Mandatory CBC Report Affect MNE BEPS (2024)

Did Private Country-by-Country Reporting Really Not Affect Tax-Motivated International
Income Shifting?
Jillian R. Adams
[email protected]
School of Accounting and Finance
University of Waterloo
May 2024
Keywords: country-by-country reporting, private disclosure, mandatory disclosure, disclosure

regulation, income shifting
This paper is based on my doctoral dissertation at the University of Waterloo. I am extremely thankful to my
supervisor, Kenneth Klassen, for his never-ending support, guidance, and encouragement. I also thank Andrew Bauer
(committee member), Jennifer Blouin, John Campbell, Ted Christensen, Lisa De Simone, Paul Demere, Elizabeth
Demers (committee member), Kaitlyn Kroeger, Tyler Menzer, Devan Mescall (discussant), Erin Towery, Cinthia Valle
Ruiz, Stefan Weck, the NHH-BU Tax Readings Group, the UConn Tax Readings Group, and participants at the 2023
AAA/Deloitte Foundation/J. Michael Cook Doctoral Consortium, the University of Waterloo, Texas A&M University,
the University of Georgia, and the 2024 Tax Policy Research Symposium for their helpful comments and suggestions.
I especially thank Shawn Porter of Deloitte for the insightful conversation. Ethics approval for my dissertation was
obtained from the Research Ethics Board at the University of Waterloo.
Electronic copy available at: https://ssrn.com/abstract=4840684

Abstract
Extant research has failed to find robust evidence that mandatory private country-by-country
reporting (CbCR) disclosure to the tax authorities has significantly decreased multinational
corporations’ (MNCs’) tax-motivated international income shifting. These results have been
widely cited and interpreted as evidence that MNCs’ income shifting has not decreased after the
implementation of private CbCR. I revisit this question by replicating the results of prior research,
lengthening the sample period, and expanding the sample beyond the EU. Using affiliate-level
income shifting models, I find robust evidence that MNCs shift significantly less income in the
post-CbCR implementation period. On average, MNCs have decreased their income shifting by
half since the implementation of CbCR, and these decreases are monotonic across MNC size and
not concentrated amongst MNCs subject to mandatory private CbCR. Altogether, I show that
CbCR generally achieved its overall objective of reducing income shifting; however, I provide
compelling evidence that the CbCR consolidated revenue threshold did not matter, making it
unlikely that it was the CbCR disclosure to the tax authorities that resulted in the decrease in
income shifting. Consistent with the overarching focus of the OECD’s BEPS framework, I find
that MNCs headquartered in countries with strong regulatory environments, an interest deduction
limitation consistent with BEPS Action 4, and no preferential tax regimes named in BEPS Action
5, exhibit greater decreases in income shifting relative to MNCs headquartered in countries with
weak regulatory environments and not compliant with Action 4 or Action 5.

1. Introduction
Tax authorities, policymakers, media, and the public scrutinize MNCs’ international tax
planning activities (e.g., The Guardian [2009], Bergin [2012], Organisation for Economic Co-
operation and Development (OECD) [2013], Wayne et al. [2014]). Due to their international
footprint and resource availability, MNCs have historically exploited opportunities or gaps in
international tax laws and shifted income amongst their geographically-dispersed affiliates.
Shifting income to low-tax or no-tax jurisdictions can improve after-tax income by reducing the
MNC’s global income tax burden. The OECD [2024a] estimates that this type of behavior is
significant, resulting in lost global tax revenues of 100-240 billion USD annually. Disclosure is
one potential method of addressing this global tax challenge (OECD [2024a]). I study the
following research question: Did mandatory private country-by-country reporting (CbCR)
disclosure to the tax authorities significantly decrease the tax-motivated international income
shifting of multinational corporations (MNCs)?
To combat base erosion and profit shifting (BEPS), the OECD developed a framework
comprising 15 action items. As of November 15, 2023, 145 countries have agreed to implement
these action items (OECD [2024a]). One of the first items implemented, Action 13 Country-by-
Country Reporting, requires large MNCs to prepare CbCR for submission to the tax authorities
(hereinafter referred to as “private CbCR”). The OECD [2015] has issued guidance and a template
on information in CbCR; this includes total and related-party revenue, pre-tax income, income
taxes paid and accrued, stated capital, accumulated earnings, and real activities (tangible assets,
number of employees) in each country. The OECD [2013] proposes that the level of real activity
in a jurisdiction is the key driver of the appropriate income reported in that jurisdiction, suggesting
that CbCR can be used to identify MNCs’ income shifting behavior. Tax authorities can then use
-3-

this information to improve their tax audit selection and increase the cost of MNCs’ income
shifting behavior.
While disclosure has the potential to alter tax planning behavior, prior research has failed
to find robust evidence that private CbCR disclosure has resulted in a significant decrease in
MNCs’ income shifting: Joshi [2020] finds some evidence that European MNCs subject to private
CbCR exhibit a small decrease in income shifting in 2018, as detailed below, whereas Nessa et al.
[2022] fail to find evidence that US MNCs decrease their income shifting any year after the
implementation of CbCR. Instead, prior research has found that MNCs, except for US MNCs
(Nessa et al. [2022]), increase their real activities in low-tax European countries after the
implementation of private CbCR (De Simone and Olbert [2022]), likely to substantiate the income
reported in these countries.
In Joshi’s [2020] primary income shifting test (Table 8 Panel A column 2), she uses a
modified version of the Huizinga and Laeven [2008] affiliate-level income shifting model on a
sample of 123,920 affiliate-years from 2010 to 2018 belonging to MNCs headquartered in the EU.
She finds a positive but insignificant coefficient on the interaction of C, the tax incentive to shift
income, and CbCR, the indicator variable for MNCs subject to private CbCR (0.513, t-stat of
1.145); a positive but insignificant coefficient on the interaction of C and Post, the indicator
variable for the years after the implementation of private CbCR in 2016 (0.349, t-stat of 0.709);
and a negative but insignificant coefficient on the interaction of C, CbCR, and Post (-0.140, t-stat
of -0.273).1 When she breaks down the C×CbCR×Post interaction into the three post years, she
finds that the coefficient on C×CbCR×2018 is positive and significant at the five percent level in
one of three regressions using her full sample of EU MNCs (2.393, t-stat of 2.330 in her Table 8
1
Note that Joshi [2020] refers to the Huizinga and Laeven [2008] C variable as “π” and reports standard errors which
I convert into t-statistics for ease of comparability.
-4-

Panel B column 1). Joshi [2020] interprets her results as private CbCR having “a limited effect on
tax-motivated income shifting” (p. 334). This impression of limited effectiveness has been widely
cited. Others, in citing her work, have inferred her results to mean that private CbCR has had little
to no impact on income shifting (e.g., Hoopes et al. [2023]; Kobbi-Fakhfakh and Driss [2022];
Laudage [2023]).
That private CbCR did not appear to significantly decrease MNCs’ income shifting is worth
further examination. International income shifting typically involves highly complex structures
spanning multiple countries and multiple layers of corporate entities. These structures are costly
to implement in terms of both time and money and, therefore, can take significant time to dismantle
and replace with other tax-efficient structures. Thus, it is plausible that the effect of private CbCR
on income shifting would be seen in the lengthening of the sample period to 2021. Additionally,
there may be country-level differences in MNCs’ income shifting responses, as evidenced by
Joshi’s [2020] evidence of a decrease in income shifting among affiliates located in countries with
stronger tax enforcement. Thus, it is also plausible that the effect of private CbCR on income
shifting would be seen in the expansion to a global sample of MNCs.
To answer my research question, I begin by replicating the results of Joshi [2020]. I then
expand beyond the sample of Joshi [2020]. First, I remove the sample restriction that all MNCs
must have an international securities identification number (ISIN) listed in Orbis, which
effectively limits the sample to very large public MNCs. Simply by eliminating this restriction to
allow for smaller public MNCs and private MNCs in the sample, I find that EU MNCs – both those
subject to private CbCR and those below the reporting size threshold – significantly decrease their
income shifting after the implementation of private CbCR on average (coefficient of 0.752 on
-5-

C×Post, t-stat of 3.308). Inferences are unchanged when I use the Huizinga and Laeven [2008]
and the De Simone et al. [2017] affiliate-level income shifting models.
Next, I lengthen the sample period from 2018 to 2021. I continue to find statistically
significant evidence that all EU MNCs – both treated and untreated – on average, exhibit a
significant decrease in income shifting in the post-CbCR implementation period. Then, I expand
the sample to MNCs headquartered in any country with a private CbCR regulation implemented
during the sample period. I again find statistically significant evidence that both treated and
untreated MNCs, on average, exhibit a significant decrease in income shifting in the post-CbCR
implementation period.
To further explore my new evidence, I refine the time and treatment measures into finer
gradations. I find that MNCs subject to private CbCR were originally less aggressive income
shifters than those not subject to private CbCR, with the smallest MNCs, based on consolidated
revenue, being the most aggressive income shifters in the pre-period. All MNCs, on average,
exhibited statistically significant decreases in income shifting from 2016 to 2019, and these
decreases are monotonic across MNC size.
Altogether, I find robust evidence that MNCs’ income shifting has significantly decreased
alongside the implementation of CbCR, with estimates suggesting that, on average, MNCs have
decreased their income shifting by approximately half.2 However, based on the difference-in-
differences coefficients and the monotonic decreases in income shifting across MNC size, I
provide compelling evidence that the CbCR consolidated revenue threshold did not matter, making
it unlikely that it was the CbCR disclosure to the tax authorities that resulted in the decrease in
income shifting.
2
Table 5 Panel B: CbCR=0: (1.155 – 0.671) / 1.155, CbCR=1: (0.786 – 0.335) / 0.786
-6-

