Corporate Tax Havens and Shareholder Value

Transcription

Corporate Tax Havens and Shareholder Value:
Evidence from Tax Information Exchange Agreements
Morten Bennedsen†
Stefan Zeume ‡
INSEAD
University of Michigan
First Draft: November 18, 2014
Abstract
Using a novel hand-collected dataset of 13,639 publicly listed firms from 52 countries and their international
subsidiaries, we provide evidence that subsidiaries in tax havens are used for tax saving motives and for
entrenchment motives. Consistent with the tax motive, we find that a 1 percentage point reduction in firms’
home-country corporate tax rate is associated with a 1.2% increase in value of firms without tax haven
subsidiary while firms with tax haven subsidiaries are unaffected. To provide direct evidence for the
entrenchment motive, we exploit the passage of Tax Information Exchange Agreements (TIEAs) between
countries and tax havens as a shock to firms’ transparency. The implementation of a TIEA increases average
shareholder value by 3.1 percent, a result confirmed in event studies of daily returns around the passage of
TIEAs. Even though tax agreements are endorsed by investors, we show that many firms engage in haven
hopping: They move subsidiaries from tax havens that entered TIEAs to tax havens that did not.
This paper also provides evidence on the use of tax haven subsidiaries by country and firm characteristic: The
use of tax haven subsidiaries is more prevalent in countries with high income tax rates, strong minority
shareholder protection, low corruption levels, and low perceived tax evasion. At the firm level, the use of tax
haven subsidiaries is positively correlated with firm size, effective tax rate, leverage, cash holdings, dividend
payout, and intangible assets (R&D, patents, trademarks) and negatively with corporate taxes faced abroad.
Keywords: Tax havens, firm value, corporate governance, entrenchment
______________________
Preliminary. Please do not cite. We have benefitted from comments offered by Denis Gromb, Maria Guadalupe,
Charlie Hadlock, and Urs Peyer. We thank the participants of brownbag seminars at INSEAD (Economics, Finance)
and the University of Michigan (Finance).
†
André and Rosalie Hoffmann Chaired Professor of Family Enterprise and Professor of Economics, INSEAD,
[email protected]
‡
Assistant Professor, Department of Finance, University of Michigan, [email protected]
1
1.
Introduction
A tax haven is a state or territory where corporate and personal tax rates are so low that
foreign companies – or individuals – have incentives to establish shell companies to shield their
income from higher tax liabilities at home. In 2007, the OECD estimated that between USD 5
trillion and USD 7 trillion were held offshore. The tax justice network estimates that between
USD 21 trillion and USD 32 trillion were held offshore in 2012. The US organization Citizens
for Tax Justice estimates that at least 362 out of the top 500 US firms are active in tax havens
and that the thirty US companies with most offshore investment collectively book USD 1.2
trillion in tax havens. In early 2014, the press uncovered prominent tax schemes involving
companies such as Apple and Starbucks. In November 2014, the Luxembourg Tax Leak brought
to light that more than 374 large international companies had made private arrangements with the
Luxembourgish tax authority to pay less than 1% in tax – the official Luxembourgish corporate
tax rate is 29%. While this will likely become a major policy issue in the European Union in the
coming years, the US has shown a strong interest in regulating the use of offshore tax havens
ever since first signing tax information exchange agreements with tax havens more than 15 years
ago.1
Given the large amount of tax shielding that takes place in international corporations and the
enormous attention this topic has received in policy and media spheres, it is surprising how little
is known about types of firms that use tax haven subsidiaries and their underlying motives. This
paper fills this gap by providing evidence from a novel dataset covering 13,639 publicly listed
firms from 52 countries and their circa 164,000 domestic and foreign subsidiaries.
The most obvious and cited motive for establishing subsidiaries in tax havens is to minimize
overall tax payments. Multinational firms can reduce their tax bill by establishing subsidiaries in
tax havens. Typically, firms pursue little operational activities in tax havens yet tax reductions
can be obtained e.g. by transferring intangible assets such as patents, licenses, brands, or
goodwill to tax havens. Operational subsidiaries in high tax environments as well as the parent
firm are charged for using these assets, reducing firms’ tax bill.
Besides this tax motive, companies may have an entrenchment motive for establishing
subsidiaries in tax havens. Tax havens are typically not very transparent and it may be hard for
1
The public debate largely focuses on the costs of tax havens for high tax countries yet some studies show that low tax regimes
have positive spillovers on nearby high tax regimes (e.g. foreign direct investment, subsidiary investment and growth, and
mitigation of tax competition – see Dharmapala 2008, Desai, Foley and Hines 2004 & 2006A, and Slemrod and Wilson 2006).
2
shareholders to clearly monitor how a firm’s assets are used in these territories. For instance, the
controlling managers may have an interest in piling cash in tax havens to finance future activities
beyond what shareholders find optimal. Hanlon, Lester and Verdi (2014) document that firms
hold cash in tax havens and use it for inefficient acquisitions. Entrenchment can also involve
resource transfers out of the corporation through third parties. These are often non-transparent
for non-controlling owners. For example, Enron CFO Andrew Fastow created a staggering 892
offshore subsidiaries, 692 of these were created in the Cayman Islands alone. Not only did this
network of subsidiaries allow Enron to avoid paying taxes but in the court cases after the Enron
collapse it was also revealed that Fastow and his friends were able to transfer considerable
resources to companies that they controlled outside Enron. Stealing and tunneling resources from
shareholders to controlling managers is easier in environments that lack transparency and
enforcement as shown in the context of Russia by Desai, Dyck and Zingales (2007) and Mironov
(2013).
In this paper, we first provide descriptive evidence for the use of tax haven and then offer
direct evidence for the tax and entrenchment motives. In order to test for the tax motive, we
use historical changes in corporate tax levels. We find that a 1 percentage point reduction in
firms’ home-country corporate tax rate is associated with a 1.2% increase in value of firms
without tax haven subsidiaries while firms with tax haven subsidiaries are unaffected.
To provide direct evidence for the entrenchment motive, we exploit the passage of Tax
Information Exchange Agreements (TIEAs) between countries and tax havens as a shock to
firms’ transparency. A TIEA is a bilateral agreement between countries to exchange information
relevant in civil or criminal tax investigations against individuals or firms. Importantly, TIEAs
do not change the tax rules in either of the involved countries.
TIEAs provide a natural experiment to separate the tax motive from the entrenchment motive.
As for the tax motive, TIEAs could leave firm value unaffected or lead to a decline in firm value:
If firms indeed followed the tax rules in both countries, passage of a TIEA should have no effect
on tax payments and shareholder value for firms with subsidiaries in a tax haven. If a TIEA
increased a country’s ability to judge whether within firm transfer pricing and other activities are
consistent with national tax rules or not, the passage of a TIEA might increase the amount of
taxes due at home, reducing shareholder value.
3
The entrenchment motive, however, suggests that the passage of TIEAs increases shareholder
value: Suppose tax havens were used to shield assets from shareholders, such as through piling
up free cash for later or inefficient use2 or tunneling assets out of the company through third
party companies that are controlled by insiders of the multinational firm or their close friends or
families. TIEAs make the actual transfers of cash and assets costlier to insiders and more
transparent to headquarter country tax authorities during tax investigations. Thus, it becomes
more difficult and costly for controlling owners and managers to use tax havens to hide or steal
cash. If entrenchment is an important motive for the use of tax havens, shareholders should
endorse TIEAs affecting firms headquartered in one signatory country and operating subsidiaries
in the other signatory country (the tax haven).
TIEAs provide an ideal experimental setting not only because they enable us to separate the
two main motives but also because they are bilateral, affecting some firms headquartered in a
country but not others. More than 500 such agreements were passed at different points in time
over the past 12 years, providing a lot of variation across time and countries. Above all, it is
relatively easy to identify counterfactuals such as publicly traded companies that are
headquartered in one signatory country but that do not have tax haven subsidiaries or that have
subsidiaries in unaffected tax havens.3
Using annual data, we show that the implementation of a TIEA increases average shareholder
value (measured by Tobin’s Q) by 3.1 percent on average. We re-confirm this result using daily
stock returns around the signing of TIEAs. This result provides strong evidence that the use of
tax havens does come with costs to shareholders: Tax havens may be used for entrenchment and
even stealing; shareholders reward increased transparency resulting from the passage of TIEAs.
In order to rule out that the positive shareholder effect is driven by a reduction on tax
uncertainty or by shareholders rewarding management for making future tax payments more
transparent, we examine how controlling owners and managers adjust the corporate structure.
We show that firms engage in haven hopping: They move their subsidiaries strategically from
tax havens that entered TIEAs to tax havens that did not. Haven hopping is consistent with the
2
Hanlon, Lester and Verdi (2014) show that US firms use cash piled in tax havens for inefficient acquisitions.
Few papers have exploited the passage of TIEAs. Johannesen and Zucman (2014) show that after the passage of TIEAs, bank
deposits are shifted from affected to unaffected tax havens. German foreign direct investment and the number of German
subsidiaries in tax havens declined after Germany passed TIEAs (Braun and Weichenrieder 2014). Bilicka and Fuest (2014)
document that TIEAs are typically passed between countries and tax havens with stronger economic links.
3
4
entrenchment hypothesis but is hard to explain from the tax saving motive, given that
shareholder value increase after the implementation of a TIEA. As a side effect, this may mean
that TIEAs benefit the least compliant tax havens. To further ensure that our result is not driven
by a reduction in uncertainty, we show that firms’ betas do not decline significantly.
Our paper is related to a number of recent papers studying the use of tax haven subsidiaries by
multinational corporations. These papers are largely country specific and provide somewhat
mixed evidence. Desai, Foley and Hines (2006B) study US firms: large firms, more international
firms, and firms with extensive intrafirm trade and high R&D intensity are more likely to use tax
havens. US firms use large tax havens to reallocate taxable income and small tax havens to avoid
taxation of foreign income in the US. Markle and Robinson (2012), studying 8,000 multinational
firms from 28 countries, confirm the result on firm size but find the opposite result for R&D
intensive firms. They also document a negative correlation between the tax rate in non-tax haven
subsidiaries and the use of tax havens. In Germany, however, foreign income is tax exempt and
consequently, manufacturing firms are more likely to use tax havens when facing high foreign
corporates taxes (Gumpert, Hines and Schnitzer 2011). Last but not least, constrained firms are
less likely to use tax haven subsidiaries (Dyreng and Markle 2013). We contribute to this
literature by documenting causal evidence on the tax savings motive and the entrenchment
motive, by collecting a larger dataset of 13,639 firms that covers 52 countries, and by exploiting
various sources of data on transferrable assets to obtain more clarity on the link between
