. As outlined above, the

deemed repatriation and elimination of the permanent deferral option makes this explanation

unlikely. Second, provisions within the TCJA incentivized foreign capital expenditures and

disincentivized domestic capital expenditures for multinational firms. Therefore, our cross-

sectional results may reflect conflicting investment incentives. We explore this alternative

explanation by separately examining post-TCJA changes in foreign and domestic capital

expenditures. As demonstrated in Appendix A, more foreign (less domestic) tangible assets

decreases the GILTI inclusion (increases the FDII deduction). Therefore, the GILTI inclusion

incentivizes investment on foreign tangible assets and the FDII deduction disincentivizes

investment in domestic tangible assets; whereas, bonus depreciation and Sec. 179 expensing

incentivize investment in domestic tangible assets. We hand-collect required segment disclosures

of net property, plant, and equipment for domestic and foreign operations (see ASC 280-10-50-

5
41) and partition the sample based on high and low potential GILTI tax inclusions. 6 Our findings

show that, for multinational firms potentially subject to greater GILTI inclusions, higher levels

of pre-TCJA foreign cash are associated with increased post-TCJA foreign property, plant, and

equipment investments. We do not find a similar increase in domestic property, plant and

equipment. These findings suggest that while the TCJA changed investment opportunity sets for

multinational firms, confounding incentives created by the GILTI inclusion and FDII deduction

counteract domestic investment incentives and introduce additional internal capital market

frictions for U.S. multinational firms.

We make at least two contributions to the literature. First, Clemons and Shevlin (2016) argue

that policymakers only consider academic research when it is directly useful to those involved in

the policymaking process, and the authors argue the most effective way for research to affect tax

policy is to specifically discuss tax policy in research papers. This paper accomplishes both of

those objectives. The findings in our study highlight an unintended consequence of the GILTI

inclusion–firms with greater foreign cash balances increased foreign capital expenditures without

a corresponding increase in domestic capital expenditures. Given that domestic capital

expenditure is important to the TCJA’s stated objective of economic growth, we believe our

findings are particularly relevant to policymakers. 7

Second, this study builds on prior literature examining the spending and investment of

internal capital resulting from a reduction in internal capital market frictions. Prior literature used

the AJCA as a setting of temporary repatriation cost reduction and documented that, on average,

6
Our cross-sectional investigation focuses on the GILTI inclusion (and not FDII exclusion) due to data availability.
7
Related to investment’s role in economic growth, Steven Mnuchin, Secretary of the Treasury expressed the
following: “The TCJA incentivized investments in equipment and intellectual capital that will continue to improve
the productivity of American workers. When Americans are more productive, they are better compensated. Today’s
capital investments are the foundation of tomorrow’s additional wage gains.” (Mnuchin 2019)

6
firms increased shareholder payouts and did not increase domestic investment (Blouin and Krull,

2009; Dharmapala et al., 2011). The exception is Faulkender and Petersen (2012) who document

an increase in capital expenditures for financially constrained firms. Our evidence suggests

financially constrained firms use the enactment of TCJA as an opportunity to return cash to

shareholders using the most flexible payout strategy (i.e., increasing repurchases instead of

dividends) post-TCJA, potentially increasing their internal capital flexibility in the future. 8

However, our findings document no change in capital expenditures for the average multinational

firm. Although many provisions in the TCJA differ from those in the AJCA, the incremental

effect on multinational firms domestic spending appears to be similar for both pieces of tax

legislation.

2. THE TAX CUTS AND JOBS ACT OF 2017

President Trump and a Republican-controlled Congress turned their attention to tax reform in

the late stages of summer 2017. The House Ways and Means Committee released a draft bill on

November 2nd and approved it on November 9th. On November 16th, the House floor passed a

revised version and the Senate Finance Committee approved a version of the tax reform bill. The

Senate passed a revised bill on December 2nd. The final TCJA was passed in the House of

Representatives and Senate on December 19th and December 20th, respectively. President Trump

then signed the TCJA into law on December 22, 2017.

Major corporate tax changes under the TCJA fall into four categories: statutory tax rate

decrease, 100 percent bonus depreciation, interest deductibility, and foreign earnings taxation.

8
A contemporaneous study by Bennett, Thakor and Wang (2019) finds that share repurchases increase for firms
with high foreign profits post-TCJA. Our study is complimentary to Bennett et al. (2019), examining other types of
spending and investment.

7
First, the TCJA reduces the corporate statutory tax rate from a graduated maximum rate of 35

percent to a flat rate of 21 percent. Second, the TCJA allows firms to deduct 100 percent of

qualified capital expenditures as bonus depreciation and increases IRC Sec. 179 expensing to

$1,000,000. Previously, firms could deduct 50 percent of capital expenditures as bonus

depreciation with the ability to expense an additional $510,000 in qualifying IRC Sec. 179

depreciation. Third, the TCJA limits the deductibility of business interest expense to business

interest income plus 30 percent of adjusted taxable income. 9 Fourth, the TCJA moves the U.S.

from a modified worldwide tax system to a quasi-territorial tax system.

To transition the U.S. to a quasi-territorial tax system, the TCJA required a deemed

repatriation of UFE from specified foreign corporations (SFCs). 10 The deemed repatriation is

subject to a dividends received deduction that effectively reduces the tax rate to 15.5 percent for

foreign cash and 8 percent for other assets with the ability to utilize foreign tax credits (FTCs) to

reduce this tax liability, which is payable over eight years (Nevius 2017). Under the new, quasi-

territorial tax system, SFCs receive a 100 percent dividends received deduction for foreign

earnings, effectively eliminating U.S. federal repatriation taxes on those future earnings. 11

The reduced U.S. corporate tax rate decreases the benefits of shifting profits abroad as the

spread between the U.S. statutory rate and foreign tax rates decreases or disappears. Nonetheless,

a shift to a quasi-territorial tax system could increase profit shifting as foreign source income is

9
Internal Revenue Code (IRC) Sec. 163(j) defines the limitation on interest deductibility. Adjusted taxable income
is computed without allowable deductions for amortization, depreciation, depletion, or business interest expense.
Additionally, firms whose average gross receipts do not exceed $25 million for the three prior years are exempt from
the business interest expense limitation.
10
SFCs are controlled foreign corporations with a domestic corporation shareholder owning 10 percent or more of
the stock for its last tax year prior to January 1, 2018. U.S. multinational companies include as subpart F income
their pro-rata share of the greater of the SFCs accumulated post-1986 deferred foreign income determined as of
November 2, 2017 or December 31, 2017.
11
U.S. multinational firms could still have state repatriation taxes and foreign withholding taxes on remitted
earnings. As a result, some firms have significantly reduced rather than eliminated internal capital market frictions
with regard to repatriated funds.

8
exempt from home country taxation. However, the TCJA contains several provisions aimed at

limiting profit shifting. First, the TCJA introduces a GILTI inclusion to discourage income

shifting from the U.S. to foreign jurisdictions or among foreign jurisdictions. Before the TCJA,

Subpart F rules were intended to prevent U.S. firms from shifting income to low-tax foreign

jurisdictions. The GILTI inclusion is a much broader inclusion regime, requiring immediate

taxation of controlled foreign corporations (CFCs) foreign earnings. IRS Sec 951(A)(b) requires

U.S. firms to calculate “tested income” or “tested loss” of each CFC; firms then net these

amounts, arriving at “net CFC tested income.” Finally, to arrive at firms’ GILTI inclusion

amount, firms must calculate their deemed tangible income return for the year. Firms calculate

the deemed tangible income return as the excess of 10 percent of the aggregate of a shareholder’s

pro rata share of the qualified business asset investment for each of its CFCs for a tax year,

which is tangible property used in a trade or business of a CFC. Firms then receive a 50 percent

deduction of their net GILTI inclusion under IRC Sec. 250(a)(1)(B). Because of the complexity

involved in the GILTI calculation, we provide examples of calculating the GILTI inclusion in

Appendix A.

In addition to the GILTI inclusion, the TCJA also creates the FDII deduction, which

incentivizes U.S. firms to export products and services to foreign markets and maintain

ownership of intellectual property in the U.S. The FDII deduction is equal to 37.5 percent of

FDII for U.S. firms. The FDII deduction proxies for intangible income from intellectual property

located in the U.S. by allowing a deduction for income exceeding a 10 percent return on

domestic tangible assets rather than calculating intangible income from intellectual property

located in the U.S. directly. When firms increase their investment in tangible domestic assets,

they decrease their FDII deduction. Because of its complexity, we provide examples of FDII

9
deduction calculations in Appendix A. Importantly, both GILTI and FDII proxy for intangible

income rather than calculating these amounts directly. Thus, to minimize overall taxes, firms can

minimize their GILTI inclusion and maximize their FDII deduction by decreasing domestic

investment and/or increasing foreign investment in tangible assets. Additionally, the TCJA

introduces the BEAT, which requires a minimum tax for multinational firms making deductible

payments like royalties, interest expense and certain service payments to foreign related

corporations.

3. HYPOTHESIS DEVELOPMENT

3.1 Internal Capital Market and Domestic Investment

Firms with efficient internal capital markets create value by reallocating resources to

business segments unable to generate sufficient funding for investment opportunities (Weston,

1970; Williamson, 1975; Stein, 1997; Billett and Mauer, 2003). For example, firms can use cash

flow from one business segment or division to fund a capital project in another business segment

or division. Efficient internal capital markets allow financially constrained business segments to

access lower cost internal capital to fund positive net present value projects.

The literature documents several frictions that create inefficient internal capital markets:

internal agency costs, external agency costs, and tax-related repatriation costs. First, Desai,

Foley, and Hines (2007) suggest that when internal agency problems between parent company

managers and foreign operations managers increase, firms are more likely to repatriate cash held

in foreign subsidiaries to avoid foreign operations managers’ self-maximizing behavior. For

firms with centralized treasury functions, internal agency costs are less of a concern. However,

external agency costs, arising from the agency conflict between parent company managers and

shareholders, may also increase the likelihood of misallocation of internal resources. The
10
literature suggests external agency costs could encourage managers to extract private benefits of

control by investing in suboptimal growth opportunities (Jensen, 1986; Hope and Thomas,

2008), make value-destroying capital expenditures (Shin and Stulz 1998) and acquisitions

(Harford, 1999), and acquiesce to rent-seeking behavior by subsidiary managers (Datta,

D’Mello, and Iskandar-Datta, 2009).

Finally, repatriation taxes represent an internal capital market friction, which may limit

multinational firms’ ability to allocate resources among their domestic and international business

operations efficiently. Theory and prior empirical work indicate that using internal capital to

fund operations within a firm is generally less costly than external capital because of information

asymmetry problems (Myers, 1984; Myers and Majluf, 1984; Shyam-Sunders and Myers, 1999).

However, large multinational firms issued debt rather than incurring costs related to the

repatriation of foreign earnings to satisfy investors’ demands of return of capital in the pre-TCJA

era. For example, to complete repurchases and dividend payments, Apple borrowed $17 billion

despite having $145 billion of cash in 2013 (Lattman and Eavis 2013), and eBay borrowed $3

billion in 2012 rather than accessing their $7 billion in foreign cash holdings (Mead and Kucera

2012). Beyer et al. (2017) provide evidence consistent with repatriation costs increasing

abnormal debt to fund shareholder payouts. De Simone and Lester (2018) document that frictions

created by repatriation taxes explain the use of external domestic debt financing by these cash-

rich firms. Because of the increased tax and financial reporting costs associated with

repatriations, research suggests a positive association between repatriation costs and cash held by

foreign subsidiaries (Foley et al., 2007; Hanlon et al., 2015).