Several possible explanations exist for the observed effect on non-treated MNCs, such as
peer effects or anticipation that the regulations would extend to smaller MNCs. I assert that it is
the OECD’s BEPS framework in general that is contributing to the reduction in income shifting.
Several other BEPS action items that target income shifting were introduced alongside private
CbCR, such as Action 4 Limitation on Interest Deductions. Notably, these action items generally
apply to all MNCs by having de minimis thresholds below that of private CbCR – meaning that
they apply to the non-treated MNCs in my sample.
In additional analysis, I explore cross-sectional differences in the regulatory environments
of MNCs’ headquarter-countries to provide evidence for this assertion. I find that MNCs
headquartered in countries with stronger regulatory environments exhibit greater decreases in
income shifting in the post-period than MNCs headquartered in countries with weaker regulatory
environments. This finding is consistent with countries with strong regulatory environments being
more likely to implement and enforce the BEPS action items, increasing the cost of MNCs’ income
shifting in those countries. In a more direct test, I find some evidence that MNCs headquartered in
countries not compliant with BEPS Action 4 Limitation on Interest Deductions or Action 5 Harmful
Tax Practices – Preferential Tax Regimes exhibit smaller decreases in income shifting in the post-
period than MNCs headquartered in countries that comply with Action 4 and Action 5. This result
is consistent with my assertion that the BEPS framework in general contributed to the reduction in
income shifting.
My results are important to governments and policymakers worldwide, particularly the
OECD. I provide much-needed and timely insight into whether disclosure is addressing the
ongoing global BEPS challenge. This evidence is critical to document before the implementation
of public CbCR in 2024, so as not to confound the impacts of the regulations. Additionally, my
-7-

results suggest that public CbCR and the OECD’s proposed BEPS Pillar II rules (global anti-base
erosion rules, e.g., the implementation of a global minimum tax) may be unnecessary since large
MNCs have already drastically decreased their income shifting levels. Pillar II is a legislated and
even more complex response to income shifting relative to the solution of private CbCR disclosure.
These results should also be of interest to academic researchers. It has been widely cited
that MNCs’ income shifting has not decreased after the implementation of private CbCR. To my
knowledge, I provide the first global evidence to amend this perception by showing that CbCR is
associated with a significant decrease in income shifting, and I reconcile this with prior results. I
note that Joshi [2020] was clear about the ISIN sample restriction and included a caveat that her
results may not hold in a broader sample of MNCs. However, the sample restriction and caveat are
generally not mentioned in subsequent citations of Joshi [2020], which may have contributed to
this perception. I further provide evidence to suggest a broader exploration of all the BEPS
provisions may yield important insights.
The remainder of the paper proceeds as follows: Section 2 reviews prior research and
develops the hypothesis; Section 3 describes the sample, data sources, and research methodology;
Section 4 discusses the results; Section 5 contains additional analyses; and Section 6 concludes.
2. Background and Hypothesis Development
2.1 Mandatory Tax Disclosure
Governments and regulators mandate tax disclosure. Most of the disclosures are private
(i.e., submitted to the tax authorities and not released publicly).3 Other disclosures, such as
financial statements, are public. Both private and public disclosures are believed to constrain tax
3
Some exceptions exist, such as the Australian Tax Office’s public disclosure of large corporations’ total income,
taxable income, and taxes payable (Hoopes et al. [2018]; Kays [2022]).
-8-

behavior by altering the net benefit of undertaking the tax planning activity via three hypothesized
mechanisms (Müller et al. [2020]). First, disclosure gives the tax authorities more information,
improving their tax audit selection and efforts to ensure compliance with the tax law. Second,
policymakers can use the information to find and revise gaps in tax legislation, potentially leading
to new, stricter laws. Third, public disclosure can result in public scrutiny and attention to overly
aggressive choices, giving firms a greater sense of accountability. Extant research has found that
mandatory tax disclosure can discipline tax behavior in many settings, especially those where firms
do not have much discretion over the quantity or quality of the information provided (see review
by Hoopes et al. [2023]).
2.2 Mandatory Private CbCR and Tax-Motivated International Income Shifting
The OECD announced mandatory private CbCR disclosure in 2013, and it was
implemented in 2016 in the EU and several other countries, including the United States, Australia,
and Canada.4 Approximately 115 countries have implemented private CbCR regulations to date
(OECD [2024b]). Appendix A details the private CbCR regulations for all countries in this study’s
global sample.
Mandatory private CbCR disclosure to the tax authorities is hypothesized to constrain tax
behavior, specifically international tax-motivated income shifting, through Müller et al.’s [2020]
first and second mechanisms. Except for the US Internal Revenue Service (IRS), which requires
foreign corporation reporting and foreign disregarded entity reporting, tax authorities worldwide
were generally not privy to the international activities of MNCs and their affiliates. Thus, CbCR
constituted new information that the tax authorities could use in risk assessments and audits. The
increased tax authority information and the threat of better-targeted audits alter the net benefit of
4
For more detailed background information on private CbCR, see Joshi [2020].
-9-

income shifting. Potential responses include decreasing aggressive income shifting behavior or
reallocating real activity to low-tax countries; the latter aims to substantiate the income reported
in those countries (Hanlon [2018]).
Prior research has failed to find robust evidence that private CbCR has resulted in a
significant decrease in MNCs’ income shifting. Using a sample of affiliate-years belonging to
MNCs headquartered in the EU from 2010 to 2018, Joshi [2020] finds some evidence that EU
MNCs subject to private CbCR exhibit a small decrease in income shifting in 2018 (coefficient of
2.393 on C×CbCR×Post, t-stat of 2.330 in her Table 8 Panel B column 1). However, this evidence
is found in only one of three regressions using her unrestricted sample of MNCs and their affiliates.
Joshi [2020] also finds that affected EU MNCs with affiliates located in countries with above-
median tax enforcement dollars (scaled by GDP) decrease their income shifting after the
implementation of private CbCR (coefficient of 2.122 on C×CbCR×Post, t-stat of 1.801 in her
Table 9 Panel A column 1).
De Simone and Olbert [2022] consider whether MNCs responded to CbCR regulations by
altering their economic footprint to better match where their income is reported. Using a sample
of EU MNCs from 2012 to 2018, they find that affected MNCs reduced their number of affiliates
located in tax haven countries, and increased their tangible assets and labor expenses in low-tax
European countries after the implementation of private CbCR. Broadly consistent with De Simone
and Olbert’s [2022] evidence, Joshi et al. [2023] use country-level aggregate CbCR data and
foreign direct investment data to show that MNCs have realigned their fixed assets across countries
since the implementation of CbCR, such that the location of their fixed assets better match where
their income is reported. These actions reduce the measured intensity of income shifting.
- 10 -

Using confidential taxpayer return data, Nessa et al. [2022] fail to find evidence that
affected US MNCs decreased their income shifting after the implementation of private CbCR.
Moreover, they do not find evidence to suggest that these US MNCs alter their real activities. The
authors posit that this may be because of the lack of new information that CbCR provides to the
IRS due to the pre-existing foreign corporation and foreign disregarded entity reporting.
There are at least two reasons why the relation between private CbCR regulation and
income shifting may differ from that found by prior research. First, Joshi [2020] finds some
evidence that affected EU MNCs decrease their income shifting beginning in 2018. This evidence
is consistent with the notion that income shifting structures may take considerable time to
dismantle and replace with other tax-efficient structures. If this is the case, stronger evidence of
decreased income shifting may be seen in a longer sample period.
Second, many countries have implemented CbCR beyond the EU and the US. Given that
Joshi [2020] finds a decrease in income shifting among affiliates located in countries with greater
tax enforcement, and Nessa et al. [2022] fail to find evidence that US MNCs decrease their income
shifting, there are country-level differences in MNCs’ income shifting response to private CbCR
regulation. Thus, it is also plausible that stronger evidence of decreased income shifting would be
seen in the expansion to a global sample of MNCs.
I state my hypothesis as follows:
H1: Relative to MNCs not subject to mandatory private CbCR, MNCs subject to mandatory
private CbCR exhibit a greater decrease in their tax-motivated international income shifting
after the implementation of mandatory private CbCR.
Alternatively, it is possible that all MNCs decreased their income shifting alongside the
implementation of private CbCR, not just MNCs subject to private CbCR disclosure to the tax
- 11 -