transferrable assets and the use of tax havens. Also, our results are robust to various definitions
of tax havens.
In terms of tax enforcement, the paper closest to ours is Desai, Dyck, Zingales (2007) who
show that stronger tax enforcement reduces income diversion by insiders. Their model features a
trade-off between tax enforcement’s impact on taxes paid and the cost of income diversion to
insiders. Empirically, Desai, Dyck and Zingales (2007) show that the Russian oil firm Sibneft
earns positive abnormal returns over five tax enforcement actions in Russia, indicating that tax
enforcement can have a positive impact on firm value. Linking this finding to private benefits of
control, the authors then document that tax enforcement actions aimed at extractive industries
lead to a reduction in the control premium compared to the reduction in control premium
experienced by firms in other industries. This confirms a notion made by Dyck and Zingales
(2004): The premium paid in block sales is negatively related to the strength of tax enforcement.
5
These studies are supported by Mironov (2013): In Russia, tax enforcement correlates positively
with operating performance. Mironov (2014) cleverly measures tax enforcement at regional level
by standardizing the number of tax employees by the number of firms.
In the corporate finance literature, Heider and Ljungqvist (2012) document an asymmetric
relation between changes in the tax rate and adjustment in leverage. Firms lever up subsequent to
increases in the tax rate but do not lever down subsequent to tax reductions. The relation between
tax sheltering and leverage has been studied by Graham and Tucker (2006) who, on a carefully
collected sample of 44 tax sheltering cases, show that firms engaged in tax sheltering have lower
leverage than matched firms at the time of detection and prior to detection of sheltering activity.
A large accounting literature has examined the link between firm-specific accounting
measures of tax avoidance and firm value. Representatively, Desai and Dharmapala (2005) show
that tax avoidance (measured at the firm level by the book-tax gap) has no effect on firm value
on average but a positive effect among strongly governed firms. These papers differ in that they
look at firms’ selected level of tax avoidance rather than tax enforcement.
Two challenges with studying tax regimes are measurement across countries and finding
shocks that affect a subsample of firms for identification. Dyck and Zingales (2004) resort to La
Porta et al.’s (1996) tax compliance measure that is based on surveys published in the Global
Competitiveness Report for 1996. To construct the measure, survey participants are asked to
assess the level of tax compliance in their country. While subsequent reports allow constructing a
time series, the measure is related to all firms in a country, making it difficult to identify a
suitable control group. Also, comparability of changes across countries is a challenge. Desai,
Dyck and Zingales (2007) resort to tax enforcement actions under Putin in Russia. In their results
on dual class shares, they work with a clean but small sample of 15 observations that they further
split in half.
The rest of this paper is organized as follows: In Section 2, we discuss the definition of tax
havens and describe our data. In Section 3, we provide descriptive statistics for the use of tax
havens at country and industry level. In Section 4, we provide firm level evidence on the use of
tax haven subsidiaries. In Section 5, we study changes in corporate tax rates as a shock to the
value of having tax haven subsidiaries. Section 6 provides our most direct evidence for the
entrenchment motive for use of tax havens: Studying TIEAs, we show that shareholders rewards
firms when TIEAs are introduced. Section 7 concludes.
6
2. Tax Havens, Subsidiary Data, and Accounting Data
In this section, we define tax havens, describe our source of subsidiary data, and define key
firm-level accounting controls.
2.1 Tax havens
A typical definition of tax havens makes reference to sovereign countries or non-sovereign
territories where certain taxes are low or zero (Dharmapala and Hines 1996). Such sovereign or
non-sovereign jurisdictions allow firms to engage in tax avoidance. By that definition, many tax
haven lists exist. We focus on five commonly cited lists. Table 1 lists all sovereign countries or
non-sovereign territories that fall under the definition of a Tax Haven by any of these lists.
--- TABLE 1 ABOUT HERE ---
First, countries and non-sovereign states that have not substantially implemented
internationally agreed tax standards constitute the OECD Grey List (see List 1 in Table 1). Using
the Grey List as of August 17, 2009, 34 territories are described as tax haven. These territories
are predominantly located in Europe and the Caribbean though some are located in Africa, the
Middle East, and the Pacific. Larger independent countries such as Hong Kong and Ireland are
not classified as a tax haven though Singapore is.
Second, while never enacted, the ‘Stop Tax Haven Abuse Act’ (S.1533) is widely cited as a
source of Tax Haven territories. The Act lists 30 territories including Hong Kong and Singapore
(see List 2 in Table 1). Third, we use the original OECD Tax Haven list, which includes 42
territories (see List 3 in Table 1). Fourth, Hines and Rice (1994) provide a more practical list
based on true corporate tax rates rather than official corporate tax rates (see List 4 in Table 1).
Luxembourg, for instance, has an official corporate tax rate of 29% and does therefore not fall
under any of the definitions used to establish the first three Lists. Yet companies can enter
private agreements on low taxes (1% and less) and Advanced Tax Agreements with the
Luxembourgish tax authorities, making it effectively a tax haven. Fifth, as our transparency
experiment focuses on Tax Information Exchange Agreements (TIEAs), we provide a list of all
7
low-tax regimes that enter such agreements as per the OECD TIEA list (see TIEA in Table 1; see
OECD Harmful Tax Practices).
It is not our intention to provide one universal list of tax havens. Rather, most of our
descriptive analysis focuses on the OECD Grey List (the most commonly cited list) though all
results are robust to using other lists. We list countries involved in TIEAs because our
experiment on transparency is based on the passage of TIEAs.
2.2 Subsidiary data
Firm-level subsidiary data is obtained from Dun and Bradstreet’s 2013/2014 Who Owns
Whom book series and supplemented by the 2008/2009 and 1998/1999 series where necessary.
This data set lists public and private ultimate owner firms, the subsidiaries they hold to 50% or
more, and subsidiaries of subsidiaries. Besides subsidiary names, the data source identifies
subsidiaries’ headquarter country, including sovereign and non-sovereign tax haven jurisdictions.
Notice that this dataset does not allow us to distinguish between operative subsidiaries located
in tax havens and subsidiaries located in tax havens and used as tax vehicles. However, we argue
that most tax haven subsidiaries are likely tax vehicles given that tax havens are small and
feature a small population on average. Thus, Table 1 shows area and population for each tax
haven (and area and population for the USA, Germany and Denmark for comparison). It turns
out that tax haven territories are small and have a low population on average.
To substantiate our argument that many tax haven subsidiaries are likely tax vehicles rather
than operative subsidiaries, we count the number of subsidiaries located in tax havens and set
that number in relation to size and area of the tax haven territories. We repeat the same analysis
on the subset of subsidiaries owned by foreign firms (see Table 1). In the US, on average, one
can find 1 foreign subsidiary per 307 km2 and per 9,946 inhabitants. In the British Virgin Islands
and the Cayman Islands, one foreign subsidiary comes with 19 and 50 inhabitants, respectively,
and less than 0.1km2 in land.4
4
Some narratives from the data tell a similar story. For instance, there are 1,759 breweries in the US, roughly 1 per 180,000
inhabitants. There are 3 breweries in the Cayman Islands, roughly 1 per 20,000 inhabitants. SAB Miller (the multinational
brewery) alone had 6 subsidiaries in the Cayman Islands in 2008/2009, some of them certainly not breweries.
8
2.3 Firm-level data and variable definitions
Starting with a list of all publicly listed firms from Datastream/Worldscope, we match firm
names to the Dun & Bradstreet Who Owns Whom 2013/2014 book series. While we are able to
match slightly more than 14,000 firms, we focus on the ones without missing industry affiliation.
We are able to match 13,639 firms and collect name and headquarter country information on
163,908 subsidiaries. We enrich this dataset by firm-level accounting data and data on
trademarks and patents from Osiris. Stock return data are obtained from Datastream/Worldscope.
Following Demsetz and Lehn (1985) and Morck, Shleifer and Vishny (1988), we use Tobin’s
Q as the main measure of firm value. Tobin’s Q is obtained from Osiris as (Enterprise
Value+Total Liabilities)/(Total Shareholder Equity (Book Value) + Total Liabilities). In order to
facilitate interpretation in the multivariate difference in difference regression framework,
regressions use the natural logarithm of Tobin’s Q.
ROA(%) is Profit & Loss before Tax / Total Assets in %. Cash Flows over Sales is Operating
Cash Flow over Total Sales in %. Profit Margin and Gross Margin are Profit&Loss before Tax /
Operating Revenue in % and Gross Profit / Operating Revenue in %, respectively. Effective Tax
Rate is Income Taxes / Earnings before Interest, Tax, Depreciation and Amortization in %.
Leverage is Total Liabilities and Debt / Total Assets. Dividend payout is dividends per share
over earnings per share. Dividend Payer Dummy is a dummy equal to one if a firm pays non-zero
dividends. Intangible Assets and R&D are intangible assets and R&D as a fraction of total assets.
# Trademarks and # Patents denote the number of registered trademarks and patents in 2013. ID
Trademark and ID Patent are dummy variables equal to one if a firm has a trademark and patent,
respectively. ln(Assets) is the natural logarithm of total assets. Cash over Assets is Total Cash
and Short Term Investments / Total Assets in %. Sales Growth is the average year-by-year
growth in sales using sales from 2004 and 2013. Mean Foreign Tax is the average maximum
corporate tax rate faced by foreign subsidiaries where each foreign subsidiary is weighted
equally. Dif(Foreign-Home Tax) is the Mean Foreign Tax less the maximum tax bracket at
home. Accounting data and data on trademarks and patents is obtained from Osiris and Orbis.
Tax data is obtained from various sources (largely government agencies and audit firms).
Part
of
the
analysis
uses
daily
and
monthly
stock
returns
obtained
from
Datastream/Worldscope. Part of the analysis uses beta, firms’ exposure to their home country
9
market measured in a one-factor regression of monthly stock excess returns over the risk free
rate on the market excess return.
3. Descriptive statistics: Country and Industry Level
We collect firm-level subsidiary information for 13,639 firms from 52 countries from Dun &
Bradstreet’s Who Owns Whom 2013/2014. These firms have a total of 163,908 subsidiaries –
partly in their home country, partly abroad. Importantly for this study, some of the foreign
subsidiaries are located in tax haven territories. We will refer to firms that have at least one
subsidiary in a tax haven as a tax haven firm. The purpose of this section is then to describe the
occurrence of tax haven firms by country-level characteristics and by industry-level
characteristics.
3.1 Country Level
Our sample consists of 52 countries for which at least one publicly listed firm with nonmissing industry affiliation in Datastream/Worldscope could be matched to Dun & Bradstreet’s
Who Owns Whom 2013/2014. Summary statistics are provided in Table 2. Countries are sorted
by percentage of publicly listed firms that have at least one subsidiary headquartered in a tax
haven as defined by the List 1 (the OECD’s ‘Grey List’ as of August 17, 2009).
Besides Singapore – where 100% of sample firms are classified as Tax Haven Firm (because
Singapore is a Tax Haven by the OECD ‘Grey List’) – tax havens are particularly wide-spread in
Switzerland, Norway, Malaysia, and the Netherlands: More than one in five firms headquartered
in these countries have at least one tax haven subsidiaries by List 1. Some countries do not have