Prior literature explores the effect of the American Jobs Creation Act (AJCA) on firm

behavior. The AJCA temporarily decreased repatriation taxes on foreign earnings, making

11
internal capital less costly to access. 12 On the one hand, Blouin and Krull (2009) and

Dharmapala et al. (2011) provide evidence consistent with firms using repatriated earnings from

the AJCA tax holiday to increase shareholder payouts instead of domestic capital investments.

On the other hand, Faulkender and Petersen (2012) find that while most firms that voluntarily

repatriated earnings increased shareholder payouts, financially constrained firms increased

domestic investment after the AJCA. Additionally, Arena and Kutner (2015) examine the

transitions from worldwide tax systems to territorial tax systems in the United Kingdom and

Japan and find that payouts increase but document no change in domestic investment.

While prior research documents that reductions in repatriation taxes largely lead to increases

in shareholder payouts and not capital investment, there are unique features of the TCJA that

could lead to differing results. First, Slemrod (2018) argues that the corporate tax rate reduction

should decrease the cost of capital, increasing the number of domestic, positive net present value

projects. However, while bonus depreciation may increase incentives for capital expenditures, it

may also dampen the impact of the corporate tax rate reduction on firm investment (Slemrod

2018). Second, while the AJCA was temporary and participation was optional, the deemed

repatriation under the TCJA and the quasi-territorial tax system are mandatory and implement

large inclusion regimes (i.e., GILTI) that are unique to the U.S. 13 The permanent reduction in

repatriation taxes grants companies access to additional cash on a long-term basis. As a result,

companies may be more likely to make long-term capital investment decisions rather than short-

12
Similar to the TCJA, the AJCA provided a dividends received deduction for repatriating firms. However, the
AJCA provided an 85 percent dividends received deduction when the statutory tax rate was 35 percent, resulting in
an effective 5.25 percent tax rate on repatriated earnings minus any available foreign tax credits. The dividends
received deduction was offered on the greater of $500 million or earnings designated as permanently reinvested in
the financial statements issued on or before June 20, 2003. Additionally, repatriation under the AJCA was optional.
13
For example, the TCJA introduces a global intangible low-taxed income (GILTI) inclusion to discourage income
shifting from the U.S. to foreign jurisdictions and among foreign jurisdictions. Importantly, GILTI proxies for
intangible income using a return on foreign tangible assets. Thus, to minimize overall taxes, firms can minimize
their GILTI inclusion by increasing foreign investment in tangible assets.

12
term shareholders payouts via share repurchases, which are generally made as a result of

temporary cash inflows (Allen and Michaely, 2003; Brav, Graham, Harvey, and Michaely, 2005;

DeAngelo, DeAngelo, and Stulz, 2006; Beyer et al., 2017). Third, in contrast to prior literature’s

settings of the United Kingdom and Japan’s transitions to a territorial tax system (Arena and

Kutner, 2015), the United States has greater opportunities for companies to invest capital. 14

Finally, Hanlon, Hoopes, and Slemrod (2019) analyze earnings conference calls and find a

number of firms stating they would increase investment post-TCJA.

Due to the significant differences between the TCJA and the AJCA and the findings from

Hanlon et al. (2019), we examine the relation between firms’ foreign cash levels pre-TCJA and

shareholder payouts, capital expenditures, and debt reductions post-TCJA. 15 The conversion to a

quasi-territorial tax system creates a permanent reduction in repatriation taxes, decreasing

internal capital market frictions, and, in turn, providing companies access to lower cost internal

capital to spend on domestic capital investments. In the pre-TCJA environment, De Waegenaere

and Sansing (2008) model that for some firms, investing the funds generated from foreign

operating assets in financial assets was optimal despite the weakly negative net present value.

This result occurred because obtaining permanent deferral (i.e., avoiding repatriation taxes)

generated greater savings than the cost of investing in financial assets. Based on these theoretical

predictions, Blouin and Krull’s (2009) predictions assume that multinational firms’ investment

sets remained constant. Consistent with these theoretical predictions, prior literature suggests the

AJCA, a temporary tax holiday, increased shareholder payout but did not increase capital

14
According to World Population Review (GDP Ranked by Country, 2019), which ranks the Gross Domestic
Product (GDP) by country, the United States has by far the highest GDP ($21.41 trillion) compared to Japan (3rd,
$5.36 trillion) and the United Kingdom (7th, $3.02 trillion).
15
We recognize that firms have incentives and/or lack of consequences that can prompt managers to make claims
that do not necessarily match their true intentions. For instance, Graham, Hanlon, and Shevlin (2010)’s survey
reports U.S. capital investment as the most prominent use of repatriated cash from the AJCA. However, Blouin and
Krull (2009) and Dharmapala et al. (2011) results do not support these claims.

13
expenditures (Blouin and Krull, 2009; Dharmapala et al., 2011). However, the TCJA eliminated

the permanent deferral option thereby eliminating the benefit of investing in financial assets for

multinational firms who have reached the optimal level of foreign operating assets. For firms

who previously employed this strategy, investment in slightly negative net present value projects

(i.e., financial assets) will no longer be an optimal investment choice.

Prior literature finds that multinational firms issued abnormal levels of domestic debt to fund

shareholder payouts (Beyer et al., 2017) in the years prior to the TCJA. Investment sets changed

post-TCJA, and because firms can issue debt as a substitute for repatriation, it is unclear whether

multinational firms will increase shareholder payouts in the post-TCJA environment. Consistent

with prior literature (Blouin and Krull 2009; Dharmapala et al. 2011), we state our first

hypothesis in the alternative as follows:

Hypothesis 1. Firms with greater pre-TCJA foreign cash balances increase shareholder
payouts in the post-TCJA period.

Prior literature finds that increases in investment only occurred for financially constrained

firms following the AJCA tax holiday (Faulkender and Petersen 2012). As outlined above, the

new U.S. tax policy enables access to lower cost internal capital, eliminates the permanent

deferral option which increases firms’ investment sets, and firms are communicating plans to

increase investment (De Waegenaere and Sansing, 2008; Blouin and Krull, 2009), suggesting

that trends documented following the temporary AJCA tax holiday may not occur again

following the TCJA. Therefore, it is unclear whether firms will alter their capital expenditures

post-TCJA. We formally state our second hypothesis in null form as follows:

Hypothesis 2. Firms with greater pre-TCJA foreign cash balances do not change capital
expenditures in the post-TCJA period.

14
 Research provides evidence that internal capital market frictions caused by repatriation costs

lead to higher levels of abnormal debt due to firms accessing debt markets to return cash to

shareholders and spend on domestic investment (Albring 2006; Beyer et al., 2017; De Simone

and Lester, 2018). Therefore, it is possible that the reduction in repatriation costs could lead to

reductions in firms’ debt holdings because debt is no longer needed as a substitute for

repatriation. However, firms with low cost external financing may forgo debt reductions in favor

of other alternative investment opportunities. We formally state the third hypothesis in the

alternative form:

Hypothesis 3. Firms with greater pre-TCJA foreign cash balances reduce debt in the post-
TCJA period.

3.2 Foreign Investment Incentives

To curb income shifting to foreign jurisdictions under the new territorial tax system, the

TCJA includes several new provisions unique to the TCJA, which distinguishes the current study

from literature examining this transition in other countries such as the U.K. and Japan (Arena

and Kutner, 2015). Legislators designed the GILTI inclusion to not only prevent income shifting

to foreign jurisdictions but also to prevent income shifting among foreign subsidiaries. The

GILTI inclusion does not directly calculate intangible income from foreign operations. Thus,

increasing investment in tangible foreign assets decreases the GILTI inclusion and, all else

equal, decreases income tax expense. At the same time, the FDII deduction was designed to

encourage exports and maintaining intellectual property in the U.S. However, as detailed in

Appendix A, because the FDII deduction does not directly calculate intangible income of the

U.S. firm, firms can maximize their FDII deduction by decreasing domestic investment in

tangible assets. Therefore, the GILTI inclusion and the FDII deduction encourage firms to

increase foreign investment in tangible assets and decrease domestic investment in tangible

15
assets, which contradicts the stated objective of the TCJA and may represent a new internal

capital market friction. 16

Given the GILTI inclusion’s incentive (FDII deduction’s penalty) for increasing foreign

(domestic) capital expenditures, foreign capital expenditures may increase post-TCJA for firms

poised to invest in foreign tangible assets. Accordingly, our final hypothesis is stated as follows:

Hypothesis 4. Among firms with tax-related incentives to increase foreign investment, those
with greater pre-TCJA foreign cash balances increase foreign capital expenditures in the
post-TCJA period.

4. RESEARCH DESIGN AND DATA

4.1 Research Design

The TCJA consisted of several broad changes (e.g., statutory tax rate decrease, 100 percent

bonus depreciation, interest deductibility, and foreign earnings taxation). We are interested in

studying the consequences of the change in foreign earnings taxation, which eliminated federal

repatriation taxes as an internal capital market friction. Therefore, our treatment group is

multinational firms. A possible control group includes firms from other countries with similar

legal systems (i.e., Canada, Japan and the United Kingdom). However, using this set of foreign

firms as the control group has at least two significant challenges in our setting. First, comparing

U.S. multinational firms to foreign firms does not allow us to isolate the change in foreign

earnings taxation, but instead, it only allows us to isolate the broader consequences of the TCJA.

Second, the parallel trends assumption is violated when we examine our dependent variables in

the pre-TCJA period (i.e., 2010 – 2016) (untabulated). Specifically, the levels of our dependent

16
Additionally, bonus depreciation is available to U.S. parent firms on qualified property placed into service after
September 27, 2017 with no distinction between foreign and domestic qualified property.

16
variables do not have similar trends for multinational firms as the selected foreign firms.

Therefore, we choose to use U.S. domestic firms as our control group. 17 We isolate the change in

foreign earnings taxation by using a difference-in-differences research design that compares

corporate spending and investment for multinational firms (i.e., treatment firms) to domestic-

only firms (i.e., control firms). Multinational and domestic firms are similarly affected by all

broad TCJA changes except the taxation of foreign earnings. The change in the taxation of

foreign earnings and associated deemed repatriation of UFE does not affect domestic firms. We

use entropy balancing in our regression analysis so that there are no statistical differences among

the first three statistical moments of the control variables when comparing multinational and

domestic firms. We use the following difference-in-differences model: 18

Expenditurei,t = α0 + α1Postt + α2MNCi × Postt + α3MNCi + + α4Book ETRi + α5Cash

ETRi + α6Lossi,t + α7Sizei,t + α8Earningsi,t + α9Cash Flowi,t +
α10Leveragei,t + α11Total Cashi,t + α12MTBi,t + α13Sales Growthi,t + (1)
α14CapEx2014i + α15R&Di,t + α16ACQi,t + α17Firm Agei,t +
α18RE/BVi,t + α19Returnsi,t + α20Optionsi,t + α21St. Dev. Earningsi,t +
α22Dividendsi,t-1 + α23Repurchasesi,t-1 + ε

Equation (1) is estimated separately for each of the following dependent variable measures

for Expenditure: Dividends, Repurchases, CapEx and Debt Reduction. Dividends is defined as

total dividends scaled by end of fiscal year 2014 total assets. 19 Repurchases is equal to the

purchase of common and preferred stock less any decrease in the redemption of preferred stock

or less any decrease in preferred stock if the redemption value of preferred stock is missing, all

scaled by end of fiscal year 2014 total assets. We use OLS regression because it allows for easier

17
The parallel trends assumption for our sample is examined and discussed in section 4.2. To increase the similarity
of treatment and control groups, we also entropy balance the sample of domestic only and multinational firms.
18
All variables are defined in Appendix B.
19
Fiscal year 2014 total assets is used as the common scalar (instead of lagged total assets) so that prior period’s
expenditure does not affect the current period’s denominator in the variable calculation. Inferences are similar if we
use prior year’s total assets.