authorities. This regulation was part of a broader groundswell of policy efforts to reduce income
shifting extensively, and all MNCs may have reacted to the group of policy efforts.
For example, BEPS Action 4 Limitation on Interest Deductions was introduced in 2015
and implemented by 67 countries as of 2019 (OECD [2024a]). Action 4 directly addresses one of
the most well-known and easiest income shifting techniques MNCs have available to them – the
structuring of debt such that the greatest interest deductions are taken in high tax jurisdictions to
offset income reported in those jurisdictions and, in the case of related party debt, the greatest
interest income is reported in low- or no-tax jurisdictions. Action 4 addresses this by generally
limiting the interest deductible for tax purposes based on the MNC’s income reported in a country.
Unlike Action 13 Country-by-Country Reporting, Action 4 generally applies to all MNCs by
having de minimis thresholds below that of Action 13.5 Thus, all MNCs may have decreased
income shifting alongside the implementation of private CbCR, given that a key income shifting
technique has been significantly limited in the same time period, which increases the cost of
income shifting.
Consistent with this conjecture, an international tax Partner at a Big 4 accounting firm said
the following in our discussion about my study:6
That the introduction of CbC reporting was followed by a decrease in profit shifting
by in-scope MNEs is not surprising. However, in my opinion, the profit shifting
behaviours of MNEs were significantly more responsive to the substantive law
changes introduced in many countries during the period covered by the study. Such
substantive law changes were largely inspired by OECD BEPS Action Reports
finalized in 2015, including BEPS Action 2 Neutralizing the effects of hybrid
mismatch arrangements (curtailing the use of hybrid entities and instruments that
had been used to achieve deduction/no inclusion and double deduction outcomes,
among other things); Action 4 Limitations on Interest Deductions (implementing
in effect a global formulary approach – based on operating income – to limit base
5
For example, the US had a de minimis gross receipts threshold of USD 25 million in 2019 for their interest limitation
rule, and the UK had a de minimis consolidated interest expense threshold of GBP 2 million in 2019 for their earnings
stripping rule.
6
Ethics approval was obtained from the Research Ethics Board at my university.
- 12 -

erosion through interest expense in any particular country); Action 5 Harmful tax
practices (adopting a modified nexus approach in an effort to better align reported
profits with local activities and control over functions and risks – even prompting
some countries to modify their patent and intellectual property box (and similar)
regimes in an effort to keep profits onshore); and Actions 8-10 Transfer Pricing
(which lead to further updates to the OECD’s Transfer Pricing Guidelines in the
ensuing years).
It is also possible that MNCs did not alter their income shifting behavior after the
implementation of private CbCR. Although private CbCR constitutes new information for most
tax authorities, it is not necessarily new information tax authorities can act on, as income shifting
is typically implemented through legal structures. Thus, MNCs may not have felt pressure to
decrease their income shifting.
3. Sample and Research Design
3.1 Sample Selection
Consistent with Joshi [2020], I begin with the population of global ultimate owner (GUO)
corporations in Orbis that have at least one affiliate located in a country other than the GUO’s
headquarter-country. I drop all GUOs that are financial institutions because they are subject to
unique regulations, including the EU’s public CbCR mandate. To remain consistent with Joshi
[2020], I also drop all GUOs in extractive industries because most are subject to ‘payments to
governments’ reporting, which includes total taxes paid by country. Then, I drop all GUOs
headquartered outside of the EU and GUOs missing consolidated revenue or total assets. I then
obtain all affiliates belonging to these MNCs. I drop all affiliates missing NACE industry codes
and required financial statement data, and I drop loss affiliates.7 As detailed in Table 1 Panel A,
my EU sample has 289,231 affiliate-years from 2010 to 2021.
7
To get a more complete picture of MNCs’ tax-motivated income shifting behavior, I do not drop affiliates operating
in extractive or financial industries. In untabulated analysis, I drop affiliates operating in extractive and financial
industries and all inferences remain the same.
- 13 -

I follow the same sample selection procedure for my global sample. However, instead of
dropping all GUOs headquartered outside of the EU, I drop all GUOs headquartered in countries
that did not implement a private CbCR regulation in the sample period.8 I then obtain all affiliates
belonging to the global sample of MNCs. After sample restrictions for missing data and loss firms,
my global sample consists of 510,324 affiliate-years from 2010 to 2021. Detailed sample selection
criteria can be found in Table 1 Panel A, and sample composition by year, country, and industry
can be found in Table 1 Panels B, C, and D, respectively.
3.2 Research Design
I test hypothesis 1 using the Huizinga and Laeven [2008] and De Simone et al. [2017]
affiliate-level income shifting models modified for a difference-in-differences design as follows:
𝑙𝑛𝐸𝐵𝐼𝑇𝑖,𝑡 = 𝛽0 + 𝛽1 𝐶𝑖,𝑡 + 𝛽2 𝐶𝑏𝐶𝑅𝑀 + 𝛽3 𝐶𝑏𝐶𝑅𝑀 × 𝑃𝑜𝑠𝑡𝑡 + 𝛽4 𝐶𝑖,𝑡 × 𝐶𝑏𝐶𝑅𝑀 + 𝛽5 𝐶𝑖,𝑡 × 𝑃𝑜𝑠𝑡𝑡

+ 𝛽6 𝐶𝑖,𝑡 × 𝐶𝑏𝐶𝑅𝑀 × 𝑃𝑜𝑠𝑡𝑡 + 𝛽7 𝑙𝑛𝐹𝑖𝑥𝑒𝑑𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + 𝛽8 𝑙𝑛𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛𝑖,𝑡
+ 𝛽9−12 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑖,𝑡 + 𝐹𝑖𝑥𝑒𝑑𝐸𝑓𝑓𝑒𝑐𝑡𝑠 + 𝜀𝑖,𝑡 (1)
where C equals either the Huizinga and Laeven [2008] revenue-weighted tax rate differential:
𝐵𝑗
𝐻𝐿
𝐶𝑖,𝑡 = 1 ∑𝑛𝑗≠𝑖( )(𝜏𝑖 − 𝜏𝑗 ) (2a)
1− 𝜏𝑗
𝐵𝑗
(1 − 𝜏𝑖 ) ∑𝑛𝑗=𝑖( )
1− 𝜏𝑗
or the De Simone et al. [2017] capital-weighted tax rate differential:

𝐷𝐾𝑆
𝐶𝑖,𝑡 = ∑𝑛≠𝑖 𝐾𝑛 (𝜏𝑖 − 𝜏𝑛 ) (2b)
∑𝑛 𝐾𝑛
In equation 1, the dependent variable, lnEBIT, equals the natural logarithm of affiliate i’s
earnings before interest and tax. C measures the affiliate’s tax incentive to shift income per
equation 2a or 2b, which is alternatively the revenue-weighted or capital-weighted tax rate
8
To get a more complete picture of MNCs’ tax-motivated income shifting behavior, I do not restrict the sample to
affiliates located in countries that implemented a private CbCR regulation in the sample period.
- 14 -

differential between the affiliate and all other affiliates of the MNC M in year t, where B is the
affiliate’s revenue, K is the affiliate’s capital, and τ is the affiliate’s statutory tax rate. CbCR is an
MNC-level indicator variable that equals one where the MNC group is subject to private CbCR
based on either an ultimate parent entity filing obligation or a local filing obligation as outlined in
Appendix A, and zero otherwise. Post is a time-period indicator variable that equals one for the
years when private CbCR is required as outlined in Appendix A, and zero otherwise. Post is not
included in equation 1 as it is subsumed by year fixed effects.
The coefficient on the interaction of C and CbCR, β4, is an empirical estimate of the average
response to the tax incentive to shift income for MNCs subject to private CbCR, compared to those
not subject to private CbCR in the period before the implementation of private CbCR. The
coefficient on the interaction of C and Post, β5, is an empirical estimate of the average response to
the tax incentive to shift income for MNCs not subject to private CbCR in the years after the
implementation of private CbCR, relative to the years before the implementation of private CbCR.
Joshi [2020] found neither of these coefficients to be statistically different from zero.
The coefficient on the interaction of C, CbCR, and Post, β6, is my coefficient of interest. β6
is the standard difference-in-differences coefficient. β6 is an empirical estimate of the differential
average response to the tax incentive to shift income for MNCs subject to private CbCR in the
years after versus before the implementation, relative to MNCs not subject to private CbCR. I
expect β6 to be positive, consistent with hypothesis 1; however, Joshi [2020] found this coefficient
to be generally statistically indistinguishable from zero. Because tax-motivated income shifting
implies β1 is negative, a positive coefficient is consistent with MNCs subject to private CbCR
decreasing their tax-motivated income shifting after the implementation of private CbCR to an
extent greater than MNCs not subject to private CbCR.
- 15 -

Based on the income shifting literature (Huizinga and Laeven [2008], De Simone et al.
[2017], De Simone et al. [2022]), I control for various factors that predict the affiliate’s economic
earnings. In the Huizinga and Laeven [2008] variation of the income shifting model, I control for
the affiliate’s fixed assets (lnFixedAssets), the affiliate’s labor expense (lnCompensation), and
industry fixed effects to control for industry profitability, as these are critical to the Cobb-Douglas
production function. I also include country-level GDP per capita (lnGDPperCapita) to proxy for
economic shocks.
In the De Simone et al. [2017] variation of the income shifting model, I control for the
affiliate’s fixed assets (lnFixedAssets), the affiliate’s labor expense (lnCompensation), and industry
profitability (IndustryROA), as these are critical to the Cobb-Douglas production function. I also
include the change in country-level GDP (ΔGDP) to proxy for economic shocks, the age of the
affiliate (lnAge), and the change in the industry market size (ΔMktSize), as these factors may affect
profitability.
In both models, year fixed effects are included and standard errors are clustered at the MNC
level. All data are in US dollars, and all variables are defined with their sources in Appendix B.
4. Results
4.1 Summary Statistics
Summary statistics are reported in Table 2 Panel A and are largely consistent with the
income shifting literature (e.g., the summary statistics of the global sample in De Simone et al.
[2022]). I separately report summary statistics for affiliates of MNCs that are and are not subject
to private CbCR. The last three columns of Table 2 Panel A indicate significant differences between
these affiliate groups. Affiliates of MNCs that are subject to private CbCR are more profitable,
- 16 -