any tax haven subsidiaries by List 1, most notably Argentina, Greece, and Russia. Notice,
however, that one in six Greek firms make use of tax haven subsidiaries when using Lists 2 and
3. 9.7% of US firms use tax haven subsidiaries; fewer Chinese firms (1.2%) use tax haven
subsidiaries though this figure increases to 11.6% when using List 4 which includes Hong Kong
and Macau. The average country has between 6.9% and 19.6% tax haven firms (by the TIEA
10
List and List 4, respectively) and the % of firms with tax haven subsidiary are positively
correlated across lists.
Table 2 also provides country-level variables for further analysis. Log (GDP per capita) is the
natural logarithm of GDP per capita in USD in 2013 (Source: World Bank). Corporate Tax Rate
is the maximum corporate tax rate bracket and Income Tax Rate is the maximum income tax
bracket in 2013, listed by Wikipedia and obtained through various sources (largely government
agencies and audit firms). Tax Evasion is obtained from the Global Competitiveness Report
conducted by the World Economic Forum: Countries’ tax evasion is rated on a scale from 1
(strongly disagree) to 7 (strongly agree) to the statement ‘Tax evasion is minimal.’. ICRG
(Property Rights Protection) captures political, economic and financial risk in 2013 and is
obtained from the International Country Risk Guide; the measure ranges from 1 to 6 and
increases in protection. Corruption Level is based on Transparency International’s Corruption
Perception Index as of 2013 (Source: Transparency International), an index that measures
corruption levels on a scale from 1 (high corruption) to 10 (low corruption).
In order to study the correlation between the use of tax haven subsidiaries and country level
characteristics, Figure 2 plots the % of firms with tax haven subsidiary by country against
country characteristics. The y-axis denotes the percentage of publicly listed firms that have at
least one tax haven subsidiary using the OECD ‘Grey List’ as of August 17, 2009 though results
are robust to using other lists. Hong Kong, Singapore and Ireland are omitted because they
constitute tax havens by that list or other official tax haven lists. The x-axis denotes country level
characteristics.
--- FIGURE 1 ABOUT HERE ---
First, while the relation between corporate tax rates of firms’ home country and the use of tax
haven subsidiaries is close to zero, the use of tax haven subsidiaries is more wide-spread in
countries with higher income tax rates. Second, the use of tax haven subsidiaries is more
prevalent in countries with stronger property rights protection and lower corruption level,
suggesting that stealing from outsiders by insiders is more difficult in these countries, making tax
haven subsidiaries more valuable. Third, tax haven firms are more wide-spread in countries with
low levels of tax evasion, suggesting that tax havens are more valuable to firms that are
11
headquartered in countries where it is harder to steal from the government. Sixth, the use of tax
havens is more prevalent among firms in more developed countries.
While these results are correlations, they may all be explained by economic development.
Table 3 addresses this concern by explaining the use of tax havens at the country level by
country level characteristics and the natural logarithm of GDP per capita. The table presents the
results of country-level logit models where the dependent variable %TH Firms denotes the % of
publicly listed firms that have at least one subsidiary headquartered in a tax haven where tax
havens are countries or non-sovereign nations that appear on the OECD ‘Grey List’ (as of
August 17, 2009). Sample countries are those used in Figure 1. Panel A reports results for
equally weighted observations; Panel B reports results for value weighted observations where
weights are determined by the % of public firms in the overall sample. Supporting the previous
notion, the use of tax havens is more pronounced in countries with higher income tax rates, lower
protection of property, lower corruption levels, and lower tax evasion even after controlling for
economic development.
Overall, these correlations in the data are suggestive (but no more than suggestive) of two
channels, one related to tax savings, one related to expropriation.
3.2 Industry Level
Table 4 provides industry-level summary statistics for the use of tax havens across 48 FamaFrench Industries. Sample firms are those used in Table 2. # Parent Firms denotes the number of
publicly listed firms headquartered in the respective industry. List 1 through to List 4 and TIEA
denote the % of parent firms that have at least one subsidiary in a tax haven where tax havens are
defined by respective lists described above.
12
Most notably, the use of tax haven subsidiaries is most prevalent among firms in shipping,
machinery, chips, tobacco, and transportation. Notice that some of the tax havens (by any of the
lists) are flags of convenience which may partly explain the large fraction of shipping and
transportation firms. None of the results of this paper are subject to the inclusion of shipping and
transportation firms. Tax havens are less common among firms in health, gold mining, other
mining, textiles, and guns.
4. Descriptive statistics: Firm Level
In this section, we study the relation between the use of tax havens and firm characteristics.
As before, our aim in this section is to provide a description of the data rather than a causal link.
We will refer to firms that have at least one subsidiary in a tax haven as a tax haven firm.
4.1 Univariate split
Table 5 presents summary statistics for the sample of publicly listed firms. For a range of
variables, the table reports the number of firms with non-missing observations in 2013, the mean
of respective variables, the mean if such firm has at least one tax haven subsidiary or no tax
haven subsidiary (using the OECD ‘Grey List’, see Table 1), and the difference in means. As we
just saw, the use of tax havens is more common in certain countries and industries which is why
the table also reports differences in mean variables after adjusting variables by industry x
country. Means are constructed from one observation per firm, such firm-level observation being
the firm-level 10-year average using 2004-2013 data (or fewer years for missing data).
12.6% of sample firms are tax haven firms by the definition of the OECD ‘Grey List’ (List
5
1) . The average sample firm has a Tobin’s Q of 1.61 though tax haven firms have a significantly
lower Tobin’s Q. However, tax haven firms are more profitable in terms of operating profit (%
5
By Lists 2, 3 and 4, 18.9%, 19.0%, and 21.1% of the firms are tax haven firms. By the TIEA List, 5.3% of firms are tax haven
firms. Notice that the TIEA List contains by and large smaller tax havens and does not include Singapore, Hong Kong and
Ireland.
13
ROA, Cash Flow / Sales, and Profit Margin) despite not having a higher gross margin. While this
may seem at odds with the result for Tobin’s Q, tax haven firms also grow 2.7 percentage points
more slowly than non-tax haven firms. Surprising at first sight, tax haven firms have a higher
effective tax rate – this may seem at odds with the tax saving motive at first sight yet notice that
these are equilibrium outcomes: The marginal benefit of having a tax haven subsidiary may be
higher when effective tax rates are high. Similarly, while leverage provides a tax shield, the
marginal benefit of having a tax haven subsidiary may be higher if leverage is high, providing
potential justification for the fact that our sample firms with tax haven subsidiary have higher
leverage than sample firms without tax haven subsidiary.
In terms of intangible (and potentially easily transferrable) assets, the univariate split supports
intuition in that tax haven firms have more intangible assets, patents, and trademarks; also, tax
haven firms are more likely to have at least one patent or one trademark. Tax haven firms are
larger on average and – counter to popular belief – have fewer cash holdings as measured by
cash over assets. Last but not least, firms with tax haven subsidiary are firms that face a lower
tax rate abroad and a larger difference between foreign taxes and home taxes. This result is
suggestive of repatriation taxes making tax havens more valuable.
4.2 Firm-level probit regressions
In order to more formally study the use of tax haven subsidiaries, employ firm-level probit
regressions with industry and country fixed effects (Table 6). The dependent variable is an
indicator variable equal to one if a firm has at least one subsidiary headquartered in a tax haven
where tax havens are countries or non-sovereign nations on the OECD Grey List (though all
results are robust to other definitions of tax haven).
On its own, size, return on assets, effective tax rate, leverage, and being a dividend payer are
positively correlated with having a tax haven subsidiary while having a lot of cash relative to
assets is negatively correlated. Controlling for all these at once, the results for size and being a
dividend payer survive while having more cash over assets becomes positively associated with
14
having a tax haven subsidiary. This latter result confirms popular belief. Adding the difference
between taxes paid abroad and taxes paid at home as an additional control provides further
evidence on the repatriation argument discussed above.
Panel B investigates whether transferability of assets, measured by intangible assets, R&D,
and the use of patents and trademarks, explains the use of tax havens. Indeed, after controlling
for all of the factors discovered in Panel A, firms with assets that are more easily transferred are
more likely to employ tax haven subsidiaries. This provides some suggestive evidence on the tax
saving motive.
Panel C investigates whether monitoring by large shareholders, measured by institutional
ownership, foreign institutional ownership, and domestic ownership, is associated with the use of
tax haven subsidiaries. While large shareholders may benefit from the tax saving motive, they
may also suffer from expropriation or managerial slack. Overall, firms with higher institutional
ownership are less likely to use tax haven subsidiaries, suggestive of the latter channel. This
result is stronger for domestic institutional owners, suggestive that these owners are more
important monitors.
Panels D and E investigate subsamples of US firms and non-US firms and while results are
the same by and large, the cash motive, trademarks, and monitoring are less reliably associated
with the use of tax havens among US firms. However, some of this may be driven by the much
smaller sample size.
Overall, this section provides correlations between firm characteristics and the use of tax
haven subsidiaries. While not causal, evidence of both tax saving motives and expropriation
motives was provided.
5. Tax Savings Motive: Changes in corporate tax rates and firm value
We have so far provided some correlations that are suggestive of the tax savings motive. In
this section, we seek to provide causal evidence for the tax savings motive. For identification, we
exploit changes in corporate tax rates over the period 2008-2013. During this period, some
countries reduced their maximum corporate tax bracket. A reduction in corporate taxes should
increase free cash flows and, all else equal, increase firm value (notice part of the full effect may
15
be offset by the effect of taxes on managerial slack). However, firms that avoid home taxes
through tax haven subsidiaries should be less positively affected.
Figure 2 plots changes in the corporate tax rate between 2008 and 2013 against changes in
firm value and changes in the use of tax haven subsidiaries, respectively. Changes in the
corporate tax rate are obtained from KPMG’ Corporate and indirect Tax Rate Survey 2014; a
negative value denotes a reduction in corporate tax rates over the five year period. On the left,
the y-axis denotes changes in the difference in Tobin’s Q from 2008 to 2013. Specifically, the
difference between Tobin’s Q of firms with tax haven subsidiaries in 2008 and firms without tax
haven subsidiaries in 2008 is deducted from the respective difference in 2013. A negative value
denotes that firms with tax haven subsidiary have become relatively less valuable over the five
year period.
--- Figure 2 ABOUT HERE ---
In line with our prediction, the difference in firm value between tax haven and non-tax haven
firms decreases in countries that reduce corporate tax rates. In order to test this more formally at
the firm level, Table 7 investigates the effect of changes in the corporate tax rate on firm value in