17
interpretation of coefficient estimates on interactions and because we can compare the statistical

difference of coefficients across regressions. 20 CapEx is capital expenditures scaled by end of

fiscal year 2014 total assets. Debt Reduction is equal to total debt reductions during the year

scaled by fiscal year 2014 total assets. 21 Post is equal to one for fiscal years ending in 2018 and

2019, and zero for fiscal years ending prior to 2017. 22 MNC is equal to one if the firm reports

either foreign income or foreign tax expense in any period 2014 – 2016, inclusive. We choose

this period to measure MNC because it is the closest period to the TCJA enactment and best

identifies firms most affected by the change in the taxation of foreign earnings. A positive

(negative) coefficient of interest, α2, indicates an incremental increase (decrease) in the

respective expenditure for multinational firms compared to domestic firms.

Control variables are based on other tax changes included in the TCJA and prior literature

that examines payout policy (Fama and French, 2001; DeAngelo et al., 2006; Grullon, Paye,

Underwood, and Weston, 2011) and capital investment (Canace, Jackson, and Ma, 2018). The

TCJA lowered the overall U.S. corporate tax rate from a maximum of 35% to 21%. Such a

lowering of the corporate rate would also benefit firms with higher effective tax rates more than

firms with low effective tax rates. Therefore, we control for pre-TCJA effective tax rates using

the ETR from 2016. Book ETR is equal to fiscal year-end 2016 income tax expense scaled by

20
In untabulated results, inferences are the same if we use Tobit regressions and the INTEFF Stata command to
compute statistical significance based on Ai and Norton’s (2003) suggestion.
21
Debt Reduction is measured using Compustat item DLTR which has the following Compustat definition: “a
reduction in long-term debt caused by long-term debt maturing (being classified as a current maturity), payments of
long-term debt and the conversion of debt to stock.” As such, this is a noisy measure of cash payments for long-term
debt, however, is the best available machine-readable data, to our knowledge. Other studies that examine debt
reduction often use a measure of net debt (i.e., total debt changes) (Caskey and Hanlon, 2013; Dudley, 2012; Chu,
2018), but those measures would have even more noise in our setting than our measure because they incorporate
debt issuances.
22
Inferences for the coefficient on the interaction of interest are similar if we include year fixed effects. We choose
to omit year fixed effects so that the coefficient on Post has more meaning and the variation is not drowned out by
including year fixed effects.

18
pretax income minus special items, truncated at zero and one. Cash ETR is equal to fiscal year-

end 2016 cash taxes paid scaled by pretax income minus special items, truncated at zero and

one. 23 We also control for loss firms by including Loss, set equal to one when income before

extraordinary items is less than zero and equal to zero otherwise.

Size is equal to the natural logarithm of total assets. Earnings is income before extraordinary

items scaled by fiscal year 2014 total assets. Cash Flow is operating cash flows scaled by fiscal

year 2014 total assets. Leverage is long term debt scaled by fiscal year 2014 total assets. 24 Total

Cash is total cash and cash equivalents scaled by fiscal year 2014 total assets. MTB is the market

value of equity scaled by book value of equity. Sales Growth is current period sales minus prior

period sales, all scaled by prior period sales. CapEx2014 is total capital expenditures scaled by

total assets, all measured in fiscal year 2014. R&D is total research and development expense

scaled by fiscal year 2014 total assets. ACQ is funds spent on mergers and acquisitions scaled by

fiscal year 2014 total assets. 25 Firm Age is equal to the natural logarithm of the number of years

that the firm has appeared in the Compustat database. RE/BV is retained earnings scaled by book

value of equity.

We also include controls for past stock returns (Returns), stock options (Options), and

volatility of earnings (St. Dev. Earnings). Returns is measured as the firm’s stock return

compounded monthly for the two-year periods ending before the current year. Options is

calculated as the annual percentage change in total diluted shares outstanding as if no

23
Book ETR and Cash ETR also control for general tax planning. Book ETR captures the effect of permanent book-
tax differences whereas Cash ETR captures the effect of both temporary and permanent book-tax differences; put
differently, the Book ETR captures the financial statement impact of tax planning whereas Cash ETR captures the
cash outlay portion of tax planning, both of which may be impacted by a lowering of the corporate tax rate.
Inferences are similar if we only control for either Book ETR or Cash ETR (untabulated).
24
In untabulated analysis, we restricted the sample to firms with positive leverage for testing debt reductions.
Inferences are similar to those presented in the tables.
25
RD and ACQ are set equal to zero when missing in Compustat.

19
repurchases occurred the current year. St. Dev. Earnings is measured as the standard deviation of

earnings for the current and previous four years. We also include lagged values of shareholder

payouts (Dividends and Repurchases) because they can be sticky from the previous year.

Industry fixed effects based on two-digit SIC codes are included in all models. Our analysis

includes robust standard errors clustered by firm. All continuous variables are winsorized at the

1st and 99th percentiles to lessen the influence of outliers and data errors.

An underlying assumption of our difference-in-differences analysis is that multinational

(domestic) firms have (do not have) greater pre-TCJA foreign cash balances. To test our

hypotheses more directly, we also perform analysis on a subset of multinational firms that

disclose their foreign cash balances. This allows us to see how corporate spending and

investment vary with multinational firms’ level of foreign cash. The model below is used for the

multinational firm sample:

Expenditurei,t = α0 + α1Postt + α2Foreign Cashi × Postt + α3Foreign Cashi + α4Domestic (2)

Cashi + Controls + ε

Foreign Cash is the foreign cash balance at the end of fiscal year 2016 (hand-collected from

10-Ks) scaled by total assets at the end of fiscal year 2014. 26 Domestic Cash is total cash minus

foreign cash (measured at the end of fiscal year 2016) scaled by total assets (measured at the end

of fiscal year 2014). We measure foreign cash at the end of fiscal year 2016 because it best

approximates the one-time cash windfall from the permanent reduction in repatriation costs as a

result of the TCJA. Foreign cash is not equivalent to permanently reinvested earnings (PRE) or

UFE. Laplante and Nesbit (2017) discuss the differences between foreign cash, PRE and UFE;

Blaylock et al. (2019) document the high correlation (>90%) between PRE and estimated UFE.

26
Inferences are similar to those presented when we estimate foreign cash using methodology provided in Campbell
et al. (2019). We acknowledge the limitation that using hand-collected levels of disclosed foreign cash only allows
us to make inferences about those multinational firms that disclose foreign cash.

20
We examine foreign cash because it is often held overseas for tax reasons, (Foley et al., 2007)

and Slemrod’s (1992) hierarchy of tax planning suggests there are significant costs associated

with liquidating real assets; therefore, foreign cash’s liquidity means that it is likely available

first for investment purposes in the post-TCJA setting. Control variables included in Equation (2)

are the same as those described above in Equation (1).

4.2 Sample Selection

Because the TCJA became law on December 22, 2017, the sample is limited to firms with a

calendar fiscal year-end and includes fiscal years 2015, 2016, 2018 and 2019. We exclude fiscal

years ending on December 31, 2017 because they include both pre- and post-TCJA activity. 27

Table 1 outlines the sample selection process. We collect data from Compustat in August of

2020 resulting in 12,850 firm-year observations. We eliminate 2,300 firm-years missing data to

scale our variables (i.e., fiscal year end 2014 assets). Next, we eliminate 2,359 firm-years

missing data to calculate Returns. We require firms to have data available in all four years of the

sample, which reduces the sample by another 2,623 firm-years. Finally, 628 firm-years are

eliminated for missing control variables. This results in a sample of 4,940 firm-year

observations. The multinational firm sample starts with the multinationals included in the full

sample (4,940×0.662=3,270) and is reduced by 1,526 firm-years that do not disclose foreign

cash, resulting in a total of 1,744 firm-year observations. The multinational firm sample is also

balanced so there are both two years pre- and post-TCJA.

A strong difference-in-differences research design relies on four assumptions. First, the

treatment group is significantly affected by the international provisions of the TCJA, but the

control group is not. The change in the taxation of foreign earnings only affects the taxation of

27
Inferences are similar if we include fiscal year 2017 in the sample.

21
multinational firms and not domestic firms. Second, the pre- and post-TCJA time period samples

should be balanced (Atanasov and Black, 2016). We have a balanced sample because our

analysis compares two years pre-TCJA to two years post-TCJA and having all four firm-year

observations is a sample selection requirement. Third, the treatment (i.e., the changes in the

taxation of foreign earnings) should have a significant effect on the treatment group. Firms that

had significant amounts of foreign cash recognized large amounts of one-time transition tax in

their financial statements leading to greater access to internal capital. For example, Cisco (who

was top five in foreign cash balances prior to the TCJA) recorded over $8 billion in transition

taxes during the 2017 and 2018 time periods. Therefore, it is reasonable to assume that the TCJA

significantly increased accessibility of internal capital via foreign cash. Fourth, the outcome

variables for the treatment and control groups should exhibit parallel trends over the pre-TCJA

period. Panel A of Table 2 charts the average level of each dependent variable for multinational

and domestic firms. The dependent variables appear to have similar trends prior to the TCJA. To

further verify this assumption is met, we estimate the following regression model for each of our

outcome variables:

Expenditurei,t = α0 + α1Time Trend + α2MNC + α3MNC × Time Trend + ε (3)

Expenditure and MNC are defined above. Time Trend is a time-trend variable calculated as

the number of days between the financial report date and January 1st, 2010, scaled by 100. In

order to compare a longer range of trends, the sample period for the parallel trends assumption is

2010 to 2016 (whereas the DiD sample is balanced from 2015 to 2019). If there are parallel

trends between the treatment and control groups then the coefficient α3 will be insignificantly

different from zero. Table 2 Panel B presents the coefficients and t-statistics from the estimation

22
of Equation (3). In all four models, α3 is not significantly different from zero, supporting the

parallel trends assumption in the pre-event period.

4.3 Descriptive Statistics

Table 3 reports the summary statistics for the variables used in the study. The mean (median)

of Dividends and Repurchases is 1.2 (0.0) and 2.7 (0.4) percent of assets, respectively,

documenting high positive skewness in these variables. The mean (median) of CapEx is 5.1 (3.3)

percent of assets. Debt Reduction mean (median) is 17.3 (3.7) percent of assets, which also

indicate a high positive skewness. As discussed in the research design section, the large amount

of debt reductions documents that the variable is measuring more than just cash payment

reductions to long term debt. By construction, the average for POST is 0.50 indicating that each

firm has four years of data included (i.e., two years pre-TCJA and two years post-TCJA). The

DiD sample is comprised of 66.2 percent multinational firms. Within the multinational sample,

firms have 7.8 (8.6) percent of their total assets held in foreign (domestic) cash, on average.

5. RESULTS

5.1 Difference-in-Differences Analysis

Table 4 presents the coefficient estimates from our baseline regression specification

described above in section 4.1. The sample includes two years pre-TCJA and two years post-

TCJA for both multinational and domestic firms. The dependent variable is Dividends,

Repurchases, CapEx, and Debt Reduction across the columns. Results show a positive

association between lagged and current measures of dividends, repurchases and capital

expenditures, which is consistent with the expectation for “sticky” shareholder payouts and

recurring investment. Column (4) shows that leverage is positively associated with debt

23
reduction, which also matches intuitive predictions. The coefficient estimate for Post is positive

and significant for Dividends and Debt Reduction, which is consistent with each of these

expenditures increasing, on average, for all firms after TCJA.

The coefficient α2 is not significantly different from zero across all of the columns. This

finding implies that, compared to domestic firms, multinational firms did not change their

spending and investment behavior in the post-TCJA periods. The lack of significant results on

average differs from prior studies that examine the spending of repatriated cash under the AJCA.