larger, have greater compensation expenses, and are older. They are more likely to operate in
countries with greater GDP but lesser GDP per capita.
In untabulated analysis, I use entropy balancing to address these observable differences
and balance the proportion of the sample groups (Hainmueller [2012]). Entropy balancing is a
form of pseudo-matching that reweights the groups to ensure that at least the first and second
moments of each group’s covariate distribution are alike. This approach helps to address measured
confounding between the groups. I entropy balance on the key control variables used in the
Huizinga and Laeven [2008] income shifting model, lnFixedAssets, lnCompensation, and
lnGDPperCapita, and all inferences remain unchanged.
Pearson correlation coefficients are reported in Table 2 Panel B, with correlations
statistically significant at the five percent level denoted in bold. Correlations are generally low
except for similar measures (e.g., the alternate definitions of C), which are not used together in the
same model, and the profitability measures with assets and compensation.
4.2 Replicating the Results of Joshi [2020]
Before testing my hypothesis, I first replicate the results of Joshi [2020]. To match Joshi’s
[2020] sample, I retain all affiliates of EU MNCs in Table 1 Panel A with observations from 2010
to 2018, which gives me 192,595 affiliate-year observations. Then, I implement Joshi’s [2020]
sample restriction that all MNCs must have an ISIN listed in Orbis, which gives me 134,070
affiliate-year observations. This is very close to Joshi’s [2020] main sample of 123,920 affiliate-
year observations in her Table 8 Panel A column 2.
I then amend the affiliate-level income shifting model in equation 1 for Joshi’s [2020]
variable definitions. Instead of using lnEBIT as the dependent variable, Joshi [2020] uses lnPTI,
the natural logarithm of the affiliate’s pre-tax income. Joshi [2020] also uses the natural logarithm
- 17 -

of GDP (lnGDP). Joshi’s [2020] Table 8 Panel A column 2 states that MNC-level fixed effects are
used; however, the CbCR indicator variable is not subsumed in the model, so I assume MNC fixed
effects are not used.9
Joshi’s [2020] Table 8 Panel A column 2 is included in Table 3 column 1 alongside my
replication in column 2. Consistent with Joshi [2020], I find positive but insignificant coefficients
on the interactions of C with CbCR and C with Post. The coefficient on the interaction of interest,
C×CbCR×Post is negative and insignificant, again consistent with the results of Joshi [2020].
These coefficients are also of similar magnitude to the coefficients reported in Joshi [2020].
Altogether, this test fails to find evidence that this sample of EU MNCs subject to private CbCR
decreases their income shifting after the implementation of private CbCR, relative to EU MNCs
not subject to private CbCR.
Next, I explore the impact of the ISIN sample restriction. Joshi [2020] implemented this
restriction to match MNC-year observations in Orbis to Compustat, as she required Compustat’s
cash taxes paid variable in order to conduct tests on the change in ETRs after the implementation
of private CbCR. As income shifting models do not require cash taxes paid, I remove this
restriction and estimate Joshi’s [2020] model on the sample of 192,595 affiliate-year observations
from 2010 to 2018 that belong to EU MNCs. I report this in Table 3 column 3.
Interestingly, the coefficient on the interaction of C and Post is positive and statistically
significant at the one percent level (0.752, t-stat of 3.308). This result suggests that EU MNCs not
subject to private CbCR exhibit a significant decrease in income shifting in the post-CbCR
implementation period. However, the coefficient on the interaction of interest, C×CbCR×Post, is
negative and statistically significant at the five percent level (-0.553, t-stat of -2.068).
9
In untabulated analysis, I include MNC fixed effects and all inferences remain unchanged.
- 18 -

Next, I use two variations of income shifting models on the sample of affiliates of EU
MNCs from 2010 to 2018 to determine if the model and variable definitions affect the results.
Results using the Huizinga and Laeven [2008] and De Simone et al. [2017] income shifting models
are reported in Table 3 columns 4 and 5, respectively, and provide similar inferences to the results
in column 3.10 Put together, these results suggest that EU MNCs subject to CbCR do not decrease
their income shifting in the post-CbCR implementation period to the same extent as EU MNCs not
subject to CbCR. However, I do not make any formal conclusions until considering the global
sample of MNCs from 2010 to 2021.
4.3 Exploring the Impact of the ISIN Sample Restriction
Given that the ISIN sample restriction drives these substantially different results, I explore
the differences in characteristics of EU MNCs with and without an ISIN listed in Orbis. As detailed
in Table 4 Panel A, affiliates of EU MNCs with an ISIN listed in Orbis are more profitable, larger,
have greater compensation expense, and are more likely to operate in countries with a higher GDP
and GDP per capita compared to affiliates of EU MNCs without an ISIN listed in Orbis. Perhaps
most notably, 83% of affiliate-year observations in the ISIN-restricted sample are subject to private
CbCR. Thus, the lack of a control group may drive the different results instead of the large firm
bias imposed by the ISIN restriction. I explore the role of MNC size in greater detail below.
In Table 4 Panel B, I detail the sample composition based on MNC headquarter-country
for the samples with and without the ISIN restriction. For brevity, I list only countries that account
for more than five percent of the broader sample. By restricting the sample to EU MNCs with an
10
In untabulated analysis, I estimate the Huizinga and Laeven [2008] and De Simone et al. [2017] income shifting
models on the sample of 134,070 affiliate-years where the EU MNCs have an ISIN listed in Orbis. Consistent with
the results of Joshi [2020], I find positive but insignificant coefficients on C×Post and negative but insignificant
coefficients on C×CbCR×Post.
- 19 -

ISIN listed in Orbis, MNCs headquartered in France and the UK are overrepresented; in contrast,
MNCs headquartered in Germany, Spain, and Italy are underrepresented.
To explore whether the overrepresentation affected results, I estimate the Huizinga and
Laeven [2008] and De Simone et al. [2017] income shifting models on MNCs headquartered in
France and the UK. I report these results in Table 4 Panel C columns 1 and 2. In both columns, the
coefficients on C×Post, C×CbCR, and C×CbCR×Post are all insignificant. Thus, I fail to find
evidence that MNCs headquartered in France and the UK have decreased their income shifting
after the implementation of private CbCR. To explore this result further, I create a 2x2 matrix of
France and UK MNCs’ C coefficients from column 1. I report this matrix in Table 4 Panel D. F-
tests of the implied coefficients on C in each of the CbCR=1, Post=0; CbCR=0, Post=1; and
CbCR=1, Post=1 quadrants reveal that all are statistically significantly different than zero at the
five percent level or greater. This result is consistent with France and UK MNCs still engaging in
significant income shifting after the implementation of private CbCR, and the overrepresentation
of these MNCs in the Joshi [2020] sample may have driven the results she found.
4.4 Test of Hypothesis 1
I formally test my hypothesis on the EU sample of MNCs from 2010 to 2021 in Table 5
Panel A columns 1 and 2, and on the global sample of MNCs from 2010 to 2021 in columns 3 and
4. I estimate the Huizinga and Laeven [2008] income shifting model in columns 1 and 3 and the
De Simone et al. [2017] income shifting model in columns 2 and 4.
In the EU sample, the coefficients on the interaction of C and CbCR are positive but not
statistically significant at conventional levels. In the global sample, the coefficient on the
interaction of C and CbCR is positive and statistically significant at the ten percent level in column
3 (0.369, t-stat of 1.710) and positive and statistically significant at the five percent level in column
- 20 -

4 (0.712, t-stat of 2.091). These tests provide some evidence that MNCs subject to private CbCR
were originally less aggressive income shifters than those not subject to private CbCR.
The coefficients on the interaction of C and Post are positive across all specifications and
are statistically significant at the five percent level or greater in the EU sample (0.661, t-stat of
3.073 in column 1; 0.678, t-stat of 1.988 in column 2) and at the one percent level in the global
sample (0.484, t-stat of 2.675 in column 3; 0.645, t-stat of 2.267). As in Table 3, this result suggests
that MNCs not subject to private CbCR exhibit a decrease in income shifting in the post-CbCR
implementation period.
The coefficient on the interaction of C, CbCR, and Post is my coefficient of interest. Per
hypothesis 1, I predict this coefficient will be positive. A positive coefficient is consistent with
MNCs subject to private CbCR decreasing their tax-motivated income shifting after the
implementation of private CbCR, relative to MNCs not subject to private CbCR. However, I do
not find evidence to support this hypothesis. The coefficients on C×CbCR×Post are negative and
not statistically significant across all specifications.
To better understand these results, I create a 2x2 matrix of the C coefficients from the global
sample of MNCs in column 3. This matrix is reported in Table 5 Panel B. These coefficients
illustrate the main results: (1) large MNCs subject to private CbCR exhibit less aggressive income
shifting behavior in the pre-period, relative to MNCs in the pre-period that are not subject to private
CbCR, and (2) after the implementation of private CbCR, all MNCs in the sample decrease their
income shifting, on average. I posit that since MNCs subject to private CbCR already had lower
income shifting in the pre-period, these MNCs do not have much more income shifting to decrease
in the post-period relative to the income shifting reductions of the MNCs not subject to private
CbCR. Thus, no significant difference is observed in the treated group in the post-period.
- 21 -