a panel of publicly listed firms from 2008 to 2013. The left hand side is Tobin’s Q. The key
control variable is Change in Tax Rate, the percentage change in corporate tax rates over the
previous year. Tax Haven Subsidiary is an indicator variable equal to one if a firm has at least
one subsidiary in a tax haven (as defined by the OECD ‘Grey List’). Columns (1) and (2) use the
full sample while Columns (3) and (4) use a sample of firms with tax haven subsidiary and
control firms matched by industry, headquarter country, the natural logarithm of assets, and the
natural logarithm of firms’ age (measured by years since foundation). All regressions control for
the natural logarithm of assets, the natural logarithm of assets squared, firm fixed effects, and
time fixed effects. Standard errors are clustered at the country and year level (2-way clustering).
Indeed, while a reduction in corporate tax rates leads to an increase in firm value, this result
only holds for the subset of firms that do not have tax haven subsidiaries. In the matched sample,
16
a 1 percentage point decrease in the corporate tax rate is associated with a 1.2% increase in the
value of firms without tax haven subsidiary but no increase in the value of tax haven firms.
(Notice the sample of tax changes is a sample abundant of tax reduction – the Change in Tax
Rate coefficient, however, is coded such that it increases in corporate tax increases.)
When faced with an increase in corporate tax rates, the marginal benefit of having a tax haven
subsidiary may increase. In the right Panel of Figure 2, the y-axis denotes the difference between
the percentage of firms with tax haven subsidiaries in 2013 and the percentage of firms with tax
haven subsidiaries in 2008. A positive value means that the fraction of firms with tax haven
subsidiary has increased over the five year period. Somewhat in line with the idea that tax haven
subsidiaries become more valuable when corporate tax rates are relatively higher, the percentage
of firms with tax haven subsidiary increases less over the 6 year sample period in countries that
reduce corporate tax rates.
In sum, this section provides evidence on the tax saving motive. Reduction in the corporate
tax rate do not benefit firms with tax haven subsidiary and changes in the corporate tax rate are
associated with firms’ tendency to open tax haven subsidiaries.
6. Entrenchment Motive: Increases in tax haven transparency and firm value
We have provided some correlations that are suggestive of the expropriation motive above. In
this section, we seek to provide causal evidence. For identification, we exploit the passage of Tax
Information Exchange Agreements (TIEAs) which constitute a shock to the transparency of tax
haven operations.
6.1 Tax Information Exchange Agreements
Tax Information Exchange Agreements are bilateral agreements between territories aimed at
promoting the exchange of tax-relevant information in civil and criminal tax investigations.
Important for the study of firms, tax-relevant information comprises bank details and ownership
details of companies, funds, and trusts. Tax Information Exchange Agreements increase
transparency: they increase the amount of information that can be used to set home country tax
17
rates. Ultimately, Tax Information Exchange Agreements help tax authorities in enforcing tax
laws – with civil and criminal consequences – more effectively.6
Since 2000, over 500 Tax Information Exchange Agreements have been signed (see left Panel
Figure 3). While these agreements have been signed largely after 2008, importantly for our
identification strategy, the number of treated firms in our sample increased substantially through
a few agreements signed in 2001 and 2002, as well as through agreements signed after 2008.
On a critical note, at least 1/3 of these agreements have been signed between two tax havens
or between tax havens and economically small (and non-sovereign) territories (such as between
the Faroe Islands and Greenland). Being interested in implications of tax enforcement for
publicly listed firms, these agreements are outside the scope of this paper. On another critical
note, some countries are not among the signatory countries, e.g. Brazil and Russia or
Luxembourg and Switzerland. These countries would provide some interesting cross-country
predictions: In Russia, for instance, legal tax avoidance or even illegal tax fraud do not require
complex tax haven constructs but can be organized e.g. in conjunction with theft (see Mironov
2013) so that a TIEA signed by Russia may have no effects on firms.
Table 8 lists Tax Information Exchange Agreements (TIEAs) involving exactly one tax haven
country (or non-sovereign nation) and one non-tax haven country (Source: OECD Harmful Tax
Practices) and affecting at least one sample firm. Listed are non-tax haven signatories (Panel A)
and Tax Haven signatories. This table lists all 362 such agreements – some sample firms may be
affected by more than one of these agreements.
6
Notably, information provided under these agreements is confidential; the requested party is obliged to collect information even
if
the
requested
party
does
not
need
that
information.
See
e.g.
oecd.org/tax/exchange-of-taxinformation/taxinformationexchangeagreementstieas.htm and hmrc.gov.uk/taxtreaties/tiea.htm for more information and a list of
all treaties.
18
6.2 Empirical Methodology
We estimate the effect of tax enforcement on firm value using a difference-in-difference
approach that follows Bertrand and Mullainathan (2003). Specifically, we estimate
y it = α i + α t + β 1TREATED it + X it + ε it
(1)
where i denotes firms, t denotes time, yit is the dependent variable of interest (e.g. Tobin’s
Q), αi and α t are firm and year fixed effects, TREATEDit is a dummy that equals one if a firm
has been affected by a Tax Information Exchange Agreement signed between its headquarter
country and a tax haven in which that firm has a tax subsidiary, Xit is a vector of controls, and
ε it is an error term. Besides year and firm fixed effects, controls comprise size and size squared.
6.3 Firm Value
In Table 9, we study studies the effect of Tax Information Exchange Agreements (TIEAs) on
firm value using OLS regressions for a panel of firms from 1995 to 2013. We follow Equation
(1) above. The left-hand side variable is the natural logarithm of Tobin’s. The key control
Treated after is an indicator variable equal to one in the years after a firm has been directly
affected by a TIEA. A firm is directly affected (treated) if it is headquartered in a country that
signs a TIEA and has a subsidiary in the other signatory country or non-sovereign nation (a tax
haven). Some firms are affected by more than one TIEA: They are counted as treated the
moment they are affected for the first time. Column (1) uses the full sample of firms. In columns
(2) and (3), one non-treated (control) firm is matched to each treated firms five years prior to the
year a TIEA is signed. In columns (4) and (5), up to 10 firms are matched to treated firms with
replacement. Control after is an indicator variable equal to one in the years after a firm is control
firm to a firm affected by a TIEA. Firms are matched by country and industry and then
additionally by the natural logarithm of assets and the natural logarithm of their age, measured as
the number of years since the founding year. All regressions control for the natural logarithm of
assets, the natural logarithm of assets squared, firm fixed effects, and time fixed effects.
Treated=Control provides the p-value from testing that the coefficient on Treated after equals
19
that on Control after. Standard errors are clustered at the country and year level (2-way
clustering).
In line with the expropriation motive, the passage of TIEAs does indeed lead to an increase in
firm value. In the full sample, the average treated firm sees an increase in Tobin’s Q by 3.1%.
While this may seem a lot at first sight, recall that Desai, Dyck and Zingales (2007) find an effect
of roughly 20% for 5 tax enforcement announcements in Russia. The effect is slightly smaller yet
still significant for samples of 1 control firm (2.5%) and 10 control firms (2.3%). Moreover, the
effect is not present in control firms, i.e. in firms that are otherwise comparable to treated firms.
In order to alleviate the concern that we are merely capturing a time trend, Figure 9 plots the
evolution of firm value of treated firms around the passage of Tax Information Exchange
Agreements (TIEAs). The x-axis denotes years around the passage of TIEAs. The y-axis shows
the interaction between year-to-event dummies and an indicator variable that equals 1 if a firm is
directly affected by a TIEA. Interaction terms are obtained from an OLS regression on a sample
of treated and control firms with the natural logarithm of Tobin’s Q on the left hand side and
controls for size and size squared as well as year and industry fixed effects on the right. Control
firms are matched to treated firms 5 years before treatment by headquarter country, industry, as
well as the natural logarithm of assets and the natural logarithm of assets squared. Indeed, the
increase in firm value occurs abruptly between year -1 and year +1 around the treatment date.
One concern up to now may be that Tobin’s Q also responds to changes in e.g. accounting
practices associated with the passage of TIEAs. Figure 5 plots returns of treated firms 100 days
around the signing of a TIEA. A firm is directly affected (treated) if it is headquartered in a
country that signs a TIEA and has a subsidiary in the other signatory country (a tax haven).
Returns are obtained from Datastream/Worldscope and cumulated; cumulative returns are
standardized to equal zero a day before the signature date. While it is impossible to nail down
precisely the first mentioning of these agreements by the press, it is likely that these agreements
20
are not discussed in public long before they are signed. Also, as these agreements are passed at
different points in time between 2001 and 2012, using raw returns is less subject to other events
(such as a positive market around the passage) driving the main result. Indeed, Figure 5 confirms
that TIEAs do have a positive effect on affected firms’ value: The result on Tobin’s Q is not
entirely driven by changes in the accounting treatment.
In order to further understand what drives the increase in firm value, Table 10 studies subcomponents of Tobin’s Q around the passage of TIEAs using OLS regressions for a panel of
firms from 1995 to 2013. The analysis follows exactly Table 9 but the left-hand side is ROA(%),
Profit Margin, Gross Margin, Effective Tax Rate, and Beta (estimated in a 1-factor model of
monthly excess stock returns on the headquarter country’s main market index’ excess return over
24 months). Odd-numbered columns report results for the whole sample; even-numbered
columns report results for a sample of treated and control firms. Control firms are matched as
before.
Treated firms experience an insignificant increase in %ROA and profit margin relative to
control firms and their gross margin increases relative to control firms. Also, their effective tax
rate is unaffected and their beta goes down though insignificantly so. Taken together, while
largely insignificant, increases in operating profit and decreases in the discount rate (through a
reduction in beta) can jointly explain an increase in firm value.
6.4 Haven Hopping
Some of the full effect of TIEAs on firm value can be offset by firms adjusting their firm
structure through so-called haven hopping: If one of their tax haven subsidiaries is affected by
more transparency through TIEAs, firms may move that subsidiary to a tax haven that has not
entered a TIEA. Table 11 constitutes a conversion matrix aimed at investigating whether firms
strategically move out of tax havens entering Tax Information Exchange Agreements (TIEAs)
21
between 2008 and 2013. The focus is on non-US sample firms for which subsidiary data was
available in both 2008 and 2013 (most TIEAs signed by the US date back to the early 2000s).
For 2008, each row gives the number of firms with no tax haven subsidiary, with tax haven
subsidiaries only in tax havens that subsequently sign a TIEA, with tax haven subsidiaries only
in tax havens that subsequently do not sign a TIEA, and with tax haven subsidiaries in both tax
havens that sign and that do not sign a TIEA. For 2013, each column gives the number of firms
with no tax haven subsidiary, with tax haven subsidiaries only in tax havens that subsequently
sign a TIEA, with tax haven subsidiaries only in tax havens that subsequently do not sign a