For example, Blouin and Krull (2009) and Dharmapala et al. (2011) document that, due to the

AJCA, repatriating firms increase share repurchases compared to non-repatriating firms. Because

repatriation under the AJCA was voluntary, prior research directly captures firms who chose to

repatriate under the AJCA. The deemed repatriation and associated toll tax under the TCJA were

mandatory. Thus, if firms held cash abroad to avoid U.S. repatriation taxes (Foley et al., 2007),

then we can assume this cash is available post-TCJA for domestic expenditure. Given the

difficult task of interpreting a null results outcome (Cready et al. 2020), we next examine various

cross-sectional settings where the spending and investment of foreign cash may be more likely.

5.2 Financial Constraints

We first examine whether financial constraints affect the relation between investment and

greater access to foreign cash. De Waegenaere and Sansing (2008) model that pre-TCJA, the best

investment strategy for firms who have reached the optimal level of foreign operating assets may

be to invest in slightly negative net present value foreign financial assets because the cost of

these investments was less than the benefit of permanent deferral of U.S. repatriation taxes. This

investment strategy provides an explanation for the stockpiling of foreign cash overseas, which is

associated with U.S. repatriation taxes as documented in Foley et al. (2007). Additionally, firms

24
that were not financially constrained had access to external financing to fund shareholder

payouts (Beyer et al., 2017) and capital investments (Albring, 2006; De Simone and Lester,

2018). To the extent that internal capital market frictions created by repatriation taxes limited

financially constrained firms from returning cash to shareholders and taking advantage of

investment opportunities, shareholder payouts and capital expenditures should increase post-

TCJA. Therefore, we expect the reduction in internal capital market frictions to primarily change

the spending and investment behavior of firms that faced greater pre-TCJA financial constraints.

Table 5 examines the cross-sectional results based on financial constraint. We follow

Faulkender and Petersen (2012) to measure financial constraint. Specifically, we measure the

level of internal capital (earnings after taxes but prior to interest) compared to investment (capital

expenditures) for the four years from 2013 through 2016. The level of financial constraint is

measured as the number of years where earnings after taxes but before interest is less than the

level of capital expenditures, scaled by four. Therefore, the financial constraint measure ranges

from zero to one, inclusive, and is equal to strict values of 0.00, 0.25, 0.50, 0.75, and 1.00. The

measure increases in financial constraint. Table 5 partitions the sample into firms that are

financially constrained (unconstrained) in Panel A (Panel B) based on the financial constraint

measure being greater than 0.50 (less than 0.50). 28

In Panel A, the coefficient α2 is positive and significant when the dependent variable is

Repurchases, and the coefficient α2 is not significantly different from zero when the dependent

variable is Dividends. This finding implies that, compared to domestic-only firms, financially

constrained multinational firms increased repurchases but not dividends in the post-TCJA period.

Blouin and Krull (2009) highlight that firms typically prefer repurchases to dividends because

28
We omit firms with financial constraint equal to 0.50.

25
repurchases are not “sticky” (Guay and Harford, 2000; Jagannathan et al., 2000). Additionally,

Blouin and Krull (2009) argue that share repurchases generate capital gains, which allows for the

deferral of taxes, a return of capital upon sale, timing of sale to match capital losses, and use of

capital loss carryforwards. Our evidence suggests that, in the post-TCJA period, financially

constrained multinationals increased repurchases without a corresponding increase in dividends,

potentially increasing their internal capital flexibility in the future. As discussed previously, the

AJCA did not change firms’ investment opportunities; thus, it is unsurprising that Blouin and

Krull (2009) and Dharamapala et al. (2011) document an increase in share repurchases, which

are more flexible and less “sticky.”

In Panel A the coefficient estimate for α1 is positive and significant while the coefficient

estimate for α2 is insignificant when the dependent variable is CapEx in column three, suggesting

both financially constrained domestic-only firms and financially constrained MNCs increased

capital expenditures following the TCJA, but changes to the U.S. taxation of foreign earnings do

not appear to be driving this result. Also in Panel A, the coefficient estimates for α1 and α2 are

relatively similar in magnitude in column 4 when the dependent variable is Debt Reduction,

however, α1 is positive and significant and α2 is negative and significant. This result indicates

that financially constrained multinational firms made no change to their debt reduction payments

in the post-TCJA period (i.e., α1+ α2=0) while domestic firms increased their debt reduction

payments in the post-TCJA period. These results suggest that eliminating internal capital market

frictions did not result in debt repayment for MNCs.

We find no significant α2 coefficient for firms that have low financial constraints (Panel B).

This suggests that reducing internal capital market frictions does not appear to affect the

spending and investment decisions of firms without capital constraints. This is most likely due to

26
their ability to generate cash from domestic operations to fund shareholder payouts and

investments and/or their ability to borrow (Beyer et al., 2017; De Simone and Lester, 2018),

reducing the reliance on foreign cash.

In untabulated results, we also examine whether the spending and investment of foreign cash

is affected by the firms’ growth opportunities. Prior literature uses the market-to-book ratio to

measure growth opportunities (Blouin and Krull, 2009; Faulkender and Petersen, 2012). We do

not find statistically significant results when we partition the sample based on the median

market-to-book ratio at the end of 2016. We are unable to disentangle domestic and foreign

growth opportunities when using the firm-level market-to-book ratio. The reduction in

repatriation costs allows firms to use foreign cash to respond to domestic growth opportunities,

whereas there should be little response to foreign growth opportunities. As another proxy for

growth opportunities, we use the level of firm-level underinvestment as measured by Biddle,

Hilary, and Verdi’s (2009) model of investment efficiency. Firms that historically underinvest

are identified as firms with the greatest domestic growth opportunities. We reason that

underinvestment captures domestic and not foreign growth opportunities because firms would

have the ability to use pre-TCJA foreign cash to reduce foreign underinvestment. Similar to the

market-to-book cross-sectional results, we find no significant results in the cross-section of

investment efficiency.

5.3 Multinational Firm Sample

Our next analyses exclude domestic firms and focus on multinational firms, allowing us to

examine how foreign cash balances affect spending and investment within multinational firms.

In addition, using this sample allows us to examine cross-sectional results in how firms most

affected by the one-time reduction in repatriation costs invested their foreign cash. The sample is

27
comprised of 1,744 firm-year observations for the time period 2015 through 2019, excluding

fiscal year 2017. Table 6 presents the coefficients from Equation (2). The coefficient α2 is

positive and significant in column 2 when the dependent variable is Repurchases. This implies

that multinational firms with greater levels of pre-TCJA foreign cash increased their repurchases

in the post-TCJA time period. Economically, a one standard deviation increase in foreign cash

(0.075) leads to an incremental increase in post-TCJA repurchases of 11.5 percent. 29 We find no

other significant effect of foreign cash on post-TCJA spending and investment. We do not

examine cross-sectional tests based on financial constraints for the multinational firm sample

because, using the Faulkender and Petersen (2012) measure of financial constraint classifies 86

percent of the multinational firms as not financially constrained. Therefore, we examine cross-

sectional tests to confirm that our results thus far are due to the changes in taxation of foreign

earnings, which eliminated the permanent deferral of U.S. repatriation taxes.

5.4 Repatriation Costs

In Table 7, we present cross-sectional analysis based on levels of repatriation costs. We

follow Nessa (2017) to measure repatriation costs by multiplying 2016 pretax foreign income by

the difference between 0.35 and the three-year foreign effective tax rate, scaled by assets. The

three-year foreign effective tax rate is cumulative foreign tax expense scaled by cumulative

foreign pretax income for the period 2014 through 2016. Firms with large amounts of foreign

cash primarily accumulated those cash balances as a result of repatriation costs (Foley et al.,

2007). If the TCJA increases repurchases, then we expect the results from Table 6 to be the

strongest for firms facing high repatriation costs because the TCJA eliminated the permanent

deferral of U.S. repatriation taxes on foreign earnings. Panel A (Panel B) includes firms that had

29
([0.075 × 0.054] ÷ 0.035). The mean of Repurchases for the multinational sample is 0.035.

28
above (below) the sample median of repatriation costs. The coefficient α2 is positive and

significant in Panel A when the dependent variable is Repurchases. The coefficient remains

insignificantly different from zero when examining the other types of spending and investment

(i.e., columns 1, 3 and 4). This evidence is consistent with multinational firms with high levels of

pre-TCJA foreign cash increasing post-TCJA repurchases as a result of the reduction in

repatriation costs. In Panel B, the coefficient α2 is not significantly different from zero in any

column, consistent with the notion that firms facing low pre-TCJA repatriation costs did not

change their spending and investing behavior in the post-TCJA period.

5.5 Trapped Cash Held for Other Reasons

Most of the literature that examines the reasons that firms hold foreign cash attribute

managers’ incentive to hold foreign cash to the pre-TCJA tax system (i.e., due to repatriation tax

costs). However, Laplante and Nesbitt (2017) document other reasons that multinational firms

hold foreign cash (e.g., R&D needs, capital needs, borrowing, leverage and growth

opportunities) and provide a model to estimate trapped cash. We use their model to examine

whether our results are driven by firms that hold foreign cash for reasons other than tax

incentives. Specifically, we use the estimated coefficients from Table 3 column 2 of Laplante

and Nesbitt (2017) to calculate trapped cash for our sample. This measure of trapped cash is

driven by non-tax factors because the specific control variables are not related to repatriation tax

costs. We then partition our sample into above and below the median of estimated trapped cash

in Table 8. 30 Panel A (Panel B) includes firms with high (low) estimated trapped cash. If foreign

cash held for tax purposes is driving our inferences in regard to consequences post-TCJA then

we should see our previous results in both Panels. The coefficient α2 is positive and significant in

30
Table 8 excludes 328 firm-year observations that are included in Tables 6 and 7 due to data limitations imposed
by the Laplante and Nesbitt (2017) estimated trapped cash calculation.

29
Panel A, but is not statistically different from the same coefficient in Panel B. Therefore, we

conclude that the positive association between foreign cash and post-TCJA repurchases is

primarily a result of the elimination of future repatriation costs and not driven by firm operating

needs.

5.6 Foreign and Domestic Capital Expenditures

Despite differences between the TCJA and AJCA as well as between tax reform in the U.S.

and abroad, our results are largely consistent with prior literature (Blouin and Krull, 2009;

Dharmapala et al., 2011; Arena and Kutner, 2015). Put differently, we find very little evidence

that firms used the cash windfall from the TCJA to increase capital expenditures. However, we

believe that the introduction of additional internal capital market frictions through the GILTI

inclusion and FDII deduction could explain the lack of significant results for capital expenditures

in Tables 4 – 8. Firms are not required to disclose disaggregated capital expenditures. However,

firms are required to include the net property, plant, and equipment levels for both foreign and

domestic operations in their annual segment disclosures within their 10-Ks (ASC 280-10-50-

41). 31 We hand-collect foreign and domestic net property, plant, and equipment for fiscal years

ending during the period 2015 – 2018 for our sample. We use these hand-collected disclosures to

compute changes in net foreign and domestic property, plant, and equipment from 2015 to 2016

and 2017 to 2018. 32 We omit the 2016 to 2017 change because the TCJA became effective in

31
The Compustat Segment database does provide foreign and domestic capital expenditures, but less than 10 percent
of firms in the database have data for these amounts due to inconsistent disclosure. Our approach of hand-collecting
foreign and domestic net property, plant, and equipment is an attempt to improve on prior literature’s challenge of
measuring foreign versus domestic investment described in Dong, Cao, Zhao, Deshmukh (2019). We acknowledge
that the change in PPE is a noisy measure for capital expenditures. However, this is the best publicly available data
for disaggregated foreign and domestic capital expenditures.
32
We acknowledge that net property, plant, and equipment is an imperfect proxy for the tax adjusted basis of
Qualified Business Asset Investment (QBAI). However, QBAI is determined using the adjusted basis of tangible
property net of depreciation. Depreciation is calculated using the alternative depreciation system (i.e., straight-line)
(Ernst and Young 2018), which is consistent with the most common method of depreciation for financial statement
purposes. Thus, we believe foreign and domestic property, plant, and equipment are meaningful proxies for QBAI.