In summary, the difference-in-differences estimator does not allow me to make a causal
statement that CbCR reporting to the tax authorities decreased treated MNCs’ income shifting.
However, on average, MNCs subject to private CbCR shift significantly less income across the
entire sample period than MNCs not subject to private CbCR. Moreover, in the post-CbCR
implementation period, all MNCs shift significantly less income, on average. Thus, CbCR
generally achieved its overall objective of reducing income shifting. I explore the role of the CbCR
consolidated revenue threshold in greater detail below.
In untabulated analysis, I restrict both the EU and global samples to MNCs with an ISIN
listed in Orbis and, consistent with Joshi [2020], I fail to find statistically significant coefficients
on C×CbCR, C×Post, and C×CbCR×Post.
4.5 Stacked Difference-in-Differences Design
When the global sample of MNCs is used, staggered treatment timing exists due to the
various CbCR implementation years listed in Appendix A. To the extent that treatment effects are
not homogenous, the difference-in-differences design employed in equation 1 would not be an
appropriate tool to use with the staggered treatment timing as the difference-in-differences
estimator uses already-treated observations as controls, which may violate the parallel trends
assumption (Barrios [2021], Goodman-Bacon [2021], Sun and Abraham [2021]). However, this
bias is unlikely to be of significant concern in my setting given that 97% of the global sample is
subject to CbCR effective 2016. Nonetheless, in untabulated analysis, I implement a stacked
variation of equation 1 following Cengiz et al. [2019] and Baker et al. [2022], where all
observations are “stacked” so that they are treated at relative time 0. All inferences remain the
same.
- 22 -

4.6 Finer Gradations on Post and CbCR
Given that both the time period and MNC size are associated with a significant decrease in
income shifting, I split the two key indicator variables, Post and CbCR, into finer gradations to
better understand these relations. In Table 6 Panel A, I explore how income shifting has changed
over time. As income shifting structures can take time to change, and there may not be significant
change year-to-year, I separate the Post indicator into three groups of two years: 2016-2017, 2018-
2019, and 2020-2021, and I also include 2014-2015, the two years before the first implementation
of private CbCR. I report the results of this analysis on the global sample, using the Huizinga and
Laeven [2008] income shifting model in column 1 and the De Simone et al. [2017] income shifting
model in column 2.11
Consistent with Table 5 Panel A, there is evidence to suggest that MNCs subject to private
CbCR were originally less aggressive income shifters than the control group. Additionally, MNCs
exhibited statistically significant decreases in income shifting from 2016 to 2019. In Appendix C
Panel A, I graph the yearly C coefficients in column 1 to illustrate the decrease in income shifting
over time for the group subject to private CbCR and the group not subject to private CbCR. The
blue and black lines represent the income shifting of the group subject to private CbCR and the
group not subject to private CbCR, respectively. The graph shows that the group subject to private
CbCR has, on average, ceased income shifting by 2020-2021, whereas the group not subject to
private CbCR continues to have a negative coefficient on C, but has still exhibited a significant
decrease in income shifting. Interestingly, the slope of the income shifting line from 2018-2019 to
2020-2021 for the group not subject to private CbCR is statistically different from zero (F-stat
11
In untabulated analysis, I perform this analysis using the EU sample only and all inferences remain unchanged.
- 23 -

3.42, 0.064), providing some evidence that these MNCs have increased their income shifting over
this period. I look forward to future research exploring whether this behavior continues.
The analysis in Table 6 Panel A and the graph in Appendix C Panel A use the staggered
difference-in-differences design. Since 97% of the global sample is subject to CbCR effective
2016, the results and the graph do not significantly differ when a stacked design is used.
In Table 6 Panel B, I explore whether the CbCR consolidated revenue threshold matters or
whether income shifting decreases consistently across MNC sizes. To do this, I remove the CbCR
indicator from the income shifting models. I split the MNC sample into four groups: two groups
not subject to private CbCR based on a median split of these MNCs’ consolidated revenue, and
two groups subject to private CbCR based on a median split of these MNCs’ consolidated revenue.
I report the results on the global sample, using the Huizinga and Laeven [2008] income shifting
model in columns 1 to 4 and the De Simone et al. [2017] income shifting model in columns 5 to
8.12
In Appendix C Panel B, I graph the C coefficients from columns 1 to 4 to illustrate the
decrease in income shifting for MNCs of different sizes. The black and green lines represent the
income shifting of the groups not subject to private CbCR, where black represents the smaller
MNCs and green represents the larger MNCs. The purple and blue lines represent the income
shifting of the groups subject to private CbCR, where purple represents the smaller MNCs and
blue represents the larger MNCs. The graphs show that the smaller MNCs not subject to private
CbCR were generally more aggressive income shifters in the pre-period than the other groups. All
MNCs, on average, decreased their income shifting alongside the implementation of private
12
In untabulated analysis, I perform this analysis using the EU sample only and all inferences remain unchanged.
- 24 -

CbCR; however, the MNCs not subject to private CbCR appear to be participating in the greatest
income shifting in the post-period.
The graph suggests that, on average, income shifting has decreased monotonically across
MNC sizes. This provides compelling evidence that the CbCR consolidated revenue threshold did
not matter, and it is not the CbCR to the tax authorities itself that resulted in the decrease in income
shifting. I assert that, consistent with practitioner opinion, it is likely the OECD’s BEPS framework
in general, along with the corresponding increased attention to MNCs’ tax planning behavior, that
has resulted in the decrease in income shifting.
5. Additional Analyses
In this section, I explore cross-sectional differences in MNCs’ headquarter-countries to
explore whether the OECD’s BEPS framework, in general, has contributed to MNCs’ decreases in
income shifting. In these analyses, I drop the CbCR indicator variable from the income shifting
models because my evidence strongly suggests that CbCR disclosure to the tax authorities is not
the main cause of the decrease in income shifting.13
5.1 High Regulation Countries
First, I explore whether MNCs headquartered in countries with stronger regulatory
environments exhibit a greater decrease in income shifting than MNCs headquartered in countries
with weaker regulatory environments. Countries with a stronger regulatory environment are more
likely to implement and enforce the BEPS action items, increasing the cost of MNCs’ income
shifting during this time period. This would be broadly consistent with Joshi’s [2020] evidence of
a decrease in income shifting among affiliates in countries with greater tax enforcement. I focus
13
I also drop the indicator variable for the years 2014-2015 as my results suggest that there is no statistically significant
difference in MNCs’ income shifting in 2014-2015 compared to 2010-2013.
- 25 -

on headquarter-countries instead of affiliate-countries given that income shifting decisions require
coordination across affiliates and are typically made at the parent level.
I proxy for high regulation by using a median-split of a country-level regulation factor
variable (HighRegn). Extant research has used various societal and institutional characteristics to
proxy for the strength of regulation in a country, such as the strength of auditing and accounting
standards, the absence of corruption, and government effectiveness. As many of these
characteristics are highly correlated, I undertake a factor analysis to draw out the common proxy
for regulation. Greater detail on the factor variable is provided in Appendix B.
I interact HighRegn with C and the year indicators. The results of this additional analysis
using the global sample are reported in Table 7 Panel A, with the Huizinga and Laeven [2008]
income shifting model in column 1 and the De Simone et al. [2017] income shifting model in
column 2. For brevity, I report only the coefficients of interest. Note that the sample decreases in
this analysis as the World Economic Forum’s Global Competitiveness Report does not cover every
country-year in the sample.
In Appendix D Panel A, I graph the yearly C coefficients in column 1 to illustrate the
decrease in income shifting over time for the group headquartered in countries with above-median
regulatory environment strength (represented by the green line) and the group headquartered in
countries with below-median regulatory environment strength (represented by the black line). The
graph shows that the group headquartered in countries with stronger regulatory environments
exhibit greater decreases in income shifting from 2016-2017 to 2020-2021, relative to the group
headquartered in countries with weaker regulatory environments (the coefficients on the
interactions of C, HighRegn, and the years of interest are positive and statistically significant at
the ten percent level or greater). The group headquartered in countries with stronger regulatory
- 26 -

environments has, on average, ceased income shifting by 2020-2021. This result is consistent with
my conjecture that MNCs headquartered in countries with a stronger regulatory environment
would exhibit greater decreases in income shifting as these countries would be more likely to
implement and enforce the various BEPS action items.
5.2 Countries with BEPS Action 4 or Action 5 “Issues”
I also explore whether MNCs headquartered in countries that the OECD has named as
having an “issue” with BEPS Action 4 Limitation on Interest Deductions or Action 5 Harmful Tax
Practices – Preferential Tax Regimes exhibit a smaller decrease in income shifting than MNCs
headquartered in countries that the OECD has not named. I refer to countries that have not
implemented an interest deduction limitation rule consistent with Action 4 or that have a harmful
preferential tax regime named in the OECD’s 2017 Action 5 progress report as having an “issue”
with the respective action item.
I posit that, consistent with the practitioner’s quote in the hypothesis development section,
MNCs headquartered in countries that have implemented an interest deduction limitation
following Action 4 and have no harmful preferential tax regimes listed in Action 5 would exhibit
a greater decrease in income shifting. Countries not implementing these action items, despite
agreeing to the BEPS framework, suggests they may not have been as concerned about income
shifting. Further, MNCs headquartered in countries with no interest deduction limitation or a
preferential tax regime have relatively more income shifting alternatives available to them and,
consequently, likely have a greater net benefit of continuing their income shifting behavior.
I focus on BEPS action items 4 and 5 as the OECD has produced publicly available
documents outlining country compliance with both. To my knowledge, the OECD has not
completed a comprehensive country survey of compliance with the other action items mentioned
- 27 -