TIEA, and with tax haven subsidiaries in both tax havens that sign and that do not sign a TIEA.
Numbers and percentages denote the number of firms and the percentage of the group moving
from a category in 2008 to a category in 2013.
Table 11 by and large confirms strategic haven hopping. For instance, 1,588 firms did not
have a tax haven subsidiary in 2008. 93% (1,480) of these firms had no tax haven subsidiary in
2013 either; of the remaining 7%, most (5% or 77 firms) established at least one tax haven
subsidiary but explicitly in tax havens that did not sign TIEAs over the five years. Similarly,
firms affected by TIEAs adjusted their subsidiary structure more dynamically: Of the 55 firms
that had subsidiaries exclusively in tax havens that later entered tax information exchange
agreements, only 42% maintained this exposure, compared to 68% of firms with exposure
exclusively to unaffected havens in 2008 and 73% of firms with exposure to both affected and
unaffected havens maintaining this exposure.
In sum, this section provides evidence of the expropriation motive: The passage of TIEAs
increased firm value though some of the full effect may have been offset by strategic haven
hopping.
7. Conclusion
Corporations use tax havens to reduce corporate taxes and to shield cash from shareholders.
Consistent with the tax motive, we find that a 1 percentage point reduction in home-country
22
corporate tax rates is associated with a 1.2% increase in value of firms without tax haven
subsidiary while firms with tax haven subsidiary are unaffected.
In order to provide direct evidence for an entrenchment motive, we exploit the passage of
Tax Information Exchange Agreements (TIEAs) between countries and tax havens as a shock to
firms’ transparency. The implementation of a TIEA increases average shareholder value by 3.1
using annual data. This positive effect is confirmed when studying daily returns around the
passage of TIEAs. Even though tax agreements are endorsed by firms’ investors, many firms
engage in haven hopping: They strategically move their subsidiaries from tax havens that have
entered TIEAs to tax havens that did not. This suggests that the OECD-led crackdown on tax
havens may benefit some of the less compliant tax havens.
Besides global evidence on this entrenchment motive, the paper provides novel global insights
into the use of tax haven subsidiaries with respect to country, industry, and firm-level
characteristics.
23
References:
Bertrand, Marianne and Sendhil Mullainathan, 2003, “Enjoying the Quiet Life? Corporate
Governance and Managerial Preferences.”, Journal of Political Economy, 111(5), 1043-1075.
Bilicka, Katarzyna and Clemens Fuest, 2014, ‘With which countries do tax havens share
information?’. Int Tax Public Finance, 21: 175-197.
Braun, Julia and Alfons Weichenrieder, 2014, ‘Does exchange of tax information between tax
authorities influence foreign direct investment into tax havens?’, Working Paper.
Demsetz, Harold and Kenneth Lehn, 1985, “The Structure of Corporate Ownership: Causes and
Consequences, Journal of Political Economy, 93(6), 1155-1177.
Desai, Mihir A. and Dhammika Dharmapala, 2005, “Corporate Tax Avoidance and Firm Value”,
American Law & Economics Association Annual Meetings, Paper 27.
Desai, Mihir A., Alexander Dyck and Luigi Zingales, 2005, “Theft and Taxes”, NBER Working
Paper, Paper No. 10978.
Desai, Mihir A., Alexander Dyck and Luigi Zingales, 2007, “Theft and Taxes”, Journal of
Financial Economics, 84, 591-623.
Desai, Mihir A., C. Fritz Foley, and James R. Hines Jr., 2004, ‘Economic Effects of Regional
Tax Havens’, NBER Working Paper #10806.
Desai, Mihir A., C. Fritz Foley, and James R. Hines Jr., 2006A, ‘Do tax havens divert economic
activity?’, Economic Letters, 90: 216-224.
Desai, Mihir A., C. Fritz Foley, and James R. Hines Jr., 2006B, ‘The demand for tax haven
operations’, Journal of Public Economics, 90: 513-531.
Dharmapala, Dhammika, 2008, ‘What problems and opportunities are created by tax havens?’,
Oxford Review of Economic Policy, 24(4): 661-679.
Dharmapala, Dhammika und Hines Jr., James R. (2006) Which Countries Become Tax Havens?
Dyck, Alexander and Luigi Zingales, 2004, “Private Benefits of Control: An International
Comparison”, The Journal of Finance, Vol. LIX(2), 537-600.
Dyreng, Scott D. and Kevin S. Markle, 2013, ‘The effect of financial constraints on taxmotivated income shifting by U.S. Multinationals, Working Paper.
Graham, John R. and Alan L. Tucker, 2006, “Tax Shelters and Corporate Debt Policy”, Journal
of Financial Economics, 81, 563-594.
24
Gumpert, Anna, James R. Hines Jr., and Monika Schnitzer, 2011, The use of tax havens in
exemption regimes, NBER Working Paper #17644.
Hanlon, Michelle, Rebecca Lester, and Rodrigo S. Verdi, 2014, “The Effect of Repatriation Tax
Costs on U.S. Multinational Investment”, SSRN Working Paper.
Heider, Florian and Alexander Ljungqvist, 2012, “As Certain as Debt and Taxes: Estimating the
Tax Sensitivity of Leverage from Exogenous State Tax Changes”, Working Paper.
Hines, James R. Jr. and Rice, “Fiscal Paradise: Foreign Tax Havens and American Business”,
NBER Working Paper No. 3477.
Johannesen, Niels and Gabriel Zucman, 2014, ‘The end of bank secrecy? An evaluation of the
G20 tax crackdown’, American Economic Journal: Economic Policy, 6(1): 65-91.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1996,
“Trust in Large Organizations”, NBER Working Paper 5864.
Markle, Kevin and Leslie Robinson, 2012, ‘Tax haven use across international tax regimes’.
Working Paper.
Mironov, Maxim. 2013, “Taxes, Theft, and Firm Performance”, The Journal of Finance, Vol.
68(4), 1441-1472.
Morck, Randall, Andrei Shleifer and Robert W. Vishny, 1988, “Management Ownership and
Corporate Performance: An Empirical Analysis”, Journal of Financial Economics, 20(1), 293315.
Slemrod, Joel and John D. Wilson, 2006, ‘Tax Competition with parasitic tax havens’, NBER
Working Paper #12225.
25
Table 1: Tax haven list
This table lists countries and non-sovereign nations that are classified as tax havens by at least one of five sources: First, by the OECD
‘Grey List’ (List 1; as of August 17, 2009). Second, by the ‘Stop Tax Haven Abuse Act’ (List 2; S.1533; not enacted). Third, by the
original OECD Tax Haven List (List 3). Fourth, by Hines and Rice (1994) (List 4). Fifth, by entering a Tax Information Exchange
Agreement (TIEA; OECD Harmful Tax Practices). Population (in 000s) denotes the population in 2013 (World Factbook). Area (km2)
denotes the land area in square kilometers (World Factbook). Pop Dens (ppl/km2) is the population divided by area (km2). #Subs is
the number of subsidiaries of public and private firms headquartered in the respective territory in 2013 (Dun & Bradstreet’s Who
Owns Whom 2013/2014). Subsidiaries are defined as companies owned by at least 50%. Pop/sub and km2/sub are population and
square kilometers per subsidiary, respectively. #Foreign Subs is the number of subsidiaries of public and private firms headquartered
in the respective territory in 2013 but ultimately owned by a foreign parent firm. Pop/ForSub and km2/ForSub denote the population
and square kilometers per foreign subsidiary, respectively.
Country Name
Andorra
Anguilla
Antigua & Barbuda
Aruba
Bahamas
Bahrain
Barbados
Belize
Bermuda
British Virgin Isl.
Cayman Islands
Channel Islands
Cook Islands
Costa Rica
Cyprus
Dominica
Gibraltar
Grenada
Guatemala
Hong Kong
Ireland
Isle of Man
Jordan
Lebanon
Liberia
Liechtenstein
Luxembourg
Macao
Maldives
Malta
Marshall Isl.
Mauritius
Region
Europe
Caribbean
Caribbean
Caribbean
Caribbean
Middle East
Caribbean
Central Am.
Pacific
Caribbean
Caribbean
Europe
Pacific
Central Am.
Europe
Caribbean
Europe
Caribbean
Central Am.
East Asia
Europe
Europe
Middle East
Middle East
West Africa
Europe
Europe
East Asia
Indian Oc.
Europe
Pacific
Indian Oc.
List 1
1
1
1
1
1
1
0
1
0
1
1
1
1
1
0
1
1
1
0
0
0
0
1
1
1
1
0
0
1
0
1
0
List 2
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
0
1
1
0
0
0
1
0
0
List 3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
0
1
1
0
0
1
1
1
1
List 4
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
0
TIEA
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
1
0
0
1
0
0
1
1
0
1
0
0
1
1
Pop (000s)
85.08
13.45
89.07
102.38
319.03
1,317.83
283.22
334.30
64.81
27.80
57.57
163.86
14.16
4,805.30
838.90
71.68
30.00
105.48
15,806.68
6,130.92
4,586.90
84.50
6,318.00
4,424.89
4,190.44
36.66
530.95
607.50
338.44
419.46
52.56
1,291.46
26
Area
(km2)
455
91
440
180
9,992
760
430
22,810
50
153
240
190
240
51,060
9,240
285
7
340
108,889
1,042
68,890
570
88,780
10,230
96,320
160
2,590
28
300
320
180
2,030
Pop Dens
(ppl/km2)
187
148
202
569
32
1,734
659
15
1,296
182
240
862
59
94
91
251
4,412
310
145
5,884
67
148
71
433
44
229
205
21,696
1,128
1,311
292
636
# Subs
42
41
36
99
481
427
343
174
1,808
3,706
1,958
2
46
462
7,862
20
1,147
29
319
35,605
32,996
Pop/
Sub
2,026
328
2,474
1,034
663
3,086
826
1,921
36
8
29
81,929
308
10,401
107
3,584
26
3,637
49,551
172
139
km2/
Sub
10.8
2.2
12.2
1.8
20.8
1.8
1.3
131.1
0.0
0.0
0.1
95.0
5.2
110.5
1.2
14.3
0.0
11.7
341.3
0.0
2.1
#Foreign
Subs
6
20
16
41
219
173
182
32
844
1,486
1,152
2
20
295
1,698
10
354
18
243
12,387
8,988
Pop/
ForSub
14,180
673
5,567
2,497
1,457
7,617
1,556
10,447
77
19
50
81,929
708
16,289
494
7,168
85
5,860
65,048
495
510
km2/
ForSub
76
5
28
4
46
4
2
713
0
0
0
95
12
173
5
29
0
19
448
0
8
141
258
45
911
13,376
317
26
2,119
67
682
44,809
17,151
93,121
40
40
1,916
13,017
198
784
1,894
629.6
39.7
2,140.4
0.2
0.2
0.1
11.5
0.2
2.7
3.0
106
133
38
144
5,154
205
20
585
13
345
59,604
33,270
110,275
255
103
2,963
16,922
717
4,043
3,743
838
77
2,535
1
1
0
15
1
14
6
Monaco
Montserrat
Nauru
Niue
Panama
Samoa
San Marino
Seychelles
Singapore
St. Kitts & Nevis
St. Lucia
St. Vinc. & Gren.
Tonga
Turks & Caicos
US Virgin Isl.
Uruguay
Vanuatu
Europe
Caribbean
Pacific
Pacific
Central Am.
Pacific
Europe
Indian Oc.
East Asia
Caribbean
Caribbean
Caribbean
Pacific
Caribbean
Caribbean
South Am.
Pacific
1
1
1
1
1
1
0
0
1
1
1
1
1
1
0
0
1
0
0
1
0
1
1
0
0
1
1
1
1
0
1
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
0
0
1
1
1
1
1
1
1
1
0
1
1
0
1
1
1
0
0
1
1
1
1
0
1
1
1
0
1
0
1
1
37.58
5.16
9.43
1.23
3,802.28
188.89
31.25
88.30
5,399.20
304.76
180.87
109.37
104.94
31.46
105.28
3,324.46
247.26
2
102
21
260
74,340
2,830
60
460
700
999
610
389
720
616
343
176,215
12,190
18,790
51
449
5
51
67
521
192
7,713
305
297
281
146
51
307
18.87
20
For Comparison
USA
Germany
Denmark
North Am
Europe
Europe
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
318,968.00
80,716.00
5,639.72
9,857,306
357,168
42,916
32
226
131
27
248
5
152
1,033
0.0
20.4
183
5
205
1,033
0
20
1,297
278
85
83
36,099
53
60
47
4
31
2,932
679
368
1,064
150
5,750
3,015
2,327
26,235
1,015
57.3
10.2
0.7
5.5
0.0
18.8
10.2
8.3
180.0
19.9
611
231
7
17
12,195
14
35
9
4
11
6,223
818
4,464
5,194
443
21,769
5,168
12,153
26,235
2,860
122
12
9
27
0
71
17
43
180
56
617
42
5,388
5,887
285.6
290.2
422
20
7,878
12,363
418
610
228,757
219,486
100,027
1,394
368
56
43.1
1.6
0.4
32,071
30,086
2,430
9,946
2,683
2,321
307
12
18
Table 2: Country level summary statistics and the use of tax haven subsidiaries around the world
This table provides country-level summary statistics. The sample consists of 52 countries for which at least one publicly listed firm
with non-missing industry affiliation in Datastream/Worldscope could be matched to Dun & Bradstreet’s Who Owns Whom
2013/2014. Countries are sorted by the % of public firms that have at least one subsidiary headquartered in a tax haven. Tax havens
are countries or non-sovereign nations that appear on the OECD ‘Grey List’ as of August 17, 2009 (this percentage is 100% for
Singapore as Singapore is a Tax Haven by that list). # Parent Firms denotes the number of publicly listed firms headquartered in the
respective country. # Subsidiary Firms denotes the number of subsidiaries owned to 50% or more by the parent firms. List 1 through
to List 4 and TIEA denote the % of parent firms that have at least one subsidiary in a tax haven where tax havens are countries or nonsovereign states on respective lists (see Table 1); this percentage is 100% if the country is defined as a tax haven by the respective list.
Log (GDP per capita) is the natural logarithm, of GDP per capita in USD in 2013 (Source: World Bank). Corporate Tax Rate is the
maximum corporate tax rate bracket and Income Tax Rate is the maximum income tax bracket in 2013, listed by Wikipedia and
obtained through various sources (largely government agencies and audit firms). Tax Evasion is obtained from the Global
Competitiveness Report conducted by the World Economic Forum: Countries’ tax evasion is rated on a scale from 1 (strongly