30
this period, and details were made public before the enactment. Therefore, our analysis compares

the 2016 foreign/domestic capital expenditures with the 2018 foreign/domestic capital

expenditures.

Table 9 presents the estimated coefficients. We add Depreciation, measured as annual

depreciation scaled by total assets at the end of 2014, as a control variable in order to try to

isolate changes in PPE due to capital expenditures. In addition, we disaggregate total earnings

into its foreign (Foreign Earnings) and domestic (Domestic Earnings) parts. All other control

variables are similar to previous tables. The dependent variables in columns 1 and 2 are foreign

and domestic capital expenditures, respectively. The coefficient α2 is not significantly different

than zero, on average. We next disaggregate the sample based on firms with positive estimated

GILTI inclusions (High GILTI) versus firms without estimated GILTI inclusions (Low GILTI).

We estimate the GILTI inclusion as the three-year average foreign after-tax income (Compustat

PIFO – Compustat TXFO) less ten percent of net foreign property, plant, and equipment (PPE) at

the end of 2016 (the year before TCJA enactment), which is hand-collected from the firm’s

segment reporting in the 10-K, and divided by total assets. GILTI is set equal to zero if the

calculated GILTI amount is negative. The coefficient α2 is positive and significant when the

dependent variable is the change in foreign PPE and the firm has a high potential GILTI tax

inclusion (column 3). A one standard deviation increase in foreign cash increases the change in

foreign PPE by 28 percent post-TCJA.33 This evidence suggests that foreign capital expenditures

increased post-TCJA; given the confounding incentives for domestic investment, it is

unsurprising that our primary analyses do not capture a change in capital expenditures for

multinational firms. This result highlights a potential unintended consequence of the GILTI

33
The GILTI sample mean Foreign ΔPPE is 0.0072 and the standard deviation of Foreign Cash is 0.0708
(untabulated). The economic significance is calculated as [(α1+α2)*0.0708] / 0.0072.

31
inclusion; the inclusion appears to incentivize foreign rather than domestic capital expenditures.

The coefficient α2 continues to not be statistically different from zero in columns 4 – 6. Our

findings suggest the TCJA created a new internal capital market friction as a result of the new

tax provisions (i.e., FDII deduction, GILTI inclusion) included within the new tax law.

6. CONCLUSION

The TCJA is a significant change to U.S. corporate tax policy and in particular, significantly

changes the U.S. taxation of foreign earnings. The TCJA provides a unique setting to examine

whether firms increase shareholder payouts, capital investment and/or reduce outstanding debt in

response to a reduction in internal capital market frictions due to repatriation costs. Increased

domestic capital investment would spur economic growth, which is a stated objective of the

TCJA. This study examines how a change in internal capital market frictions from the enactment

of the TCJA affects firms’ spending and investment behavior.

Because the TCJA included other sweeping provisions, we first use a difference-in-

differences design to isolate the changes to the U.S. taxation of foreign earnings. In these

analyses, we find that, compared to domestic-only firms, financially constrained multinational

firms increased share repurchases. Next, we focus on a sample of only multinational firms to see

how responses differed based on the cash available for domestic expenditure post-TCJA.

Consistent with findings from the prior literature (Blouin and Krull, 2009; Dharmapala et al.,

2011; Faulkender and Petersen, 2012; Arena and Kutner, 2015), we find no evidence of an

increase in capital expenditures post-TCJA for firms with higher levels of pre-TCJA foreign cash

in the broad cross-section. Additionally, we find evidence consistent with an increase in

repurchases post-TCJA for firms with higher levels of pre-TCJA foreign cash. To verify that our

32
results capture the elimination of U.S. repatriation taxes on foreign earnings, we partition our

sample based on median estimated repatriation taxes. As expected, we find that the increase in

repurchases for firms with higher levels of pre-TCJA foreign cash is concentrated in firms with

above-median repatriation tax costs.

Although the insignificant results for capital expenditures is consistent with most prior

literature (Blouin and Krull, 2009; Dharamapala et al., 2011; Arena and Kutner, 2015), the

GILTI inclusion and FDII deduction are unique to the TCJA and may incentivize foreign rather

than domestic capital expenditures (See Appendix A) despite other measures (i.e., bonus

depreciation and Sec. 179 expensing) designed to incentivize domestic spending. When we

partition our results into foreign versus domestic capital expenditures and high versus low

likelihood of a firm having a GILTI inclusion, we find a positive and significant association post-

TCJA between foreign cash and foreign capital expenditures for firms most likely to have a

GILTI inclusion. This finding highlights an unintended consequence of the TCJA – namely, that

the GILTI inclusion may inadvertently incentivize foreign rather than domestic investment for

some multinational firms, introducing additional internal capital market frictions for U.S

multinational firms.

We make several contributions to the literature. First, our findings should be of use to

policymakers as they evaluate the effectiveness of the TCJA in encouraging domestic capital

investment. The TCJA created a significant debate regarding its ability to increase domestic

investment. Our findings indicate that multinational firms do not increase domestic investment

more than domestic-only firms and that for some multinational firms, the TCJA’s GILTI

inclusion incentivized foreign rather than domestic capital expenditures. We also provide

evidence that financially constrained firms use the internal capital markets shock as an

33
opportunity to change their shareholder payout policy to increase flexibility in their capital

structure. Second, prior research examines the investment consequences of the AJCA in addition

to tax reform in both the United Kingdom and Japan, and our results add to this existing stream

of literature. Finally, we contribute to the literature on the effect of taxes on firm decision-

making. Our findings imply that taxes influence managers’ investment decisions.

34
 REFERENCES

Ai, C., and E. Norton. 2003. Interaction terms in logit and probit models. Economics Letters 80:
123–129.
Albring, S. M. 2006. The effects of the cost of foreign internal funds on the probability that a
firm issues domestic debt. Journal of American Tax Association 28 (1): 25–41.
Allen, F., and R. Michaely. 2003. Payout policy. In: Constantinides, G., Harris, M., Stulz, R.
(Eds.), Handbook of the Economics of Finance. North-Holland, Amsterdam: 337–429.
Amberger, H. J., K. S. Markle, and D. M. P. Samuel. 2020. Repatriation taxes, internal agency
conflicts and subsidiary-level investment efficiency. The Accounting Review
forthcoming: https://doi.org/10.2308/TAR-2019-0259.
Arena, M. P., and G. W. Kutner. 2015. Territorial tax system reform and corporate financial
policies. Review of Financial Studies 28 (8): 2250–2280.
Atanasov, V. A., and S. B. Black. 2016. Shock-based causal inference in corporate finance and
accounting research. Critical Finance Review 5: (2): 207–304.
Bennett, B., A. Thakor, and Z. Wang, 2019, Stock repurchases and the 2017 Tax Cuts and Jobs
Act. Working Paper. Available at SSRN: https://ssrn.com/abstract=3443656
Beyer, B., J. Downes, and E. T. Rapley. 2017. Internal capital market inefficiencies, shareholder
payout, and abnormal leverage. Journal of Corporate Finance 43: 39–57.
Biddle, G. C., G. Hilary, and R. S. Verdi. 2009. How does financial reporting quality relate to
investment efficiency? Journal of Accounting and Economics 48 (2–3): 112–131.
Billett, M. T., and D. C. Mauer. 2003. Cross-subsidies, external financing constraints, and the
contribution of internal capital markets to firm value. Review of Financial Studies 16 (4):
1167–1201.
Blaylock, B., J. F. Downes, M. Mathis, and S. White. 2019. The effect of repatriation taxes on
bond pricing: Evidence from the 2016 election. Working paper.
Bloomberg. 2017. These are the corporate winners and losers in the GOP’s final tax bill
http://fortune.com/2017/12/16/gop-tax-bill-winners-and-losers/ (last accessed 5/20/20).
Blouin, J. L., and L. K. Krull. 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 (4): 1027–1059.
Brav, A., J. Graham, C. Harvey, and R. Michaely. 2005. Payout policy in the 21st century.
Journal of Financial Economics 77 (3): 483–527.
Campbell, J.L., D.S. Dhaliwal, L.K. Krull, and C.M. Schwab. 2019. U.S. multinational
corporations’ foreign cash holdings: An empirical estimate and its valuation
consequences. Working Paper.
Canace, T. G., S. B. Jackson, and T. Ma. 2018. R&D investments, capital expenditures, and
earnings thresholds. Review of Accounting Studies 23 (1): 265–295.

35
Caskey, J., and M. Hanlon. 2013. Dividend policy at firms accused of accounting fraud.
Contemporary Accounting Research 30 (2): 818–850.
Chu, Y. 2018. Shareholder-creditor conflict and payout policy: Evidence from mergers between
lenders and shareholders. Review of Financial Studies 31 (8): 3098–3121.
Clemons, R., and T. Shevlin. 2016. Tax policy debate: increasing the policy impact of academic
tax accounting research. Journal of the American Taxation Association 38 (1): 29–37.
Cready, W. M., J. He, W. Lin, C. Shao, D. Wang, and Y. Zhang. 2020. Is there a confidence
interval for that? A critical examination of null outcome reporting in accounting research.
Working paper available at SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3131251.
Datta, S., R. D’Mello, and M. Iskandar-Datta. 2009. Executive compensation and internal capital
market efficiency. Journal of Financial Intermediation 18 (2): 242–258.
Davis, L., and S. Chandra. 2018. U.S. cash repatriation plunges 50%, defying Trump’s tax
forecast. Bloomberg News December 19.
DeAngelo, H., L. DeAngelo, and R. Stulz. 2006. Dividend policy and the earned/contributed
capital mix: a test of the life-cycle theory. Journal of Financial Economics 81 (2): 227–
254.
Desai, M. A., C. F. Foley, and J. R. Hines. 2007. Dividend policy inside the multinational firm.
Financial Management 36 (1): 5–26.
De Simone, L., and R. Lester. 2018. The effect of foreign cash holdings on internal capital
markets and firm financing (October 22, 2018). Stanford University Graduate School of
Business Research Paper No. 18–32. Working paper available at
SSRN: https://ssrn.com/abstract=3199295.
De Waegenaere, A., and R. Sansing. 2008. Taxation of international investment and accounting
valuation. Contemporary Accounting Research 25 (4): 2299–2336.
Dharmapala, D., C. F. Foley, and K. J. Forbes. 2011. Watch what I do, not what I say: The
unintended consequences of the Homeland Investment Act. Journal of Finance 66 (3):
753–787.
Dong, Q., Y. Cao, X. Zhao, A. Deshmukh. 2019. Responses of US multinational firms to a
temporary repatriation tax holiday: A literature review and synthesis. Journal of
Accounting Literature 43: 108–123.
Dudley, E. 2012. Capital structure and large investment projects. Journal of Corporate Finance
18: 1168–1192.
Edwards, A., T. Kravet, and R. Wilson. 2016. Trapped cash and the profitability of foreign cash
acquisitions. Contemporary Accounting Research 33 (1): 44–77.
Ernst and Young. 1 September 2018. US proposed GILTI regulations implement international
tax reform changes. Global Tax Alert.
Fama, E. F., and K. R. French. 2001. Disappearing dividends: changing firm characteristics or
lower propensity to pay? Journal of Financial Economics 60 (1): 3–43.