by the practitioner, Action 2 Neutralising the Effects of Hybrid Mismatch Arrangements and
Actions 8-10 Transfer Pricing.
I review the OECD’s 2019 review of interest limitation rules and the 2017 progress report
on preferential tax regimes. I use these reports as they reflect “issues” in the early-to-middle stage
of my post-period. If I had chosen “issues” as of 2016, many more countries would fall into this
subset, given that the interest limitation rules were not released until 2015. Similarly, I do not
choose “issues” as of 2021, as very few countries still have outstanding “issues” as of this time.
Thus, these analyses would not be insightful.
To conduct my analysis, I create an indicator variable, Action4or5issue, which equals one
if one of the following is met: (1) the MNC’s headquarter-country does not have an interest
deduction limitation rule in place as of the OECD’s 2019 review, or (2) if the MNC’s headquarter-
country had a harmful preferential tax regime that was named in the OECD’s 2015 BEPS Action
5 report and still had not been amended or abolished by the time of the OECD’s 2017 progress
report. Action4or5issue equals zero if neither of the preceding conditions are met.
I interact Action4or5issue with C and the year indicators. The results of this additional
analysis using the global sample are reported in Table 7 Panel B, with the Huizinga and Laeven
[2008] income shifting model in column 1 and the De Simone et al. [2017] income shifting model
in column 2. For brevity, I report only the coefficients of interest.
In Appendix D Panel B, I graph the yearly C coefficients in column 1 to illustrate the
decrease in income shifting over time for the group headquartered in countries with no Action 4 or
Action 5 issue (represented by the blue line) and the group headquartered in countries with an
Action 4 or Action 5 issue (represented by the purple line). The graph shows that the group
headquartered in countries that comply with Action 4 and Action 5 has, on average, ceased income
- 28 -

shifting by 2020-2021. The group headquartered in countries that do not comply with Action 4 or
Action 5 has still exhibited decreases in income shifting in the post-period. However, there is
evidence to suggest that these decreases are not as large as the decreases of the group headquartered
in countries compliant with Action 4 and Action 5 (the coefficients on C×Action4or5issue×2016-
2017 and C×Action4or5issue×2020-2021 are negative and statistically significant at the ten
percent level, one-tailed). This result is broadly consistent with the practitioner’s quote and my
conjecture that the BEPS framework in general contributed to the reduction in income shifting;
however, a more detailed exploration of the various BEPS action items is necessary before a formal
conclusion can be made.
6. Conclusion
Extant research has failed to find robust evidence that private CbCR disclosure to the tax
authorities has significantly decreased MNCs’ tax-motivated international income shifting. In this
study, I revisit this question by replicating the results of prior research, expanding the sample to
include smaller public MNCs and private MNCs, lengthening the sample period to 2021, and
expanding to a global sample of MNCs. Using affiliate-level income shifting models, I find robust
evidence that MNCs shift significantly less income in the post-CbCR implementation period, and
these decreases are monotonic across MNC size. Altogether, I show that CbCR generally achieved
its overall objective of reducing income shifting; however, because I observe significant decreases
in income shifting among MNCs not subject to private CbCR, it is unlikely that the CbCR
disclosure to the tax authorities resulted in the decrease in income shifting. Additional tests on
other BEPS action items lend credibility to this conclusion.
To my knowledge, I provide the first global evidence to correct the misperception in the
literature that MNCs’ income shifting has not decreased after the implementation of private CbCR.
- 29 -

I show that CbCR is associated with a significant decrease in income shifting although this relation
is not causal. This evidence has important implications for the literature and policymakers
worldwide. Given that MNCs have already drastically decreased their income shifting, my
evidence questions whether public CbCR and Pillar II are necessary, particularly considering the
complexity and the compliance burden of Pillar II’s proposed global minimum tax rules.
Finally, given that estimates suggest that MNCs have, on average, decreased their income
shifting by half, I look forward to future research exploring the implications of this change on other
tax planning strategies. I also look forward to future research on the individual BEPS action items
that appear to contribute to the significant reduction in income shifting.
- 30 -

References
Baker, A. C., D. F. Larcker, and C. C. Y. Wang. 2022. ‘How much should we trust staggered
difference-in-differences estimates?’ Journal of Financial Economics, 144, 370-395.
Barrios, J. M. 2021. ‘Staggeringly problematic: A primer on staggered DiD for accounting
researchers.’ Working paper.
Bergin, T. 2012, October 15. Special report: How Starbucks avoids UK taxes. Reuters.
https://www.reuters.com/article/us-britain-starbucks-tax-idUSBRE89E0EX20121015
Cengiz, D., A. Dube, A. Lindner, and B. Zipperer. 2019. ‘The effect of minimum wages on low-
wage jobs.’ The Quarterly Journal of Economics, 134(3), 1405-1454.
De Simone, L., K. J. Klassen, and J. K. Seidman. 2017. ‘Unprofitable affiliates and income
shifting behavior.’ The Accounting Review, 92(3), 113-136.
De Simone, L., K. J. Klassen, and J. K. Seidman. 2022. ‘The effect of income-shifting
aggressiveness on corporate investment.’ Journal of Accounting and Economics, 74(1), 1-
20.
De Simone, L., and M. Olbert. 2022. ‘Real effects of private country-by-country disclosure.’ The
Accounting Review, 97(6), 201-232.
Goodman-Bacon, A. 2021. ‘Difference-in-differences with variation in treatment timing.’
Journal of Econometrics, 225(2), 254-277.
The Guardian. 2009, February 2. Firms’ secret tax avoidance schemes cost UK billions.
https://www.theguardian.com/business/2009/feb/02/tax-gap-avoidance
Hainmueller, J. 2012. ‘Entropy balancing for causal effects: A multivariate reweighting method
to produce balanced samples in observational studies.’ Political Analysis, 20, 25-46.
Hanlon, M. 2018. ‘Country-by-country reporting and the international allocation of taxing
rights.’ International Bureau of Fiscal Documentation (IBFD), forthcoming.
Hoopes, J. L., L. Robinson, and J. B. Slemrod. 2018. ‘Public tax-return disclosure.’ Journal of
Accounting and Economics, 66(1), 142-162.
Hoopes, J. L., L. Robinson, and J. B. Slemrod. 2023. ‘Corporate tax disclosure.’ Journal of the
American Taxation Association, forthcoming.
Huizinga, H., and L. Laeven. 2008. ‘International profit shifting within multinationals: A multi-
country perspective.’ Journal of Public Economics, 92(5-6), 1164-1182.
Joshi, P. 2020. ‘Does private country-by-country reporting deter tax avoidance and income
shifting? Evidence from BEPS Action Item 13.’ Journal of Accounting Research, 58(2),
333-381.
Joshi, P., K. Markle, and L. A. Robinson. 2023. ‘Private country-by-country reporting and the
misalignment between profits and economic activity.’ Working paper.
Kays, A. 2022. ‘Voluntary disclosure responses to mandated disclosure: Evidence from
Australian corporate tax transparency.’ The Accounting Review 97(4), 317-344.
- 31 -

Kobbi-Fakhfakh, S., and F. Driss. 2022. ‘Mandatory extraction payment disclosures and tax
haven use: Evidence from United Kingdom.’ Accounting and Management Information
Systems 21(3): 431-451.
Laudage, S. 2023. ‘The BEPS Project: Achievements and remaining challenges.’ IDOS Policy
Brief 22, German Institute of Development and Sustainability (IDOS), Bonn.
Müller, R., C. Spengel, and H. Vay. 2020. ‘On the determinants and effects of corporate tax
transparency: Review of an emerging literature.’ Working paper.
Nessa, M., A. V. Persson, J. Z. Song, E. Towery, and M. Vernon. 2022. ‘The effect of U.S.
country-by-country reporting on U.S. multinationals’ tax-motivated income shifting and
real activities.’ Working paper.
OECD. 2013. Addressing base erosion and profit shifting. OECD Publishing.
OECD. 2015. OECD/G20 base erosion and profit shifting project Action 13: Country-by-country
reporting implementation package. https://www.oecd.org/ctp/transfer-pricing/beps-action-
13-country-by-country-reporting-implementation-package.pdf
OECD. 2024a. Base erosion and profit shifting. https://www.oecd.org/tax/beps/
OECD. 2024b. Country-specific information on country-by-country reporting implementation.
https://www.oecd.org/tax/automatic-exchange/country-specific-information-on-country-by-
country-reporting-implementation.htm
Sun, L., and S. Abraham. 2021. ‘Estimating dynamic treatment effects in event studies with
heterogeneous treatment effects.’ Journal of Econometrics, 225(2), 175-199.
Wayne, L., K. Carr, M. Walker Guevara, M. Cabra, M. Hudson, M. Williams, E. Perrin, E. Díaz-
Struck, D. Reuter, F. Zalac, H. Cashore, L. Bové, K. Clerix, J. Stein, T. Plattner, M.
Stäuble, M. Knus-Galán, M. Caruana Galizia, R. Carvajal, C. Lütgert, and N. Chenoweth.
2014, November 5. Leaked documents expose global companies’ secret tax deals in
Luxembourg. International Consortium of Investigative Journalists.
https://www.icij.org/investigations/luxembourg-leaks/leaked-documents-expose-global-
companies-secret-tax-deals-luxembourg/
- 32 -