disagree) to 7 (strongly agree) to the statement ‘Tax evasion is minimal.’. ICRG (Property Rights Protection) captures political,
economic and financial risk in 2013 and is obtained from the International Country Risk Guide; the measure ranges from 1 to 6 and
increases in protection. Corruption Level is based on Transparency International’s Corruption Perception Index as of 2013 (Source:
Transparency International), an index that measures corruption levels on a scale from 1 (high corruption) to 10 (low corruption).
Country
# Parent
Firms
# Subsidiary
Firms
331
116
88
549
70
16
1,380
37
26
78
75
305
14
7
73
884
16
56
931
306
3,849
3,650
1,620
3,485
2,763
85
17,540
1,683
95
2,425
2,008
8,426
528
9
1,249
3,797
27
1,069
23,308
1,962
List 1
Singapore
Switzerland
Norway
Malaysia
Netherlands
Chile
Japan
Austria
Saudi Arabia
Spain
Finland
France
Portugal
Bangladesh
Denmark
India
Pakistan
Belgium
UK
Hong Kong
100.00%
37.93%
23.86%
22.04%
20.00%
18.75%
17.39%
16.22%
15.38%
15.38%
14.67%
14.43%
14.29%
14.29%
13.70%
12.78%
12.50%
12.50%
12.14%
12.09%
% of Firms with Tax Haven Subsidiary
(100% if country is TH by respective list)
List 2
List 3
List 4
100.00%
42.24%
26.14%
25.14%
28.57%
18.75%
25.58%
21.62%
15.38%
16.67%
18.67%
19.02%
21.43%
28.57%
19.18%
15.38%
12.50%
17.86%
16.86%
100.00%
100.00%
42.24%
26.14%
25.14%
28.57%
18.75%
25.65%
21.62%
19.23%
17.95%
18.67%
20.00%
21.43%
28.57%
19.18%
16.40%
12.50%
17.86%
16.97%
100.00%
100.00%
46.55%
27.27%
25.14%
34.29%
18.75%
25.80%
24.32%
15.38%
23.08%
25.33%
25.57%
28.57%
28.57%
24.66%
15.84%
12.50%
37.50%
24.27%
100.00%
28
Log (GDP
per capita)
Corporate
Tax Rate
Income
Tax Rate
Tax
Evasion
10.85
11.28
11.51
9.25
8.43
9.65
10.75
10.75
10.13
10.25
10.73
10.59
9.91
6.62
10.94
7.32
7.14
10.68
8.27
10.51
19.0%
25.0%
28.0%
25.0%
25.0%
20.0%
38.0%
25.0%
20.0%
30.0%
20.0%
33.3%
25.0%
20.0%
13.2%
47.8%
26.0%
52.0%
40.0%
50.0%
50.0%
5.05
4.49
3.96
4.34
3.40
4.20
4.41
3.60
52.0%
51.0%
45.0%
54.0%
25.0%
51.7%
33.0%
35.0%
55.0%
45.0%
15.0%
1.91
3.53
3.86
2.18
ICRG
Corruption
Index
TIEA
3.63%
26.72%
4.55%
2.00%
21.43%
18.75%
3.26%
10.81%
15.38%
16.67%
6.67%
15.41%
28.57%
0.00%
5.48%
3.39%
6.25%
28.57%
9.02%
10.78%
25.0%
30.0%
35.0%
34.0%
24.0%
16.5%
3.70
2.16
2.27
4.67
4.56
4.50
4.50
5.00
2.40
5.00
3.70
3.40
5.00
2.00
3.90
6.00
3.60
3.80
1.60
5.50
2.30
1.60
3.80
4.30
4.00
9.20
9.00
7.90
5.10
8.90
6.90
7.30
8.10
3.50
6.50
9.00
6.90
6.10
2.10
9.30
3.40
2.50
7.30
7.70
8.10
Philippines
Venezuela
Germany
New Zealand
Indonesia
USA
Italy
Korea
Australia
Thailand
Egypt
Brazil
Vietnam
Israel
Sweden
Canada
South Africa
Ireland
Turkey
China
Poland
Argentina
Colombia
Czech Rep.
Greece
Hungary
Kazakhstan
Mexico
Nigeria
Peru
Russia
Ukraine
Sum / Mean
71
9
385
49
118
2,940
81
560
1,126
207
15
32
17
159
138
629
202
32
65
989
205
21
1
10
94
12
1
8
10
1
74
20
611
45
9,136
298
512
39,812
1,632
2,764
10,129
952
30
294
33
976
2,745
3,059
1,698
652
360
5,367
1,077
103
1
38
990
196
2
169
15
3
565
66
11.27%
11.11%
10.65%
10.20%
10.17%
9.69%
8.64%
8.39%
8.35%
7.73%
6.67%
6.25%
5.88%
5.03%
4.35%
4.29%
3.47%
3.13%
1.54%
1.21%
0.49%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
15.49%
11.11%
14.29%
12.24%
13.56%
13.27%
9.88%
15.18%
10.48%
9.66%
6.67%
6.25%
5.88%
7.55%
10.87%
6.84%
5.45%
6.25%
1.54%
11.53%
0.49%
0.00%
0.00%
0.00%
17.02%
8.33%
0.00%
0.00%
0.00%
0.00%
1.35%
0.00%
15.49%
11.11%
14.29%
12.24%
13.56%
13.50%
9.88%
15.18%
10.48%
9.66%
6.67%
6.25%
5.88%
7.55%
11.59%
6.84%
5.94%
6.25%
1.54%
11.53%
0.49%
0.00%
0.00%
0.00%
17.02%
8.33%
0.00%
0.00%
0.00%
0.00%
1.35%
0.00%
15.49%
11.11%
17.14%
12.24%
13.56%
15.92%
24.69%
15.89%
11.37%
9.66%
6.67%
12.50%
5.88%
8.81%
10.87%
8.59%
6.44%
100.00%
3.08%
11.63%
1.95%
0.00%
0.00%
0.00%
20.21%
8.33%
0.00%
0.00%
0.00%
0.00%
1.35%
0.00%
11.27%
11.11%
6.49%
0.00%
0.85%
7.45%
22.22%
1.25%
1.07%
0.00%
6.67%
12.50%
0.00%
3.14%
3.62%
4.61%
2.48%
6.25%
1.54%
0.30%
0.98%
0.00%
0.00%
0.00%
5.32%
0.00%
0.00%
12.50%
0.00%
0.00%
0.00%
0.00%
7.86
9.45
10.66
10.56
8.18
10.85
10.43
7.72
11.12
6.31
8.09
7.03
7.47
10.39
8.61
10.84
8.90
10.74
9.27
8.71
7.15
9.36
8.96
7.53
10.02
7.14
9.40
9.19
7.35
8.82
9.55
8.26
30.0%
34.0%
29.8%
28.0%
25.0%
39.0%
31.4%
22.0%
30.0%
20.0%
20.0%
34.0%
25.0%
26.5%
22.0%
31.0%
28.0%
25.0%
20.0%
25.0%
19.0%
35.0%
33.0%
19.0%
33.0%
19.0%
17.5%
30.0%
30.0%
30.0%
20.0%
25.0%
32.0%
34.0%
45.0%
33.0%
30.0%
55.9%
43.0%
41.8%
45.0%
35.0%
20.0%
27.5%
35.0%
52.0%
57.0%
50.0%
40.0%
41.0%
35.0%
45.0%
32.0%
35.0%
33.0%
15.0%
42.0%
16.0%
10.0%
30.0%
24.0%
30.0%
13.0%
15.0%
1.83
1.56
3.41
5.00
2.53
4.47
1.77
3.29
4.58
3.41
3.57
2.14
13,639
163,908
10.75%
15.59%
15.76%
19.55%
6.90%
9.26
26.55%
36.33%
3.14
29
3.69
3.39
3.77
2.40
3.55
2.07
2.51
2.19
2.41
2.11
2.54
2.36
1.97
2.46
2.66
1.43
2.00
1.40
4.80
5.50
1.80
4.40
2.40
2.60
4.70
1.50
1.70
2.30
2.10
3.20
5.10
4.80
2.30
3.40
2.50
2.00
2.30
2.50
2.80
2.50
2.20
3.00
1.50
2.00
1.30
2.50
1.90
1.60
2.30
1.90
7.90
9.30
2.60
7.30
4.80
5.60
8.70
3.50
2.80
3.50
2.70
6.00
9.30
8.70
4.90
7.70
4.60
3.60
4.60
2.90
3.80
5.20
4.70
5.10
2.20
3.60
2.70
3.60
2.10
2.50
3.13
5.49
Table 3: Country level regressions with log(GDP pc) control
This table presents the results of country-level logit models. The dependent variable %TH Firms denotes the
% of publicly listed firms that have at least one subsidiary headquartered in a tax haven where tax havens are
countries or non-sovereign nations that appear on the OECD ‘Grey List’ (as of August 17, 2009). Sample
countries are those listed in Table 2 with the exception of countries that are a Tax Haven by any of the
different tax haven definitions given in Table 1. Panel A reports results for equally weighted observations;
Panel B reports results for value weighted observations where weights are determined by the % of public
firms in the overall sample. All regressions control for log(GDP per capita). Log (GDP per capita) is the
natural logarithm, of GDP per capita in USD in 2013 (Source: World Bank). Corporate Tax Rate is the
maximum corporate tax rate bracket and Income Tax Rate is the maximum income tax bracket in 2013, listed
by Wikipedia and obtained through various sources (largely government agencies and audit firms). Tax
Evasion is obtained from the Global Competitiveness Report conducted by the World Economic Forum:
Countries’ tax evasion is rated on a scale from 1 (strongly disagree) to 7 (strongly agree) to the statement
‘Tax evasion is minimal.’. ICRG (Property Rights Protection) captures political, economic and financial risk
in 2013 and is obtained from the International Country Risk Guide; the measure ranges from 1 to 6 and
increases in protection. Corruption Level is based on Transparency International’s Corruption Perception
Index as of 2013 (Source: Transparency International), an index that measures corruption levels on a scale
from 1 (high corruption) to 10 (low corruption). T-statistics for tests of significance based on robust standard
errors are reported below coefficients. ***, ** and * indicate significance at the 1%, 5% and 10% level.
Panel A: Equally weighted
Log (GDP pc)
(1)
%TH Firms
(2)
%TH Firms
(3)
%TH Firms
(4)
%TH Firms
(5)
%TH Firms
(6)
%TH Firms
0.313
(1.26)
0.375
(1.41)
-0.610
(-0.09)
-0.118
(-0.34)
0.153
(0.43)
-0.231
(-0.66)
-0.253
(-0.70)
Corporate Tax Rate
Income Tax Rate
13.659***
(3.09)
Tax Evasion
1.447**
(2.12)
ICRG (Property Rights)
1.206**
(2.22)
Corruption Level
Constant
Observations
Pseudo R2
-1.575
(-0.71)
49
0.032
0.706
(0.42)
48
0.042
-3.070**
(-2.36)
48
0.317
30
-2.499
(-1.53)
42
0.200
-1.445
(-1.31)
49
0.168
0.657**
(2.48)
-1.288
(-1.37)
49
0.194
Panel B: Value weighted
Log (GDP pc)
(1)
%TH Firms
(2)
%TH Firms
(3)
%TH Firms
(4)
%TH Firms
(5)
%TH Firms
(6)
%TH Firms
0.211
(0.87)
-0.094
(-0.23)
11.109
(0.72)
-0.068
(-0.19)
-0.805*
(-1.73)
-1.963*
(-1.73)
-1.888*
(-1.92)
Corporate Tax Rate
Income Tax Rate
14.077**
(2.52)
Tax Evasion
2.994***
(4.01)
ICRG (Property Rights)
4.850**
(2.26)
Corruption Level
Constant
Observations
Pseudo R2
1.915
(0.91)
49
0.010
1.602
(0.72)
48
0.034
-0.707
(-0.19)
48
0.275
31
3.174
(0.78)
42
0.428
9.810
(1.44)
49
0.402
1.823***
(3.88)
12.592*
(1.73)
49
0.415
Table 4: The use of tax haven subsidiaries by Industry
This table provides industry-level summary statistics for 48 Fama-French Industries. Sample firms are those
used in Table 2. # Parent Firms denotes the number of publicly listed firms headquartered in the respective
industry. List 1 through to List 4 and TIEA denote the % of parent firms that have at least one subsidiary in a
tax haven where tax havens are defined by respective lists: First, by the OECD ‘Grey List’ (List 1; as of
August 17, 2009). Second, by the ‘Stop Tax Haven Abuse Act’ (List 2; S.1533; not enacted). Third, by the
original OECD Tax Haven List (List 3). Fourth, by Hines and Rice (1994) (List 4). Fifth, by entering a Tax
Information Exchange Agreement (TIEA; OECD Harmful Tax Practices).
#Parent
Firms
SHIPS
MACH
CHIPS
SMOKE
TRANS
ELCEQ
RUBBR
CHEM
LABEQ
COMPS
PAPER
INSUR
STEEL
CNSTR
BOOKS
PERSV
WHLSL
HSHLD
AERO
MEDEQ
TOYS
FOOD
BOXES
TELCM
MEALS
SODA
BANKS
BUSSV
OIL
AUTOS
RTAIL
FABPR
BLDMT
UTIL
OTHER
AGRIC
COAL
FIN
RLEST
DRUGS
BEER
CLTHS
FUN
HLTH
GOLD
MINES
TXTLS
GUNS
Sum / Mean
% of Firms with Tax Haven Subsidiary
(100% if country is TH by respective list)
List 1
List 2
List 3
List 4
TIEA
47
503
637
14
378
203
175
406
122
185
115
135
301
385
93
136
579
251
36
169
90
272
53
376
194
57
584
1,719
486
237
429
48
413
258
41
156
57
972
529
422
63
133
184
140
309
351
188
8
38.30%
23.06%
21.98%
21.43%
21.16%
19.21%
18.29%
17.73%
17.21%
16.76%
16.52%
16.30%
15.95%
15.06%
15.05%
14.71%
14.34%
13.94%
13.89%
13.61%
13.33%
13.24%
13.21%
12.50%
12.37%
12.28%
11.64%
11.58%
10.70%
10.55%
10.49%
10.42%
10.41%
10.08%
9.76%
9.62%
8.77%
8.13%
7.94%
6.87%
6.35%
6.02%
5.98%
5.71%
4.85%
4.56%
4.26%
0.00%
42.55%
29.62%
33.59%
21.43%
27.78%
25.12%
26.86%
23.15%
22.95%
28.11%
21.74%
23.70%
21.26%
17.66%
17.20%
17.65%
21.42%
25.50%
13.89%
17.75%
36.67%
19.12%
20.75%
18.62%
21.13%
17.54%
16.61%
16.70%
13.99%
17.30%
16.55%
18.75%
15.25%
17.44%
17.07%
11.54%
14.04%
14.40%
16.45%
12.32%
15.87%
30.83%
12.50%
9.29%
6.15%
5.98%
12.23%
0.00%
42.55%
29.62%
33.59%
21.43%
28.04%
25.62%
26.86%
23.65%
22.95%
28.11%
21.74%
23.70%
21.26%
17.66%
18.28%
18.38%
21.59%
25.50%
13.89%
17.75%
36.67%
19.49%
20.75%
18.88%
21.65%
17.54%
16.78%
16.81%
14.20%
17.72%
17.25%
20.83%
15.74%
17.83%
17.07%
11.54%
14.04%
14.51%
16.45%
12.32%
17.46%
31.58%
12.50%
9.29%
6.15%
5.98%
12.23%
0.00%
44.68%
31.81%
35.16%
21.43%
29.37%
25.12%
29.14%
23.89%
26.23%
29.19%
22.61%
28.89%
22.59%
21.30%
20.43%
19.12%
23.14%
25.50%
16.67%
21.89%
37.78%
21.69%
24.53%
22.34%
22.68%
21.05%
19.35%
20.24%
15.64%
18.14%
19.11%
18.75%
16.22%
19.38%
21.95%
12.82%
17.54%
16.56%
17.96%
14.22%
19.05%
34.59%
19.02%
12.14%
7.77%
7.12%
13.30%
0.00%
4.26%
4.77%
5.49%
14.29%
7.14%
4.43%
1.71%
6.40%
6.56%
4.86%
5.22%
19.26%
3.65%
4.94%
5.38%
3.68%
3.11%
6.77%
11.11%
6.51%
8.89%
5.51%
5.66%
6.38%
6.70%
3.51%
9.76%
3.90%
5.35%
5.91%
6.06%
4.17%
3.39%
7.36%
4.88%
5.77%
5.26%
6.58%
3.40%
4.27%
4.76%
8.27%
4.89%
1.43%
3.88%
2.56%
1.60%
0.00%
13,639
12.84%
19.25%
19.49%
21.44%
5.62%
32
Table 5: Firm Level Summary Statistics
This table presents summary statistics for the sample of publicly listed firms. For a range of variables, the table reports the number
of firms with non-missing observations in 2013, the mean of respective variables, the mean if such firm has at least one tax haven
subsidiary or no tax haven subsidiary (using the OECD ‘Grey List’, see Table 1), the difference in means, and the difference in
industry-country-adjusted means. Means are constructed from one observation per firm, such firm-level observation being the
firm-level 10-year average using 2004-2013 data (or fewer years for missing data). TH Subsidiary (Dummy) is a dummy variable
equal to one if a firm has a tax haven subsidiary. Tobin’s Q is obtained from Osiris as (Enterprise Value+Total Liabilities)/(Total
Shareholder Equity (Book Value) + Total Liabilities). ROA(%) is Profit & Loss before Tax / Total Assets in %. Cash Flows over
Sales is Operating Cash Flow over Total Sales in %. Profit Margin and Gross Margin are Profit&Loss before Tax / Operating
Revenue in % and Gross Profit / Operating Revenue in %, respectively. Effective Tax Rate is Income Taxes / Earnings before
Interest, Tax, Depreciation and Amortization in %. Leverage is Total Liabilities and Debt / Total Assets. Dividend payout is
dividends per share over earnings per share. Dividend Payer Dummy is a dummy equal to one if a firm pays non-zero dividends.
Intangible Assets and R&D are intangible assets and R&D as a fraction of total assets. # Trademarks and # Patents denote the
number of registered trademarks and patents in 2013. ID Trademark and ID Patent are dummy variables equal to one if a firm has