36
Faulkender, M., and M. Petersen. 2012. Investment and capital constraints: Repatriations under
the American Jobs Creation Act. Review of Financial Studies 25 (11): 3351–3388.
Foley, C. F., J. C. Hartzell, S. Titman, and G. Twite. 2007. Why do firms hold so much cash? A
tax-based explanation. Journal of Financial Economics 86 (3): 579–607.
Garber, J. 2017. Here’s what America’s biggest companies plan to do with all that cash come
back to the U.S. Business Insider December 20.
GDP Ranked by Country 2019. World Population Review,
http://worldpopulationreview.com/countries/countries-by-gdp/.
Graham, J. R., M. Hanlon, and T. Shevlin. 2010. Barriers to mobility: The lockout effect of U.S.
taxation of worldwide corporate profits. National Tax Journal 63 (4): 1111–1144.
Grullon, G., B. Paye, S. Underwood, and J. P. Weston. 2011. Has the propensity to pay out
declined? Journal of Financial and Quantitative Analysis 46 (1): 1–24.
Guay, W., and J. Harford. 2000. The cash-flow permanence and information content of dividend
increases versus repurchase. Journal of Financial Economics 57: 385–415.
Hanlon, M., J. L. Hoopes, and J. B. Slemrod. 2019. Tax reform made me do it! Tax Policy and
the Economy 33: 33−80.
Hanlon, M., R. Lester, and R. Verdi. 2015. The effect of repatriation tax costs on U.S.
multinational investment. Journal of Financial Economics 116 (1): 179–196.
Harford, J., 1999, Corporate cash reserves and acquisitions, Journal of Finance 54 (6): 1969–
1997.
Hope, O.-K., and W. Thomas. 2008. Managerial empire building and firm disclosure. Journal of
Accounting Research 46 (3): 591–626.
Jagannathan, M., C. Stephens, and M. Weisbach. 2000. Financial flexibility and the choice
between dividends and stock repurchases. Journal of Financial Economics. 57 (3): 355–
384.
Jensen, M. 1986. Agency costs of free cash flow, corporate finance and takeovers. The American
Economic Review 76 (2): 323–329.
Laplante S., K., and W. L. Nesbit. 2017. The relation among trapped cash, permanently
reinvested earnings, and foreign cash. Journal of Corporate Finance 44: 126–148.
Lattman, P., and P. Eavis. 2013. To satisfy its investors, cash-rich Apple borrows money. New
York Times, April 30.
Mead, C., and D. Kucera. 2012. eBay skirts cash repatriation with bond issue: Corporate finance.
Bloomberg.
Mnuchin, S. 2019. Steven Mnuchin: The Tax Cuts and Jobs Act is working. CNN Opinion. April
15. Available at: https://www.cnn.com/2019/04/15/opinions/tax-day-benefit-american-
family-mnuchin/index.html (last accessed 5/21/20).
Myers, S. 1984. The capital structure puzzle. Journal of Finance 39 (3): 574–592.
Myers, S., and N. Majluf. 1984. Corporate financing decisions when firms have investment
information that investors do not. Journal of Financial Economics 13 (2): 187–221.

37
Nessa, M. L. 2017. Repatriation tax costs and U.S. multinational companies’ shareholder
payouts. The Accounting Review 92 (4): 217–241.
Nevius, A. M. 2017. How tax overhaul would change business taxes. Journal of Accountancy.
PR Newswire. 2019. S&P 500 Q4 2018 buybacks set 4th consecutive quarterly record at $223
billion; 2018 sets record $806 billion. https://www.prnewswire.com/news-releases/sp-
500-q4-2018-buybacks-set-4th-consecutive-quarterly-record-at-223-billion-2018-sets-
record-806-billion-300817734.html (last accessed 5/20/20).
Shin, H., and R. M. Stulz. 1998. Are internal capital markets efficient? Quarterly Journal of
Economics 113 (2): 531–552.
Shyam-Sunders, L., and S. Myers. 1999. Testing static tradeoff against pecking order models of
capital structure. Journal of Financial Economics 51 (2): 219–244.
Slemrod, J. 1992. Do Taxes Matter? Lessons from the 1980s. The American Economic Review
82 (2): 250-256.
Slemrod, J. 2018. Is this tax reform or just confusion? Journal of Economic Perspectives 32 (4):
73–96.
Stein, J. C. 1997. Internal capital markets and the competition for corporate resources. Journal of
Finance 52 (1): 111–134.
Weston, J. F. 1970. Mergers and acquisitions in business planning, Rivista Int. Sci. Econ. Com.,
(April): 309–320.
Whitehouse.gov. 2018. The tax cuts and jobs act. Available at: https://www.whitehouse.gov/wp-
content/uploads/2018/02/WH_CuttingTaxesForAmericanWorkers_Feb2018.pdf
Williamson, O. E. 1975. Markets and hierarchies: Analysis and antitrust implications. Collier
Macmillan Publishers, Inc. New York.

38
 Appendix A

The TCJA incentivizes foreign investment in tangible property through the GILTI inclusion.
Specifically, IRC Sec 951(A) requires firms to calculate their pro rata share of tested income
from CFCs minus any tested loss from CFCs. This “net CFC tested income” is then reduced by
the “net deemed tangible income return,” which is 10 percent of CFC qualified business asset
investment. CFC QBAI is tangible property used in trade or business. Firms then receive a
deduction for 50 percent of the net GILTI inclusion, which is taxed at the U.S. corporate rate of
21 percent.
The TCJA also disincentives domestic investment in tangible property through the FDII
deduction. Specifically, IRC Sec. 250 details that firms receive a deduction for 37.5 percent of
their FDII. The FDII calculation is as follows: Deemed Intangible Income × inclusion
percentage, where the inclusion percentage is foreign-derived deduction eligible income divided
by total deduction eligible income. Deduction eligible income is a U.S. corporation’s gross
income minus specific items including the GILTI inclusion. Foreign Derived Deduction Eligible
Income is a subset of deduction eligible income from the sale of property intended for foreign
use or services provided to persons, not in the U.S. Deemed Intangible income is the excess of
Deduction eligible income over the deemed tangible income return. The deemed tangible income
return is 10 percent of qualified business asset investment, which is generally tangible domestic
assets that generate deduction eligible income.
Given the complexities of both the GILTI inclusion and the FDII deduction calculations, we
provide an illustrative example below. The firm has $700,000 of gross income of which
$150,000 is from foreign sales. The firm has one controlled foreign corporation (CFC) with an
effective foreign tax rate of 5 percent and income of $100,000. The firm also has $500,000 to
invest in tangible assets. Scenario A illustrates the post-TCJA outcome of making this
investment domestically. Scenario B illustrates the post-TCJA outcome of making this
investment in a foreign jurisdiction. Critically, the effective tax rate decreases when the firm
invests in the foreign subsidiary. 34

34
IRC Sec. 951 defines tested income, which is equivalent to gross income from a foreign CFC minus allocable
deductions. For simplicity, we assume allocable deductions are zero in this example, but we note that foreign tax
credit rules could exacerbate the effect we demonstrate when firms have large domestic research and development
costs, administrative costs, and interest expense, and they have operations in high-tax countries (Rubin 2018).

39
 Appendix A (continued)
Scenario 1: Scenario 2:
$500,000 PPE $500,000 PPE
Investment in Investment in
Domestic Foreign
Parent Subsidiary

GILTI Inclusion:
Net CFC Tested Income $100,000 $100,000
Foreign Taxes Paid (5% foreign tax rate) 5,000 5,000
Net CFC Tested Income minus Foreign Taxes Paid 95,000 95,000

CFC Qualified Business Asset Investment 0 500,000

IRC Sec. 951A QBAI Return Percentage 10% 10%
Net Deemed Tangible Income Return of CFC 0 50,000

GILTI Inclusion 95,000 45,000

GILTI Inclusion % 100% 47.4%
IRC Sec. 78 Gross Up for FTC35 5,000 2,368
Total GILTI Inclusion 100,000 47,368
50% Deduction for GILTI (50,000) (23,684)
Net GILTI Inclusion 50,000 23,684
U.S. Tax on Net GILTI Inclusion (at 21% statutory rate) 10,500 4,974

Allowable FTC (80% of IRC Sec. 78 Gross Up for FTC) (4,000) (1,895)
Residual U.S. Tax on GILTI 6,500 3,079

FDII Deduction:
U.S. Gross Income 700,000 700,000
Minus: Subpart F and GILTI inclusions (100,000) (47,368)
Deduction Eligible Income (DEI) 600,000 652,632

U.S. Qualified Business Asset Investment 500,000 0

IRC Sec. 250 QBAI Return Percentage 10% 10%
Net Deemed U.S. Tangible Income Return 50,000 0

Deemed Intangible Income 550,000 652,632

Foreign Derived Deduction Eligible Income (FDDEI) 150,000 150,000
Inclusion Percentage: FDDEI/DEI 25% 23%
Foreign Derived Intangible Income 137,500 150,000
FDII Deduction (37.5% × Foreign Derived Intang. Income) (51,563) (56,250)
Taxable Income after FDII Deduction 648,437 643,750
U.S. Tax (at 21% U.S. statutory rate) 136,172 135,188

Total U.S. Tax with GILTI and FDII 142,672 138,267

Effective U.S. Tax Rate 20.38% 19.75%

Global Effective Tax Rate 18.46% 17.91%

35
GILTI Inclusion % × Foreign Taxes Paid

40
 Appendix B
Variable Definitions

Variable Name Definition

Dividends total dividends paid scaled by end of fiscal year 2014 total assets, set equal to zero when
dividends are missing in Compustat.
Repurchases total repurchases scaled by end of fiscal year 2014 total assets. A repurchase is identified as a
positive value for purchases of common and preferred stock less any decrease in the
redemption value of preferred stock in the prior year, or minus the decrease in preferred stock
in the prior year, if the redemption value is missing.
CapEx capital expenditures scaled by end of fiscal year 2014 total assets.
Debt Reduction reduction in long-term debt caused by long-term debt maturing (being classified as a current
maturity), payments of long-term debt and the conversion of debt to stock scaled by end of
fiscal year 2014 total assets.
Post equal to one for fiscal years ending after 2017, zero otherwise
MNC 1 if company is a multinational company and zero otherwise. Companies are considered
multinational if they report non-missing foreign pretax income or foreign tax expense in any
year between 2014 and 2016, inclusive.
Foreign Cash foreign cash at the end of fiscal year 2016 scaled by end of fiscal year 2014 total assets.
Domestic Cash total cash minus foreign cash at the end of 2016, scaled by end of fiscal year 2014 total
assets.
Book ETR fiscal year 2016 current federal tax expense scaled by pretax income minus special items,
truncated at zero and one.
Cash ETR fiscal year 2016 current taxes paid scaled by pretax income minus special items, truncated at
zero and one.
Loss equal to 1 when income before extraordinary items is less than zero and equal to zero
otherwise.
Size the natural logarithm of total assets.
Earnings income before extraordinary items scaled by end of fiscal year 2014 total assets.
Cash Flow operating cash flow scaled by end of fiscal year 2014 total assets.
Leverage long term debt scaled by end of fiscal year 2014 total assets.
Total Cash total cash scaled by end of fiscal year 2014 total assets.
MTB market value of shares outstanding divided by the total book value of common equity.
Sales Growth total sales minus lagged sales, scaled by lagged sales.
CapEx 2014 total fiscal year 2014 capital expenditures scaled by end of fiscal year 2014 total assets.
R&D equal to research and development expense scaled by end of fiscal year 2014 total assets.
R&D is equal to zero when missing in Compustat.
ACQ total expenditures on mergers and acquisitions scaled by end of fiscal year 2014 total assets.
ACQ is equal to zero when missing in Compustat.
Firm Age the natural logarithm of the number of years the firm has been in Compustat.
RE/BV ratio of retained earnings to book value of common equity.
Returns monthly compounded stock return for the two years prior to period t.
Options the annual percentage change in total diluted shares outstanding of firm as if no repurchases
were made during year t.
St. Dev. Earnings the standard deviation of earnings for the current and previous 4 years.