Appendix A – Private CbCR Regulations for Countries in the Sample
Ultimate Parent Entity Local Filing Consolidated Group
Country Filing Effective Date Effective Date Revenue Threshold
AE (UAE) January 1, 2019 N/A AED 3.15 billion
AT (Austria) January 1, 2016 January 1, 2017 EUR 750 million
AU (Australia) January 1, 2016 January 1, 2016 AUD 1 billion
BA (Bosnia) January 1, 2018 N/A EUR 750 million
BE (Belgium) January 1, 2016 January 1, 2016 EUR 750 million
BG (Bulgaria) January 1, 2016 January 1, 2017 EUR 750 million
BM (Bermuda) January 1, 2016 N/A EUR 750 million
BR (Brazil) January 1, 2016 Suspended BRL 2.26 billion
BS (Bahamas) January 1, 2019 N/A USD 850 million
CA (Canada) January 1, 2016 January 1, 2016 EUR 750 million
CH (Switzerland) January 1, 2018 January 1, 2018 CHF 900 million
CI (Côte d’Ivoire) January 1, 2018 N/A XOF 491,967,750,000
CL (Chile) January 1, 2016 January 1, 2016 EUR 750 million
CN (China) January 1, 2016 Suspended RMB 5.5 billion
CO (Colombia) January 1, 2016 January 1, 2016 UVT 81 million
CZ (Czechia) January 1, 2016 January 1, 2017 EUR 750 million
DE (Germany) January 1, 2016 January 1, 2017 EUR 750 million
DK (Denmark) January 1, 2016 January 1, 2017 DKK 5.6 billion
EE (Estonia) January 1, 2016 January 1, 2017 EUR 750 million
EG (Egypt) FYE December 31, 2018 N/A EGP 3 billion
ES (Spain) January 1, 2016 January 1, 2016 EUR 750 million
FI (Finland) January 1, 2016 January 1, 2016 EUR 750 million
FR (France) January 1, 2016 January 1, 2016 EUR 750 million
GB (UK) January 1, 2016 January 1, 2016 EUR 750 million
GI (Gibraltar) January 1, 2016 January 1, 2017 EUR 750 million
GR (Greece) January 1, 2016 January 1, 2016 EUR 750 million
HK (Hong Kong) January 1, 2018 January 1, 2018 HKD 6.8 billion
HR (Croatia) January 1, 2016 January 1, 2017 EUR 750 million
HU (Hungary) January 1, 2016 January 1, 2017 EUR 750 million
ID (Indonesia) January 1, 2016 January 1, 2016 IDR 11 trillion
IE (Ireland) January 1, 2016 January 1, 2016 EUR 750 million
IN (India) April 1, 2016 January 1, 2016 INR 5,500 crore
IS (Iceland) January 1, 2017 January 1, 2017 ISK 100 billion
IT (Italy) January 1, 2016 January 1, 2016 EUR 750 million
JP (Japan) April 1, 2016 April 1, 2017 JPY 100 billion
KR (Korea) January 1, 2016 January 1, 2016 KRW 1 trillion
KY (Cayman Islands) January 1, 2016 N/A USD 850 million
LK (Sri Lanka) April 1, 2018 April 1, 2018 EUR 750 million
LT (Lithuania) January 1, 2016 January 1, 2016 EUR 750 million
LU (Luxembourg) January 1, 2016 January 1, 2016 EUR 750 million
LV (Latvia) January 1, 2016 January 1, 2016 EUR 750 million
MA (Morocco) January 1, 2021 January 1, 2021 MAD 8.1225 billion
MT (Malta) January 1, 2016 January 1, 2017 EUR 750 million
MX (Mexico) January 1, 2016 January 1, 2016 MXN 12 billion
MY (Malaysia) January 1, 2017 N/A MYR 3 billion
- 33 -

NL (Netherlands) January 1, 2016 January 1, 2016 EUR 750 million
NO (Norway) January 1, 2016 January 1, 2017 NOK 6.5 billion
NZ (New Zealand) January 1, 2016 N/A EUR 750 million
OM (Oman) January 1, 2020 January 1, 2020 OMR 300 million
PA (Panama) January 1, 2018 N/A EUR 750 million
PK (Pakistan) January 1, 2016 January 1, 2017 EUR 750 million
PL (Poland) January 1, 2016 January 1, 2017 EUR 750 million
PT (Portugal) January 1, 2016 January 1, 2017 EUR 750 million
QA (Qatar) January 1, 2018 Suspended QAR 3 billion
RO (Romania) January 1, 2016 January 1, 2017 EUR 750 million
RS (Serbia) January 1, 2020 N/A EUR 750 million
SA (Saudi Arabia) January 1, 2018 January 1, 2018 SAR 3.2 billion
SE (Sweden) January 1, 2016 January 1, 2016 SEK 7 billion
SG (Singapore) January 1, 2017 N/A SGD 1.125 billion
SI (Slovenia) January 1, 2016 January 1, 2017 EUR 750 million
SK (Slovakia) January 1, 2016 January 1, 2017 EUR 750 million
TH (Thailand) January 1, 2021 suspended THB 28 billion
TN (Tunisia) January 1, 2020 January 1, 2020 TND 1.6388 billion
TR (Türkiye) January 1, 2019 January 1, 2019 EUR 750 million
UA (Ukraine) FYE December 31, 2021 FYE December 31, 2021 EUR 750 million
US (USA) June 30, 2016 N/A USD 850 million
VG (British Virgin Islands) January 1, 2018 January 1, 2018 EUR 750 million
VN (Vietnam) May 1, 2017 May 1, 2017 VND 18 trillion
ZA (South Africa) January 1, 2016 January 1, 2017 ZAR 10 billion
Source: OECD [2024b]
- 34 -

Appendix B – Variable Definitions
Variable Definition Source
Action4or5issue indicator which equals one if one of the following is met: OECD
(1) the MNC’s headquarter-country does not have an interest
deduction limitation rule in place as of the OECD’s 2019 review, or
(2) if the MNC’s headquarter-country had a harmful preferential tax
regime that was named in the OECD’s 2015 BEPS Action 5 report
and still had not been amended or abolished by the time of the
OECD’s 2017 progress report; zero otherwise
lnAge natural logarithm of year t minus the year the firm incorporated Orbis
C for the Huizinga and Laeven [2008] income shifting model: Huizinga and
measure of the affiliate’s tax incentive to shift income, which equals Laeven [2008],
the revenue-weighted average differential statutory tax rate between De Simone et
the affiliate and all other related affiliates in the same group-year al. [2017],
for the De Simone et al. [2017] income shifting model: Orbis, Tax
measure of the affiliate’s tax incentive to shift income, which equals Foundation
the capital-weighted average differential statutory tax rate between
the affiliate and all other related affiliates in the same group-year
CbCR indicator which equals one where the MNC is subject to private Appendix A
CbCR based on either an ultimate parent entity filing obligation or a
local filing obligation (i.e., prior year consolidated revenue greater
than or equal to the consolidated group revenue threshold detailed in
Appendix A); zero otherwise
lnCompensation natural logarithm of compensation expense Orbis
lnEBIT natural logarithm of earnings before interest and tax Orbis
lnFixedAssets natural logarithm of fixed assets Orbis
ΔGDP change in gross domestic product for country c from year t-1 to t, World Bank
scaled by lagged gross domestic product
lnGDP natural logarithm of country c’s gross domestic product World Bank
lnGDPperCapita natural logarithm of country c’s gross domestic product per capita World Bank
HighRegn median-split of a country-level regulation factor variable see below
IndustryROA country-year-industry median ROA for all affiliates and standalone Orbis
firms in Orbis
ΔMktSize change in country-year-industry sum of all affiliates and standalone Orbis
firm sales in Orbis from year t-1 to t, scaled by 1,000,000
Post indicator which equals one for the years when private CbCR is Appendix A
required (see Appendix A for effective dates); zero otherwise
lnPTI natural logarithm of pre-tax income Orbis
Revenue MNC’s consolidated revenue Orbis
Size natural logarithm of MNC’s consolidated total assets Orbis
- 35 -

B.1 Country-level Regulation Factor Variable
Extant research has used various societal and institutional characteristics to proxy for the strength of
regulation in a country. As many of these characteristics are highly correlated, I undertake a factor analysis
to draw out the common proxy for regulation.
Variables Used:
• Strength of auditing and accounting standards (aastd) captures the strength of the auditing and
accounting standards. Data are from the World Economic Forum Global Competitiveness Report.
• Rule of Law (rulelaw) captures the extent to which agents have confidence in and abide by the rules
of the society. Data are from the World Bank.
• Absence of corruption (abscorrupt) is an index of the controls over corruption. Data are from the
World Bank.
• Legal system (legalsys) is the efficiency of legal system in settling disputes. Data are from the World
Economic Forum Global Competitiveness Report.
• Regulatory quality (regqual) is the ability of the government to formulate and implement sound
policies and regulations that permit and promote private sector development. Data are from the World
Bank.
• Government effectiveness (govteffect) is an overall index of the quality of public services, the quality
of the civil service and the degree of its independence from political pressures, the quality of policy
formulation and implementation, and the credibility of the government’s commitment to such policies.
Data are from the World Bank.
Factor Analysis:
Table B.1 Panel A reports the eigenvalues for the first three factors. I determine that there is one principal
factor for these six variables.
Table B.1: Factor Analysis

Panel A. Eigenvalues for Unrotated Principal Factor Analysis
Factor Eigenvalue Difference Proportion Cumulative
Factor1 4.951 4.561 0.940 0.940
Factor2 0.391 0.348 0.074 1.014
Factor3 0.042 0.054 0.008 1.022
Panel B. Factor Loadings

Variable Factor1
aastd 0.829
rulelaw 0.968
abscorrupt 0.972
legalsys 0.762
regqual 0.938
govteffect 0.960
Notes: There are 712 observations for the six variables used in this analysis. Panel A presents the eigenvalues and the proportion
of total variation explained by each factor, including incremental eigenvalue and variations. Panel B shows the factor loadings.
- 36 -