a trademark and patent, respectively. ln(Assets) is the natural logarithm of total assets. Cash over Assets is Total Cash and Short
Term Investments / Total Assets in %. Sales Growth is the average year-by-year growth in sales using sales from 2004 and 2013.
Mean Foreign Tax is the average maximum corporate tax rate faced by foreign subsidiaries where each foreign subsidiary is
weighted equally. Dif(Foreign-Home Tax) is the Mean Foreign Tax less the maximum tax bracket at home. Accounting data and
data on trademarks and patents is obtained from Osiris and Orbis. Tax data is obtained from various sources (largely government
agencies and audit firms). ***, ** and * indicate significance of differences in mean at the 1%, 5% and 10% level.
Split by Tax Haven Dummy
Does firm have TH Sub?
All Firms
#Firms
Variable
TH Subsidiary (Dummy)
Tobin's Q
ROA(%)
Cash Flows over Sales
Profit Margin
Gross Margin
Effective Tax Rate
Leverage
Dividend Payout
Dividend Payer Dummy
Intangible Assets
R&D
#Trademarks
ID Trademark
#Patents
ID Patent
ln(Assets)
Cash over Total Assets
Sales Growth
Mean Foreign Tax
Dif (Foreign-Home Tax)
13,639
8,787
7,777
10,090
11,014
10,959
6,873
8,531
7,435
11,599
8,539
8,668
13,460
13,639
13,460
13,639
12,053
11,730
10,871
5,585
5,579
Mean in
2013
yes
no
Difference
12.60%
1.61
4.0%
8.2%
2.9%
43.3%
20.4%
49.5%
19.7%
42.1%
10.2%
1.5%
4.8
33.5%
31.4
32.4%
11.3
17.6%
11.9%
26.3%
-4.2%
1.00
1.44
4.9%
10.1%
6.0%
42.5%
22.2%
51.8%
23.2%
61.1%
11.8%
1.6%
11.1
46.5%
92.0
42.5%
12.7
16.1%
9.5%
23.5%
-6.2%
0
1.65
3.8%
7.9%
2.4%
43.4%
20.0%
49.0%
19.2%
39.3%
9.9%
1.5%
4.0
31.7%
23.1
30.9%
11.1
17.9%
12.2%
27.4%
-3.4%
(0.21)
1.1%
2.2%
3.6%
-0.9%
2.2%
2.8%
4.0%
21.8%
1.9%
0.1%
7.1
14.9%
68.9
11.5%
1.6
-1.7%
-2.7%
-3.9%
-2.8%
33
Industry-Country
Adjusted
Difference
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
(0.09)
1.0%
2.9%
3.3%
0.0%
1.0%
2.9%
3.3%
13.7%
2.0%
-0.1%
6.2
12.8%
54.1
9.4%
1.3
-1.5%
-0.3%
-2.5%
-2.5%
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
***
Table 6: What explains the use of TH? (probit regressions)
This table presents the result of firm-level probit regressions. The dependent variable is an indicator variable equal to one if a firm has
at least one subsidiary headquartered in a tax haven where tax havens are countries or non-sovereign nations on the OECD Grey List.
Control variables in Panels A and B are defined in Table 5. In Panel C, Institutional Ownership, Foreign Institutional Ownership and
Domestic Ownership are the fraction of shares owned by institutional, foreign institutional and domestic institutional owners (from
Orbis/Osiris), respectively. All Panels include industry fixed effects. Panels A, B, C and E include country fixed effects. T-statistics
for tests of significance based on robust standard errors are reported below coefficients. ***, ** and * indicate significance at the 1%,
5% and 10% level.
Panel A: All firms
(1)
TH Sub
(Dummy)
Log (Assets)
(2)
TH Sub
(Dummy)
(3)
TH Sub
(Dummy)
(5)
TH Sub
(Dummy)
(6)
TH Sub
(Dummy)
(7)
TH Sub
(Dummy)
0.810***
(17.22)
0.262***
(13.83)
0.028
(0.09)
-0.074
(-0.37)
0.236*
(1.92)
0.670***
(3.08)
0.301***
(4.01)
0.244***
(20.07)
Return on Assets
0.854***
(4.70)
Effective tax rate
0.630***
(4.28)
Leverage
Cash / Total
Assets
Dividend Payer
(Dummy)
Foreign –
Home Tax
Country FE
Industry FE
Observations
Pseudo R2
(4)
TH Sub
(Dummy)
0.367***
(5.48)
-0.585***
(-4.63)
Yes
Yes
11875
0.194
Yes
Yes
7610
0.091
Yes
Yes
6692
0.097
Yes
Yes
8404
0.095
Yes
Yes
11553
0.092
Yes
Yes
11416
0.130
34
Yes
Yes
5265
0.210
(8)
TH Sub
(Dummy)
(9)
TH Sub
(Dummy)
-7.138***
(-18.91)
Yes
Yes
5308
0.158
0.221***
(9.84)
-0.214
(-0.58)
-0.105
(-0.43)
0.140
(0.88)
0.481*
(1.67)
0.341***
(3.56)
-8.799***
(-11.18)
Yes
Yes
2463
0.228
Panel B: Transferable Assets
Log (Assets)
Return on Assets
Effective tax rate
Leverage
Cash / Total Assets
Dividend Payer (Dummy)
Intangible Assets
(1)
TH Sub
(Dummy)
(2)
TH Sub
(Dummy)
(3)
TH Sub
(Dummy)
(4)
TH Sub
(Dummy)
(5)
TH Sub
(Dummy)
(6)
TH Sub
(Dummy)
(7)
TH Sub
(Dummy)
(8)
TH Sub
(Dummy)
0.255***
(13.35)
0.033
(0.11)
-0.080
(-0.39)
0.233*
(1.88)
0.767***
(3.46)
0.303***
(4.00)
0.670***
(4.07)
0.256***
(13.52)
0.082
(0.27)
-0.041
(-0.20)
0.265**
(2.13)
0.446*
(1.92)
0.318***
(4.22)
0.246***
(12.89)
0.018
(0.06)
-0.054
(-0.27)
0.282**
(2.27)
0.602***
(2.74)
0.295***
(3.90)
0.228***
(11.29)
-0.022
(-0.07)
-0.008
(-0.04)
0.300**
(2.37)
0.538**
(2.37)
0.255***
(3.31)
0.245***
(12.80)
0.039
(0.13)
-0.090
(-0.45)
0.262**
(2.10)
0.586***
(2.64)
0.302***
(4.04)
0.218***
(10.81)
0.009
(0.03)
0.077
(0.38)
0.315**
(2.50)
0.541**
(2.36)
0.266***
(3.46)
0.237***
(12.30)
0.031
(0.10)
-0.072
(-0.36)
0.291**
(2.32)
0.551**
(2.47)
0.297***
(3.94)
0.211***
(10.11)
0.008
(0.03)
0.068
(0.33)
0.340***
(2.67)
0.542**
(2.34)
0.251***
(3.21)
R&D/Assets
2.115***
(3.11)
Trademark (Dummy)
0.291***
(5.20)
Ln(#Trademarks)
0.216***
(3.68)
0.163***
(7.62)
Parent (Dummy)
0.085***
(3.46)
0.343***
(5.70)
Ln(#Patents)
Country FE
Industry FE
Observations
Pseudo R2
Yes
Yes
5167
0.213
Yes
Yes
5240
0.212
Yes
Yes
5265
0.215
Yes
Yes
5143
0.214
Yes
Yes
5265
0.217
35
0.278***
(4.41)
0.130***
(7.93)
Yes
Yes
5146
0.211
Yes
Yes
5265
0.219
0.099***
(5.27)
Yes
Yes
5058
0.208
Panel C: Institutional Ownership
Log (Assets)
Return on Assets
Effective tax rate
Leverage
Cash / Total Assets
Dividend Payer (Dummy)
Institutional Ownership
(1)
TH Sub
(Dummy)
(2)
TH Sub
(Dummy)
(3)
TH Sub
(Dummy)
0.316***
(12.05)
0.258
(0.65)
0.019
(0.07)
0.220
(1.32)
0.364
(1.34)
0.302***
(2.98)
-2.406*
(-1.95)
0.325***
(12.18)
0.292
(0.71)
-0.003
(-0.01)
0.241
(1.44)
0.404
(1.48)
0.275***
(2.74)
0.313***
(12.03)
0.261
(0.68)
-0.024
(-0.08)
0.190
(1.16)
0.496*
(1.74)
0.314***
(3.07)
Foreign Institutional Ownership
-3.573
(-1.12)
Domestic Institutional Ownership
Country FE
Industry FE
Observations
Pseudo R2
Yes
Yes
5355
0.319
Yes
Yes
5355
0.319
-4.372***
(-3.23)
Yes
Yes
5355
0.317
36
Panel D: Only US Firms
Log (Assets)
Return on Assets
Effective tax rate
Leverage
Cash / Total Assets
Dividend Payer
(Dummy)
Foreign-Home Tax
(1)
TH Sub
(Dummy)
(2)
TH Sub
(Dummy)
(3)
TH Sub
(Dummy)
(4)
TH Sub
(Dummy)
(5)
TH Sub
(Dummy)
(6)
TH Sub
(Dummy)
(7)
TH Sub
(Dummy)
(8)
TH Sub
(Dummy)
(9)
TH Sub
(Dummy)
0.244***
(6.57)
0.492
(0.84)
-0.612
(-1.44)
0.275
(1.11)
0.198
(0.52)
0.519***
(3.49)
0.174***
(4.51)
0.609
(0.91)
-1.050**
(-2.03)
0.006
(0.02)
-0.571
(-1.19)
0.499***
(2.89)
-9.529***
(-4.56)
0.236***
(6.26)
0.513
(0.87)
-0.690
(-1.58)
0.225
(0.90)
0.394
(1.01)
0.538***
(3.61)
0.242***
(6.47)
0.522
(0.84)
-0.602
(-1.41)
0.267
(1.07)
0.044
(0.11)
0.537***
(3.61)
0.236***
(6.38)
0.447
(0.75)
-0.619
(-1.46)
0.286
(1.11)
0.061
(0.16)
0.500***
(3.34)
0.240***
(5.41)
0.343
(0.51)
-0.150
(-0.33)
0.295
(1.03)
0.295
(0.68)
0.290*
(1.75)
0.238***
(5.88)
-0.313
(-0.55)
0.107
(0.23)
0.539**
(2.15)
0.145
(0.34)
0.471***
(3.20)
0.230***
(5.62)
-0.341
(-0.60)
0.115
(0.25)
0.575**
(2.28)
0.158
(0.37)
0.436***
(3.01)
0.237***
(5.85)
-0.309
(-0.54)
0.112
(0.25)
0.538**
(2.14)
0.128
(0.30)
0.476***
(3.23)
Intangible Assets
0.704**
(2.41)
R&D/Assets
0.748
(0.69)
Trademark (Dummy)
-0.077
(-0.58)
Ln(#Trademarks)
-0.008
(-0.20)
Parent (Dummy)
0.514***
(3.81)
Ln(#Patents)
Institutional
Ownership
Foreign Institutional
Ownership
Domestic Inst.
Ownership
Industry FE
Observations
Pseudo R2
0.171***
(4.27)
-2.408
(-1.22)
40.315
(1.23)
Yes
961
0.207
Yes
584
0.211
Yes
939
0.209
Yes
955
0.206
Yes
961
0.222
37
Yes
855
0.219
Yes
987
0.177
Yes
989
0.176
-2.171
(-1.07)
Yes
985
0.177
Panel E: Only non-US firms
Log (Assets)
Return on Assets
Effective tax rate
Leverage
Cash / Total Assets
Dividend Payer
(Dummy)
Foreign-Home Tax
(1)
TH Sub
(Dummy)
(2)
TH Sub
(Dummy)
(3)
TH Sub
(Dummy)
(4)
TH Sub
(Dummy)
(5)
TH Sub
(Dummy)
(6)
TH Sub
(Dummy)
(7)
TH Sub
(Dummy)
(8)
TH Sub
(Dummy)
(9)
TH Sub
(Dummy)
0.282***
(12.99)
-0.162
(-0.45)
0.003
(0.01)
0.233*
(1.65)
0.667**
(2.41)
0.227**
(2.54)
0.262***
(9.14)
-0.647
(-1.38)
0.237
(0.83)
0.152
(0.79)
0.994**
(2.50)
0.181
(1.50)
-8.810***
(-10.24)
0.276***
(12.59)
-0.153
(-0.42)
0.024
(0.10)
0.238*
(1.68)
0.727***
(2.59)
0.227**
(2.52)
0.276***
(12.69)
-0.130
(-0.36)
0.020
(0.09)
0.279**
(1.97)
0.497*
(1.74)
0.239***
(2.66)
0.252***
(11.39)
-0.189
(-0.53)
0.010
(0.04)
0.302**
(2.12)
0.574**
(2.03)
0.225**
(2.50)
0.215***
(9.28)
-0.196
(-0.56)
0.039
(0.16)
0.366**
(2.56)
0.505*
(1.78)
0.246***
(2.67)
0.395***
(13.24)
0.780
(1.40)
-0.089
(-0.23)
0.091
(0.40)
0.225
(0.55)
0.156
(1.08)
0.409***
(13.52)
0.857
(1.50)
-0.105
(-0.26)
0.106
(0.46)
0.260
(0.63)
0.148
(1.01)
0.386***
(12.98)
0.670
(1.24)
-0.193
(-0.49)
0.043
(0.19)
0.524
(1.19)
0.156
(1.10)
Intangible Assets
0.590***
(2.87)
R&D/Assets
3.007***
(3.06)
Trademark (Dummy)
0.328***
(4.90)
Ln(#Trademarks)
0.155***
(4.96)
Parent (Dummy)
0.178**
(2.40)
Ln(#Patents)
Institutional
Ownership
Foreign Institutional
Ownership
Domestic Inst.
Ownership
Country FE
Industry FE
Observations
Pseudo R2
0.062***
(2.90)
-1.330
(-0.85)
-4.369
(-1.12)
Yes
Yes
4224
0.225
Yes
Yes
4150
0.227
Yes
Yes
4205
0.229
Yes
Yes
4224
0.235
38
Yes
Yes
4116
0.222
Yes
Yes
4122
0.392
Yes
Yes
4158
0.395
-4.968**
(-2.42)
Yes
Yes
4119
0.389
Table 7 Changes in Corporate Taxes and Firm Value
This table investigates the effect of changes in the corporate tax rate on firm value in a panel of publicly
listed firms from 2008 to 2013. The left hand side is Tobin’s Q, obtained from Osiris as (Enterprise
Value+Total Liabilities)/(Total Shareholder Equity (Book Value) + Total Liabilities). The key control
variable is Change in Tax Rate, the percentage change in corporate tax rates over the previous year obtained
from KPMG’ Corporate and indirect Tax Rate Survey 2014. Tax Haven Subsidiary is an indicator variable
equal to one if a firm has at least one subsidiary in a tax haven (as defined by the OECD ‘Grey List’).
Columns (1) and (2) use the full sample while Columns (3) and (4) use a sample of firms with tax haven
subsidiary and control firms matched by industry, headquarter country, the natural logarithm of assets, and
the natural logarithm of firms’ age (measured by years since foundation). All regressions control for the
natural logarithm of assets, the natural logarithm of assets squared, firm fixed effects, and time fixed effects.
T-statistics for tests of significance of coefficients based on robust standard errors clustered at the country
and year level (2-way clustering) are reported below coefficients. ***, ** and * indicate significance at the
1%, 5% and 10% level.
DV:
Sample:
Change in Tax Rate
Change in Tax Rate
* Tax Haven Subsidiary
Ln(Assets)
Ln(Assets) sqr
Firm FE
Time FE
N
r2_a
(1)
ALL
Tobin’s Q
All
-0.858
(-1.52)
(2)
ALL
Tobin’s Q
All
-0.872
(-1.55)
0.696*
(1.79)
0.071**
(2.15)
-0.004***
(-3.12)
Y
Y
37414
0.813
0.071**
(2.15)
-0.004***
(-3.11)
Y
Y
37414
0.813
39
(3)
ALL
Tobin’s Q
Matched
-0.871**
(-2.31)
0.211***
(8.74)
-0.009***
(-5.23)
Y
Y
5587
0.851
(4)
ALL
Tobin’s Q
Matched
-1.217***
(-3.14)
1.027**
(2.68)
0.211***
(8.68)
-0.009***
(-5.22)
Y
Y
5587
0.851
Table 8 Tax Information Exchange Agreements passed by Non-Tax Haven Countries
This table lists Tax Information Exchange Agreements (TIEAs) involving exactly one tax haven country (or
non-sovereign nation) and one non-tax haven country (Source: OECD Harmful Tax Practices) and affecting
at least one sample firm. Listed are non-tax haven signatories (Panel A) and Tax Haven signatories. This
table lists all 362 such agreements – some sample firms may be affected by more than one of these
agreements.
Panel A: Non-Tax Havens
Panel B: Tax Havens
# TIEA
Partners
Coded
ISO
ISO
Country
ARG
Argentina
2
ABW
Aruba
8
AUS
Australia
29
AIA
Anguilla
11
AUT
Austria
4
AND
Andorra
12
BEL
Belgium
12
ANT
Netherlands Antilles
7
CAN
Canada
7
ATG
Antigua
11
CZE
Czech Rep
5
BHR
Bahrain
5
DEU
Germany
13
BHS
The Bahamas
14
DNK
Denmark
38
BLZ
Belize
11
ESP
Spain
4
BMU
Bermuda
16
FIN
Finland
36
BRB
Barbados
1
FRA
France
20
COK
Cook Islands
11
GBR
UK
18
CRI
Costa Rica
7
IND
India
2
CYM
Cayman Islands
18
IRL
Ireland
15
DMA
Dominica
11
ISL
Iceland
37
GGY
Guernsey
12
JPN
Japan
3
GIB
Gibraltar
16
MEX
Mexico
3
GRD
Grenada
9
NLD
Netherlands
12
GTM
Guatemala
4
NOR
Norway
34
IMN
Isle of Man
11
NZL
New Zealand
15
JEY
Jersey
12
PRT
Portugal
14
KNA
St. Kitts & Nevis
21
SVN
Slovenia
1
LBR
Liberia
8
SWE
Sweden
34
LCA
St. Lucia
13
USA
United States
4
All non-TH Partners
362
Country
# TIEA
Partners
Coded
LIE
Liechtenstein
12
MAC
Macao
6
MCO
Monaco
9
MHL
Marshall Islands
7
MSR
Monserrat
7
MUS
Mauritius
5
PAN
Panama
1
SMR
San Marino
12
SYC
The Seychelles
5
TCA
Turks & Caicos
12
URY
Uruguay
5
VCT
St. Vincent & Grenadines
5
VGB
British Virgin Islands
13
VUT
Vanuatu
6
WSM
Samoa
8
All TH Partners
40
362
Table 9: Tobin’s Q around the passage of Tax Information Exchange Agreements