41
Table 1
Sample Selection

August 2020 Compustat download meeting the following criteria: 12,850

December 31st Fiscal Year End
Non-Financial Firms
Location of Incorporate = United States of America
Non-missing Assets, Book Value of Equity, Stock Price and Earnings
Less:
Firm-years without assets in fiscal year 2014 used as the scalar (2,300)
Missing Data to calculate Returns (2,359)
Require firms to have observations in the pre- and post-TCJA periods (2,623)
Missing Compustat or other lagged variables (628)
DiD Sample:
2015 - 2019 4,940

MNC Sample:
2015 - 2019 3,270
Less:
Missing Foreign Cash Data (1,526)
Final MNC Sample 1,744

____________
This table presents the selection of the sample used in the study.

42
Table 2
Parallel Trends Assumption
Panel A: Time Trend Charts

Dividends Repurchases
0.025 0.07
0.06
0.02
0.05
0.015 0.04
0.01 0.03
0.02
0.005
0.01
0 0
2010 2011 2012 2013 2014 2015 2016 2010 2011 2012 2013 2014 2015 2016

MNC Domestic MNC Domestic

CapEx Debt Reduction

0.14 0.35
0.12 0.3
0.1 0.25
0.08 0.2
0.06 0.15
0.04 0.1
0.02 0.05
0 0
2010 2011 2012 2013 2014 2015 2016 2010 2011 2012 2013 2014 2015 2016

MNC Domestic MNC Domestic

43
Panel B: Time Trend Regression
(1) (2) (3) (4)
Dividends Repurchases CAPX Debt Reduction
Time Trend (α1) 0.000*** 0.001*** 0.002*** 0.008***
(5.63) (6.30) (6.70) (5.38)
MNC (α2) 0.005*** 0.007** -0.007 -0.003
(3.01) (1.99) (-1.31) (-0.11)
MNC * Time Trend (α3) -0.000 0.000 -0.001 0.000
(-0.70) (1.21) (-1.54) (0.01)
Intercept (α0) 0.008*** 0.015*** 0.024 0.072***
(7.04) (5.91) (1.49) (3.22)

N 9,899 9,899 9,899 9,899

adj. R-sq. 0.0095 0.0179 0.0302 0.0064
Log Likelihood 18738.465 10834.188 8443.488 -10983.946

____________
Panel A charts the sample average of the dependent variables from 2010 to 2016. Panel B presents regression
coefficient estimates from Equation (3) used to test the parallel trends assumption of the difference-in-differences
analysis. Variables are defined in Appendix B.

44
Table 3
Summary Statistics

N Mean St. Dev. Q1 Median Q3

Dividends 4,940 0.012 0.019 0.000 0.000 0.018
Repurchases 4,940 0.027 0.044 0.000 0.004 0.033
CapEx 4,940 0.051 0.050 0.016 0.033 0.068
Debt Reduction 4,940 0.173 0.431 0.000 0.037 0.172
Post 4,940 0.500 0.500 0.000 1.000 1.000
MNC 4,940 0.662 0.473 0.000 1.000 1.000
Foreign Cash 1,744 0.078 0.075 0.023 0.052 0.103
Domestic Cash 1,744 0.086 0.104 0.011 0.038 0.126
Book ETR 4,940 0.194 0.205 0.000 0.182 0.293
Cash ETR 4,940 0.151 0.190 0.000 0.098 0.236
Loss 4,940 0.358 0.480 0.000 0.000 1.000
Size 4,940 6.980 1.936 5.567 7.076 8.409
Earnings 4,940 -0.044 0.430 -0.049 0.034 0.086
Cash Flow 4,940 0.068 0.170 0.028 0.091 0.157
Leverage 4,940 0.229 0.196 0.027 0.211 0.363
Total Cash 4,940 0.228 0.265 0.042 0.120 0.304
MTB 4,940 3.595 3.525 1.351 2.464 4.554
Sales Growth 4,940 0.069 0.218 -0.045 0.042 0.148
CapEx 2014 4,940 0.050 0.050 0.015 0.032 0.064
R&D 4,940 0.069 0.123 0.000 0.006 0.076
ACQ 4,940 0.033 0.073 0.000 0.000 0.018
Firm Age 4,940 3.586 0.539 3.258 3.664 3.951
RE/BV 4,940 -0.586 2.979 -0.816 0.311 0.842
Returns 4,940 0.039 0.385 -0.244 0.020 0.296
Options 4,940 0.027 0.084 -0.016 0.004 0.035

St. Dev. Earnings 4,940 0.062 0.070 0.016 0.034 0.077

____________
This table presents the descriptive statistics. Variables are defined in Appendix B.

45
Table 4
Multinational and Domestic Firm Sample: Difference-in-Differences
(1) (2) (3) (4)
Dividends Repurchases CapEx Debt Reduction
Post (α1) 0.001* (1.74) -0.002 (-0.61) 0.003 (1.63) 0.042** (1.98)
MNC × Post (α2) -0.001 (-1.46) 0.005 (1.36) 0.000 (0.26) 0.006 (0.26)
MNC (α3) 0.000 (1.03) -0.003 (-1.35) -0.001 (-0.82) -0.013 (-0.74)
Book ETR (α4) -0.001 (-1.03) -0.000 (-0.07) -0.002 (-0.63) 0.034 (0.84)
Cash ETR (α5) 0.004 (1.25) -0.000 (-0.13) 0.000 (0.02) -0.047* (-1.85)
Loss (α6) -0.001 (-1.41) -0.004*** (-2.61) -0.004*** (-2.69) -0.054** (-2.14)
Size (α7) 0.000*** (3.43) 0.002*** (3.18) 0.000 (0.06) -0.012*** (-2.78)
Earnings (α8) -0.001* (-1.95) 0.003 (0.85) -0.006*** (-2.68) -0.186 (-1.42)
Cash Flow (α9) 0.004*** (3.16) 0.047*** (3.96) 0.031*** (4.27) 0.201 (1.28)
Leverage (α10) -0.003*** (-3.11) -0.015*** (-3.04) 0.001 (0.32) 0.335*** (7.92)
Total Cash (α11) 0.001 (0.50) 0.016*** (2.63) 0.009*** (3.01) -0.183*** (-4.44)
MTB (α12) 0.000 (0.89) 0.002*** (5.14) 0.000*** (2.68) -0.004** (-2.11)
Sales Growth (α13) 0.000 (0.42) 0.005 (1.12) 0.028*** (8.49) 0.015 (0.41)
CapEx 2014 (α14) -0.007* (-1.86) -0.044** (-2.48) 0.812*** (47.15) 0.420** (2.13)
R&D (α15) -0.002 (-1.57) -0.010 (-0.81) 0.001 (0.10) 0.129 (1.24)
ACQ (α16) -0.004* (-1.74) -0.048*** (-2.65) 0.009 (1.05) 0.405*** (4.53)
Firm Age (α17) 0.001 (1.36) 0.001 (0.42) -0.015 (-1.00)
RE/BV (α18) -0.000 (-0.06) 0.000 (0.92) 0.005* (1.92)
Returns (α19) 0.000 (0.51) -0.006** (-2.46) -0.015 (-0.71)
Options (α20) 0.011*** (3.03) -0.012 (-1.04) 0.164* (1.69)
St. Dev. Earnings (α21) -0.002 (-0.64) 0.025 (1.64) 0.340** (2.32)
Dividends Lag (α22) 0.957*** (57.24) 0.045 (0.56) -0.433 (-1.16)
Repurchase Lag (α23) 0.005 (0.83) 0.519*** (12.87) 0.060 (0.41)
Intercept (α0) -0.005** (-2.34) -0.015 (-1.58) 0.011*** (3.00) 0.076 (1.14)
N 4,940 4,940 4,940 4,940
adj. R2 0.8657 0.4781 0.7367 0.1401
____________
This table presents coefficient estimates from Equation (1) for a sample of domestic-only and multinational firms. Dependent variables in columns 1, 2, 3 and 4 are
Dividends, Repurchases, CapEx and Debt Reduction, respectively. T-statistics are presented in parentheses to the right of the coefficient estimate. Industry fixed effects
are included in all models. Standard errors are robust standard errors. ***, **, ** represent 1 percent, 5 percent, 10 percent two-tailed statistical significance,
respectively. Variables are defined in Appendix B.

46
Table 5
Financial Constraint Cross-Section

(1) (2) (3) (4)

Dividends Repurchases CapEx Debt Reduction
Panel A: Financial Constraint=High
Post (α1) 0.001 (1.21) 0.002 (0.80) 0.010** (2.34) 0.397** (2.29)
MNC × Post (α2) -0.001 (-1.45) 0.008** (1.98) -0.000 (-0.08) -0.386* (-1.90)
MNC (α3) 0.001* (1.79) -0.005** (-1.98) -0.000 (-0.07) 0.155 (1.16)
Intercept (α0) -0.009** (-2.17) 0.015 (1.27) -0.030*** (-2.91) 0.716* (1.84)
Controls Yes Yes Yes Yes
N 844 844 844 844
adj. R2 0.5060 0.3369 0.6881 0.4807

Panel B: Financial Constraint=Low

Post (α1) 0.001* (1.67) -0.002 (-0.56) -0.001 (-0.62) 0.043 (1.34)
MNC × Post (α2) -0.001 (-0.82) 0.002 (0.55) 0.003 (1.35) 0.003 (0.09)
MNC (α3) 0.001* (1.82) 0.001 (0.20) -0.002 (-1.57) -0.028 (-1.27)
Intercept (α0) -0.005 (-1.30) -0.023* (-1.90) 0.008 (0.77) 0.228** (2.12)
Controls Yes Yes Yes Yes
N 2,964 2,964 2,964 2,964
adj. R2 0.8930 0.5006 0.7638 0.2246

p-value: 0.094* 0.319

F-test α2 across Panels
____________
This table presents coefficient estimates from Equation (1) for a sample of domestic-only and multinational firms.
Dependent variables in columns 1, 2, 3 and 4 are Dividends, Repurchases, CapEx and Debt Reduction, respectively.
Panel A (Panel B) includes firms with high (low) financial constraints measured using the Faulkender and Petersen
(2012) measure of financial constraints. T-statistics are presented in parentheses to the right of the coefficient
estimate. Industry fixed effects are included in all models. Standard errors are robust standard errors. ***, **, **
represent 1 percent, 5 percent, 10 percent two-tailed statistical significance, respectively. Variables are defined in
Appendix B.

47
Table 6
Multinational Firm Sample – Foreign Cash
(1) (2) (3) (4)
Dividends Repurchases CapEx Debt Reduction
Post (α1) 0.000 (0.50) -0.002 (-1.04) 0.000 (0.38) 0.076*** (3.57)
Foreign Cash × Post (α2) -0.001 (-0.14) 0.054** (2.08) 0.007 (0.74) -0.292 (-1.16)
Foreign Cash (α3) 0.008 (1.44) -0.000 (-0.02) -0.018*** (-2.62) -0.157 (-0.80)
Domestic Cash (α4) -0.003 (-1.11) -0.010 (-0.87) 0.003 (0.50) -0.726*** (-8.74)
Book ETR (α5) 0.000 (0.47) 0.001 (0.37) -0.002 (-1.22) -0.034 (-1.02)
Cash ETR (α6) -0.000 (-0.20) 0.002 (0.67) 0.002 (1.01) -0.013 (-0.41)
Loss (α7) -0.001 (-1.38) -0.002 (-1.12) -0.002* (-1.79) -0.019 (-0.83)
Size (α8) 0.000** (2.38) 0.001* (1.84) -0.000 (-0.06) -0.004 (-0.65)
Earnings (α9) -0.002 (-1.45) 0.014 (1.31) -0.001 (-0.16) -0.092 (-1.12)
Cash Flow (α10) 0.014*** (3.48) 0.111*** (6.44) 0.032*** (3.77) 0.202 (1.53)
Leverage (α11) -0.004*** (-2.70) -0.015** (-2.40) -0.005* (-1.79) 0.166*** (3.03)
MTB (α12) 0.000*** (2.69) 0.001*** (2.71) 0.000* (1.92) 0.000 (0.11)
Sales Growth (α13) -0.001 (-0.91) 0.007 (1.13) 0.026*** (7.54) -0.029 (-0.53)
CapEx 2014 (α14) -0.025*** (-3.88) -0.077** (-2.29) 0.837*** (38.66) 0.230 (0.92)
R&D (α15) -0.000 (-0.10) 0.057*** (2.73) 0.003 (0.33) 0.029 (0.25)
ACQ (α16) -0.001 (-0.66) -0.045*** (-4.75) 0.007 (1.40) 0.318*** (2.60)
Firm Age (α17) -0.000 (-0.22) -0.002 (-0.85) -0.049** (-2.53)
RE/BV (α18) 0.000 (0.67) 0.002** (2.37) 0.007 (1.19)
Returns (α19) -0.000 (-0.03) -0.013*** (-5.50) 0.018 (0.99)
Options (α20) 0.011** (1.99) -0.115*** (-4.55) 0.439** (2.37)
St. Dev. Earnings (α21) -0.003 (-0.58) 0.065*** (2.63) 0.131 (0.74)
Dividends Lag (α22) 0.965*** (40.02) -0.067 (-1.11) -0.662 (-1.61)
Repurchase Lag (α23) -0.001 (-0.34) 0.491*** (15.68) -0.142 (-0.75)
Intercept (α0) -0.001 (-0.19) 0.004 (0.36) 0.009 (1.50) 0.251*** (3.27)
N 1,744 1,744 1,744 1,744
adj. R2 0.8848 0.5644 0.7685 0.1634
____________
This table presents coefficient estimates from Equation (2) for a sample of multinational firms. Dependent variables in columns 1, 2, 3 and 4 are Dividends,
Repurchases, CapEx and Debt Reduction, respectively. T-statistics are presented in parentheses to the right of the coefficient estimate. Industry fixed effects are
included in all models. Standard errors are robust standard errors. ***, **, ** represent 1 percent, 5 percent, 10 percent two-tailed statistical significance,
respectively. Variables are defined in Appendix B.

48
Table 7
Repatriation Cost Cross-Section
(1) (2) (3) (4)
Dividends Repurchases CapEx Debt Reduction
Panel A: High Repatriation Costs
Post (α1) 0.000 (0.69) -0.005 (-1.49) 0.001 (0.61) 0.034 (1.51)
Foreign Cash × Post (α2) -0.000 (-0.04) 0.070** (2.02) 0.006 (0.52) -0.208 (-1.43)
Foreign Cash (α3) 0.007 (1.44) -0.008 (-0.31) -0.013* (-1.79) -0.482*** (-4.25)
Intercept (α0) -0.003 (-0.68) -0.009 (-0.50) 0.008 (0.95) 0.174* (1.85)
Controls Yes Yes Yes Yes
N 908 908 908 908
adj. R2 0.9058 0.5759 0.7819 0.3670

Panel B: Low Repatriation Costs

Post (α1) 0.000 (0.27) 0.001 (0.29) 0.000 (0.18) 0.100** (2.56)
Foreign Cash × Post (α2) -0.004 (-0.33) 0.024 (0.58) 0.006 (0.33) -0.476 (-0.65)
Foreign Cash (α3) 0.007 (0.54) 0.017 (0.56) -0.024 (-1.54) 0.533 (0.91)
Intercept (α0) 0.005 (0.54) 0.013 (0.68) 0.008 (0.93) 0.029 (0.24)
Controls Yes Yes Yes Yes
N 836 836 836 836
adj. R2 0.8294 0.4997 0.7481 0.1289

p-value: 0.092*
F-test α2 across Panels

____________
This table presents coefficient estimates from Equation (2) for a sample of multinational firms. Dependent variables
in columns 1, 2, 3 and 4 are Dividends, Repurchases, CapEx and Debt Reduction, respectively. Panel A (Panel B)
includes firms with high (low) repatriation costs measured using the Nessa (2017) measure of repatriation costs. T-
statistics are presented in parentheses to the right of the coefficient estimate. Industry fixed effects are included in all
models. Standard errors are robust standard errors. ***, **, ** represent 1 percent, 5 percent, 10 percent two-tailed
statistical significance, respectively. Variables are defined in Appendix B.

49
Table 8
Estimated Trapped Cash Cross-Section
(1) (2) (3) (4)
Dividends Repurchases CAPX Debt Reduction
Panel A: High Trapped Cash Estimate
Post (α1) 0.000 (0.07) -0.005 (-1.20) 0.001 (0.47) 0.100*** (3.42)
Foreign Cash × Post (α2) 0.003 (0.51) 0.077* (1.72) 0.016 (1.24) -0.291 (-1.12)
Foreign Cash (α3) 0.002 (0.38) 0.059*** (2.69) -0.026*** (-2.78) -0.063 (-0.35)
Intercept (α0) 0.002 (0.47) -0.028* (-1.77) -0.004 (-0.25) 0.105 (0.91)
Controls Yes Yes Yes Yes
N 708 708 708 708
adj. R2 0.7695 0.5991 0.7389 0.1317

Panel B: Low Trapped Cash Estimate

Post (α1) 0.000 (0.17) -0.006 (-1.05) -0.000 (-0.04) 0.048* (1.92)
Foreign Cash × Post (α2) 0.006 (0.33) 0.087 (1.29) 0.006 (0.53) -0.171 (-1.30)
Foreign Cash (α3) 0.010 (0.67) -0.017 (-0.35) -0.011 (-1.38) -0.375*** (-3.10)
Intercept (α0) -0.001 (-0.07) 0.032 (1.39) -0.001 (-0.14) 0.151 (1.41)
Controls Yes Yes Yes Yes
N 708 708 708 708
adj. R2 0.8127 0.5244 0.8187 0.4573

p-value: 0.453
F-test α2 across Panels
____________
This table presents coefficient estimates from Equation (2) for a sample of multinational firms. Dependent variables
in columns 1, 2, 3 and 4 are Dividends, Repurchases, CapEx and Debt Reduction, respectively. Panel A (Panel B)
includes firms with high (low) estimated trapped foreign cash (using models from Laplante and Nesbitt, 2017 to
estimate). T-statistics are presented in parentheses to the right of the coefficient estimate. Industry fixed effects are
included in all models. Standard errors are robust standard errors. ***, **, ** represent 1 percent, 5 percent, 10
percent two-tailed statistical significance, respectively. Variables are defined in Appendix B.

50
Table 9
GILTI

(1) (2) (3) (4) (5) (6)

Foreign Domestic Foreign ΔPPE Foreign ΔPPE Domestic ΔPPE Domestic ΔPPE
ΔPPE ΔPPE High GILTI Low GILTI High GILTI Low GILTI
Post (α1) -0.014* (-1.93) -0.005 (-0.55) -0.015 (-1.59) -0.000 (-0.00) -0.012 (-1.01) 0.011 (0.57)
Foreign Cash×Post (α2) 0.025 (1.15) 0.020 (0.70) 0.044* (1.89) -0.064 (-1.25) 0.029 (0.91) 0.001 (0.02)
Foreign Cash (α3) -0.013 (-0.97) -0.017 (-1.03) -0.015 (-0.98) -0.010 (-0.26) -0.025 (-1.40) 0.003 (0.06)
Domestic Cash (α4) -0.019 (-1.43) 0.031* (1.92) -0.008 (-0.51) -0.060* (-1.93) 0.031* (1.81) 0.000 (0.01)
Book ETR (α5) 0.004 (1.19) -0.004 (-1.35) 0.005 (1.47) 0.002 (0.28) -0.002 (-0.51) -0.005 (-0.99)
Cash ETR (α6) -0.003 (-0.97) -0.002 (-0.49) 0.002 (0.40) -0.009 (-1.49) -0.000 (-0.04) -0.007 (-1.51)
Loss (α7) 0.000 (0.03) -0.002 (-0.70) 0.000 (0.10) 0.002 (0.42) -0.004 (-1.54) 0.005 (1.14)
Size (α8) 0.000 (0.93) -0.000 (-0.02) 0.001 (0.85) -0.001 (-0.51) -0.000 (-0.69) 0.001 (0.45)
Foreign Earnings (α9) 0.070*** (2.82) -0.006 (-0.21) 0.092*** (3.00) 0.084 (1.58) 0.001 (0.04) 0.037 (0.64)
Domestic Earnings (α10) 0.012 (0.71) 0.036* (1.85) 0.008 (0.41) 0.020 (0.52) 0.029 (1.33) 0.073* (1.93)
Cash Flow (α11) -0.006 (-0.38) -0.028 (-1.63) -0.020 (-1.04) 0.035 (1.16) -0.047** (-2.40) -0.006 (-0.18)
Leverage (α12) -0.011** (-2.36) -0.006 (-1.18) -0.010* (-1.82) -0.014 (-1.41) -0.007 (-1.22) -0.012 (-0.98)
MTB (α13) -0.000 (-0.63) 0.001** (2.17) -0.000 (-1.02) -0.000 (-0.49) 0.001** (2.12) 0.001 (1.11)
Sales Growth (α14) 0.025*** (3.88) 0.037*** (4.63) 0.033*** (3.97) 0.011 (1.05) 0.028*** (3.03) 0.053*** (3.46)
CapEx 2014 (α15) 0.163*** (4.60) 0.261*** (6.13) 0.227*** (4.79) 0.030 (0.49) 0.240*** (5.01) 0.262*** (3.02)
R&D (α16) 0.004 (0.26) -0.029 (-1.64) 0.010 (0.59) 0.010 (0.25) -0.007 (-0.37) -0.044 (-1.01)
ACQ (α17) 0.072*** (5.45) 0.101*** (6.76) 0.069*** (4.45) 0.078*** (2.93) 0.100*** (6.32) 0.097*** (2.98)
Depreciation (α18) -0.033 (-0.73) -0.051 (-0.92) -0.090 (-1.55) -0.006 (-0.07) -0.028 (-0.46) -0.079 (-0.68)
Intercept (α0) -0.008 (-1.12) -0.002 (-0.24) -0.005 (-0.52) 0.008 (0.67) 0.005 (0.37) -0.006 (-0.41)
N 1,139 1,139 747 392 747 392
adj. R2 0.2161 0.2411 0.2398 0.1885 0.2493 0.2894
____________
This table presents coefficient estimates from Equation (2) for a sample of domestic-only and multinational firms. Dependent variables in columns 1 and 2 are
the change in Foreign PPE and Domestic PPE, respectively. The dependent variable for column 3 (column 4) is the change in foreign PPE for high GILTI (low
GILTI) multinational firms. The dependent variable for column 4 (column 5) is the change in domestic PPE for high GILTI (low GILTI) multinational firms. T-
statistics are presented in parentheses to the right of the coefficient estimate. Industry fixed effects are included in all models. Standard errors are robust standard
errors. ***, **, ** represent 1 percent, 5 percent, 10 percent two-tailed statistical significance, respectively. Variables are defined in Appendix B.

51