This table studies the effect of Tax Information Exchange Agreements (TIEAs) on firm value using OLS
regressions for a panel of firms from 1995 to 2013. The left-hand side variable is the natural logarithm of
Tobin’s, calculated as (Enterprise Value+Total Liabilities)/(Total Shareholder Equity (Book Value) + Total
Liabilities). The key control Treated after is an indicator variable equal to one in the years after a firm has
been directly affected by a TIEA. A firm is directly affected (treated) if it is headquartered in a country that
signs a TIEA and has a subsidiary in the other signatory country or non-sovereign nation (a tax haven). Some
firms are affected by more than one TIEA: They are counted as treated the moment they are affected for the
first time. Column (1) uses the full sample of firms. In columns (2) and (3), one non-treated (control) firm is
matched to each treated firms five years prior to the year a TIEA is signed. In columns (4) and (5), up to 10
firms are matched to treated firms with replacement. Control after is an indicator variable equal to one in the
years after a firm is control firm to a firm affected by a TIEA. Firms are matched by country and industry
and then additionally by the natural logarithm of assets and the natural logarithm of their age, measured as
the number of years since the founding year. All regressions control for the natural logarithm of assets, the
natural logarithm of assets squared, firm fixed effects, and time fixed effects. Treated=Control provides the
p-value from testing that the coefficient on Treated after equals that on Control after. T-statistics for tests of
significance of coefficients based on robust standard errors clustered at the country and year level (2-way
clustering) are reported below coefficients. ***, ** and * indicate significance at the 1%, 5% and 10% level.
Treated after
(1)
ALL
Ln(Tobin’s Q)
(2)
1 Match
Ln(Tobin’s Q)
(3)
1 Match
Ln(Tobin’s Q)
(4)
10 Match
Ln(Tobin’s Q)
(5)
10 Matches
Ln(Tobin’s Q)
0.031***
(3.26)
0.025*
(1.82)
0.033*
(2.07)
0.023*
(1.94)
0.028**
(2.13)
Control after
-0.013
(-1.01)
-0.010
(-0.99)
Ln(Assets)
0.097***
(3.30)
-0.056
(-1.45)
-0.054
(-1.36)
0.115***
(3.77)
0.116***
(3.73)
Ln(Assets) Sqr
-0.004***
(-3.22)
-0.000
(-0.00)
-0.000
(-0.05)
-0.005***
(-5.18)
-0.005***
(-5.11)
Y
Y
75841
0.715
Y
Y
4052
0.752
Y
Y
4052
0.752
0.083
Y
Y
13280
0.723
Y
Y
13280
0.724
0.070
Firm FE
Time FE
N
Adjusted R2
Treated=Control
41
Table 10 - Other Components of Tobin’s Q around the passage of Tax Information Exchange Agreements
This table studies the effect of Tax Information Exchange Agreements (TIEAs) on various contributors to firm value using OLS
regressions for a panel of firms from 1995 to 2013. The analysis follows exactly Table 9 but the left-hand side is ROA(%) (Profit &
Loss before Tax / Total Assets in %), Profit Margin (Profit&Loss before Tax / Operating Revenue in %), Gross Margin (Gross Profit /
Operating Revenue in %), Effective Tax Rate (Income Taxes / Earnings before Interest, Tax, Depreciation and Amortization in %),
and Beta (estimated in a 1-factor model of monthly excess stock returns on the headquarter country’s main market index’ excess
return over 24 months). Odd-numbered columns report results for the whole sample; even-numbered columns report results for a
sample of treated and control firms .Control firms are matched by country and industry and then additionally by the natural logarithm
of assets and the natural logarithm of their age, measured as the number of years since the founding year. All regressions control for
the natural logarithm of assets, the natural logarithm of assets squared, firm fixed effects, and time fixed effects. Treated=Control
provides the p-value from testing that the coefficient on Treated after equals that on Control after. T-statistics for tests of significance
of coefficients based on robust standard errors clustered at the country and year level (2-way clustering) are reported below
coefficients. ***, ** and * indicate significance at the 1%, 5% and 10% level.
Treated after
(1)
ALL
(2)
Match 1
ROA(%)
-0.262
(-0.93)
Control after
(4)
Match 1
Profit
Margin (%)
(5)
ALL
Gross
Margin (%)
(6)
Match 1
Gross
Margin (%)
(7)
ALL
Effective
Tax
(8)
Match 1
Effective
Tax
(9)
ALL
(10)
Match 1
ROA(%)
(3)
ALL
Profit
Margin (%)
Beta
Beta
0.363
(0.74)
-0.752
(-1.31)
-0.172
(-0.16)
0.876*
(1.93)
1.294**
(2.21)
0.003
(0.53)
0.007
(0.82)
0.003
(0.05)
-0.042
(-1.20)
-0.482
(-0.89)
-0.990
(-1.02)
-0.568
(-1.18)
-0.000
(-0.00)
0.003
(0.08)
Assets
15.191***
(8.23)
9.963***
(3.47)
9.447***
(5.14)
9.492***
(3.17)
-3.199**
(-2.77)
-0.189
(-0.02)
0.010*
(2.02)
0.004
(0.45)
-0.121***
(-4.07)
-0.137
(-0.74)
Assets Sqr
-0.507***
(-6.84)
-0.334***
(-3.73)
-0.297***
(-4.20)
-0.285***
(-3.17)
0.135***
(2.98)
0.039
(0.14)
0.000
(0.06)
-0.000
(-0.05)
0.008***
(6.86)
0.008
(1.20)
Y
Y
71243
0.572
Y
Y
3981
0.547
0.383
Y
Y
64202
0.486
Y
Y
3859
0.494
0.687
Y
Y
64221
0.848
Y
Y
3793
0.885
0.077
Y
Y
71594
0.316
Y
Y
3934
0.377
0.654
Y
Y
38940
0.339
Y
Y
2193
0.379
0.506
Firm FE
Time FE
N
Adjusted R2
Treated=Control
42
Table 11: Haven Hopping
This conversion matrix investigates whether firms strategically move out of tax havens entering Tax
Information Exchange Agreements (TIEAs) between 2008 and 2013. The focus is on non-US sample firms
for which subsidiary data was available in both 2008 and 2013 (most TIEAs signed by the US date back to
the early 2000s). For 2008, each row gives the number of firms with no tax haven subsidiary, with tax haven
subsidiaries only in tax havens that subsequently sign a TIEA, with tax haven subsidiaries only in tax havens
that subsequently do not sign a TIEA, and with tax haven subsidiaries in both tax havens that sign and that
do not sign a TIEA. For 2013, each column gives the number of firms with no tax haven subsidiary, with tax
haven subsidiaries only in tax havens that subsequently sign a TIEA, with tax haven subsidiaries only in tax
havens that subsequently do not sign a TIEA, and with tax haven subsidiaries in both tax havens that sign
and that do not sign a TIEA. Numbers and percentages denote the number of firms and the percentage of the
group moving from a category in 2008 to a category in 2013. For instance, 1,588 firms did not have a tax
haven subsidiary in 2008. 1,480 of these firms had no tax haven subsidiary in 2013 and 77 of these firms
established at least one subsidiary in tax havens but only in tax havens that did not sign TIEAs over the five
years, accounting for 5% of the 1,588 firms.
43
Figure 1: The use of tax haven subsidiaries and country characteristics
This Figure illustrates the use of tax haven subsidiaries at the country level. The y-axis denotes the
percentage of publicly listed firms that have at least one tax haven subsidiary. Subsidiary data is collected
from Dun & Bradstreet’s Who Owns Whom 2013/2014 book series. Tax havens are sovereign countries or
non-sovereign nations that appear on the OECD grey list (as of August 17, 2009); Hong Kong, Singapore
and Ireland are omitted because they constitute tax havens by that list or other official tax haven lists. The xaxis denotes country level characteristics. Corporate Tax Rate is the maximum corporate tax rate bracket and
Income Tax Rate is the maximum income tax bracket in 2013, obtained through various sources (largely
government agencies and audit firms). ICRG (Property Rights Protection) captures political, economic and
financial risk in 2013 and is obtained from the International Country Risk Guide; the measure ranges from 1
to 6 and increases in protection. Corruption Level is based on Transparency International’s Corruption
Perception Index as of 2013 (Source: Transparency International), an index that measures corruption levels
on a scale from 1 (high corruption) to 10 (low corruption). Tax Evasion is obtained from the Global
Competitiveness Report conducted by the World Economic Forum: Countries’ tax evasion is rated on a scale
from 1 (strongly disagree) to 7 (strongly agree) to the statement ‘Tax evasion is minimal.’. Log (GDP per
capita) is the natural logarithm, of GDP per capita in USD in 2013 (Source: World Bank). Each country
observation is represented by an ‘X’; the line of best fit for equally weighted observations is shown.
44
45
Figure 2: Value and use of tax haven subsidiaries around changes in corporate tax rates
This figure plots changes in the corporate tax rate between 2008 and 2013 against changes in firm value and
changes in the use of tax haven subsidiaries, respectively. Changes in the corporate tax rate are obtained
from KPMG’ Corporate and indirect Tax Rate Survey 2014; a negative value denotes a reduction in
corporate tax rates over the five year period. On the left, the y-axis denotes changes in the difference in
Tobin’s Q from 2008 to 2013. Specifically, the difference between Tobin’s Q of firms with tax haven
subsidiaries in 2008 and firms without tax haven subsidiaries in 2008 is deducted from the respective
difference in 2013. A negative value denotes that firms with tax haven subsidiary have become relatively less
valuable over the five year period. Subsidiary data is collected from Dun & Bradstreet’s Who Owns Whom
2013/2014 book series. Tax havens are sovereign countries or non-sovereign nations that appear on the
OECD grey list (as of August 17, 2009). Tobin’s Q is obtained from Osiris as (Enterprise Value+Total
Liabilities)/(Total Shareholder Equity (Book Value) + Total Liabilities). On the right, the y-axis denotes the
difference between the percentage of firms with tax haven subsidiaries in 2013 and the percentage of firms
with tax haven subsidiaries in 2008. A positive value means that the fraction of firms with tax haven
subsidiary has increased over the five year period. Each country observation is represented by an ‘X’; the
line of best fit for equally weighted observations is shown.
46
Figure 3: Tax Information Exchange Agreements and Treated Firms over time
This figure shows the evolution of passed Tax Information Exchange Agreements (TIEAs) and treated
firms over time. The graph on the left shows all TIEAs passed between two countries or non-sovereign
nations (Source: OECD Harmful Tax Practices). The graph on the right shows the number of publicly
listed firms directly affected by TIEAs at any point in time. A firm is directly affected (treated) if it is
headquartered in a country that signs a TIEA and has a subsidiary in the other signatory country (a tax
haven). Some firms are affected by more than one TIEA: They are counted as treated the moment they
are affected for the first time.
47
Figure 4: Firm Value around the Passage of Tax Information Exchange Agreements (TIEAs)
This figure shows the evolution of firm value of treated firms relative to control firms around the passage of
Tax Information Exchange Agreements (TIEAs). The x-axis denotes years around the passage of TIEAs. The
y-axis shows the interaction between year-to-event dummies and an indicator variable that equals 1 if a firm
is directly affected by a TIEA. Interaction terms are obtained from an OLS regression on a sample of treated
and control firms with the natural logarithm of Tobin’s Q on the left hand side and controls for size and size
squared as well as year and industry fixed effects on the right. Control firms are matched to treated firms 5
years before treatment by headquarter country, industry, as well as the natural logarithm of assets and the
natural logarithm of assets squared.
48
Figure 5: Daily returns of affected firms around the passage of Tax Information Exchange
Agreements (TIEAs)
This figure plots cumulative returns of firms affected by Tax Information Exchange Agreements (TIEAs)
over the 100 days surrounding the signing of a TIEA. A firm is directly affected (treated) if it is
headquartered in a country that signs a TIEA and has a subsidiary in the other signatory country (a tax
haven). Some firms are affected by more than one TIEA: They are counted as treated the moment they are
affected for the first time. Event dates are spread over 10 years (2002 to 2011). Returns are obtained from
Datastream/Worldscope and cumulated; cumulative returns are standardized to equal zero a day before the
signature date.
49

Corporate Tax Havens and Shareholder Value

Transcription

Similar documents

Tropical Haven of Melbourne, Florida

Plisic, Ljiljana

ScanGate document

june 28 july12 july19 july 26 august 9 august 2

Issue 242 - Legal IT Insider

2015-June - MidWest America Federal Credit Union

Annual Report - CHNCT Foundation

An open letter to Jim Ratcliffe, Ineos` owner

Thema 8 Unit 3 File - Camden Haven High School e

a PDF version of our Corporate Gifts brochure.