Appendix C – Tax-Motivated Income Shifting of MNCs Over Time
Panel A. By Year
Panel B. By Size
- 37 -

Appendix D – Cross-Sectional Differences in the Tax-Motivated Income Shifting of MNCs
Panel A. High Regulation Countries
Panel B. Countries with BEPS Action 4 or Action 5 “Issues”
- 38 -

Table 1: Sample Selection and Composition
Panel A. Sample Selection
Unique MNCs
EU Sample Global Sample
All global ultimate owner (GUO) corporations in Orbis with at least one affiliate
located in a country other than the GUO’s headquarter-country 295,824 295,824
Less: GUOs with missing NACE codes (214,657) (214,657)
Less: GUOs in financial industries (18,136) (18,136)
Less: GUOs in extractive industries (926) (926)
Less: GUOs headquartered outside the EU (32,354) -
Less: GUOs headquartered in countries with no private CbCR regulation - (3,775)
Less: GUOs missing revenue or total assets (22,181) (39,370)
Sample of MNCs 7,570 18,960
Affiliate-Years
All affiliate-years in Orbis belonging to the above MNCs from 2010-2021 1,064,995 2,252,506
Less: affiliates with missing NACE codes (74,539) (163,082)
Less: affiliates missing required data (e.g., EBIT, revenue, assets,
compensation expense) (596,776) (1,407,487)
Less: affiliates with EBIT < 0 (104,449) (171,613)
Sample of affiliate-years 289,231 510,324
Panel B. Affiliate-Years by Year

Affiliates of MNCs Affiliates of Affiliates of Affiliates of
Subject to Private MNCs Never MNCs Subject to MNCs Never
CbCR in the Subject to Private CbCR in Subject to
Year Sample Period Private CbCR Total the Sample Period Private CbCR Total
2010 802 104 906 1,676 178 1,854
2011 2,355 322 2,677 4,758 527 5,285
2012 13,614 6,209 19,823 24,773 7,656 32,429
2013 15,720 8,067 23,787 31,562 10,345 41,907
2014 16,627 9,025 25,652 33,884 11,674 45,558
2015 17,753 9,605 27,358 35,963 12,521 48,484
2016 18,530 10,366 28,896 37,949 13,716 51,665
2017 19,502 11,556 31,058 40,258 15,256 55,514
2018 19,886 12,552 32,438 41,287 16,440 57,727
2019 20,175 13,198 33,373 41,689 17,359 59,048
2020 18,753 12,732 31,485 38,782 16,735 55,517
2021 18,807 12,971 31,778 38,303 17,033 55,336
Total 182,524 106,707 289,231 370,884 139,440 510,324
- 39 -

Panel C. Affiliate-Years by Country of Affiliate
Affiliates of Affiliates of Affiliates of Affiliates of
MNCs Subject to MNCs Never MNCs Subject to MNCs Never
Private CbCR in Subject to Private CbCR in Subject to
Country the Sample Period Private CbCR Total the Sample Period Private CbCR Total
AL (Albania) 17 7 24 18 7 25
AR (Argentina) 7 - 7 8 - 8
AT (Austria) 4,492 1,327 5,819 6,841 1,389 8,230
AU (Australia) 2,919 205 3,124 10,150 1,367 11,517
BA (Bosnia) 266 106 372 415 127 542
BD (Bangladesh) 1 - 1 1 - 1
BE (Belgium) 10,125 6,367 16,492 17,694 6,913 24,607
BG (Bulgaria) 1,882 929 2,811 3,160 1,076 4,236
BM (Bermuda) - - - 1 6 7
BR (Brazil) 50 1 51 67 1 68
CH (Switzerland) 9 - 9 72 25 97
CI (Côte d’Ivoire) - - - 12 - 12
CL (Chile) 19 2 21 21 3 24
CO (Colombia) 2 - 2 2 - 2
CZ (Czechia) 5,298 2,286 7,584 9,674 3,366 13,040
DE (Germany) 14,675 6,063 20,738 28,315 7,448 35,763
DK (Denmark) 2,081 1,957 4,038 3,789 2,438 6,227
DZ (Algeria) 60 7 67 83 8 91
EC (Ecuador) 3 - 3 6 - 6
EE (Estonia) 871 917 1,788 1,313 1,053 2,366
EG (Egypt) - - - 1 - 1
ES (Spain) 18,382 12,524 30,906 28,201 13,539 41,740
FI (Finland) 4,316 6,516 10,832 7,036 6,900 13,936
FR (France) 33,133 9,688 42,821 49,838 11,178 61,016
GB (UK) 19,154 10,085 29,239 42,808 13,028 55,836
GE (Georgia) 3 - 3 6 - 6
GH (Ghana) 2 - 2 2 - 2
GY (Guyana) 4 - 4 4 - 4
HK (Hong Kong) 2 - 2 5 - 5
HR (Croatia) 1,339 705 2,044 2,178 761 2,939
HU (Hungary) 3,027 1,169 4,196 5,583 1,344 6,927
IE (Ireland) 2,267 790 3,057 5,770 1,035 6,805
IN (India) 2,878 673 3,551 13,151 4,492 17,643
IS (Iceland) 57 27 84 213 197 410
IT (Italy) 12,999 14,001 27,000 23,864 15,329 39,193
JP (Japan) 282 74 356 27,431 4,526 31,957
KR (Korea) 1,935 191 2,126 10,835 1,838 12,673
LK (Sri Lanka) 18 - 18 54 95 149
LU (Luxembourg) 804 205 1,009 1,550 251 1,801
LV (Latvia) 114 20 134 129 27 156
- 40 -

MA (Morocco) 820 186 1,006 1,262 238 1,500
ME (Montenegro) 62 26 88 126 42 168
MK (N. Macedonia) 81 47 128 115 47 162
MT (Malta) 99 29 128 112 37 149
MU (Mauritius) - - - 1 - 1
MY (Malaysia) 6 - 6 76 17 93
NL (Netherlands) 1,699 588 2,287 3,744 809 4,553
NO (Norway) 4,543 3,892 8,435 10,830 9,808 20,638
NZ (New Zealand) 660 38 698 2,163 281 2,444
PA (Panama) 8 - 8 8 - 8
PE (Peru) 1 - 1 5 - 5
PK (Pakistan) 79 - 79 176 40 216
PL (Poland) 7,572 4,517 12,089 13,126 5,284 18,410
PT (Portugal) 4,569 3,040 7,609 7,407 3,181 10,588
PY (Paraguay) 1 - 1 1 - 1
RO (Romania) 3,885 2,295 6,180 6,607 2,619 9,226
RS (Serbia) 1,079 473 1,552 1,786 712 2,498
RU (Russia) - - - 1 - 1
SE (Sweden) 9,166 12,696 21,862 14,623 14,161 28,784
SG (Singapore) - - - 4 - 4
SI (Slovenia) 923 326 1,249 1,540 357 1,897
SK (Slovakia) 2,927 1,423 4,350 4,913 1,687 6,600
SV (El Salvador) - - - 7 - 7
TH (Thailand) 34 - 34 431 43 474
TN (Tunisia) - - - - 3 3
TR (Türkiye) 7 - 7 7 - 7
TZ (Tanzania) 2 - 2 5 - 5
UA (Ukraine) 785 289 1,074 1,466 307 1,773
US (USA) - - - 5 - 5
UY (Uruguay) 14 - 14 22 - 22
ZA (South Africa) 9 - 9 9 - 9
ZM (Zambia) - - - 3 - 3
ZW (Zimbabwe) - - - 2 - 2
Total 182,524 106,707 289,231 370,884 139,440 510,324
- 41 -

Panel D. Affiliate-Years by Industry of Affiliate
Affiliates of Affiliates of Affiliates of Affiliates of
MNCs Subject to MNCs Never MNCs Subject to MNCs Never
Private CbCR in Subject to Private CbCR in Subject to
Industry the Sample Period Private CbCR Total the Sample Period Private CbCR Total
Agriculture, Forestry, Fishing 908 1,285 2,193 1,619 1,481 3,100
Mining, Quarrying 1,527 304 1,831 2,425 493 2,918
Manufacturing 41,227 30,159 71,386 93,925 39,920 133,845
Electricity, Gas, Steam, Air 4,030 776 4,806 5,221 1,290 6,511
Water, Sewerage, Waste 4,242 671 4,913 4,989 815 5,804
Construction 14,711 4,107 18,818 21,993 5,403 27,396
Trade, Vehicle Repair 47,705 25,982 73,687 103,536 33,747 137,283
Transportation, Storage 8,302 5,060 13,362 14,234 6,568 20,802
Accommodation, Food 1,662 1,752 3,414 4,169 2,070 6,239
Information, Communication 15,169 12,245 27,414 34,765 17,076 51,841
Financial, Insurance 2,801 1,168 3,969 6,057 1,480 7,537
Real Estate 2,892 2,874 5,766 5,243 3,532 8,775
Professional, Scientific, Tech 17,145 10,302 27,447 35,445 13,238 48,683
Administrative, Support 10,128 5,943 16,071 21,709 7,570 29,279
Public Admin, Defence 191 22 213 324 33 357
Education 600 603 1,203 1,259 751 2,010
Human Health, Social Work 7,145 2,359 9,504 9,628 2,578 12,206
Arts, Entertainment, Rec 877 511 1,388 1,690 683 2,373
Other Service 1,235 567 1,802 2,616 692 3,308
Other Activities 27 17 44 37 20 57
Total 182,524 106,707 289,231 370,884 139,440 510,324
Notes: This table provides sample selection in Panel A, and sample composition, by year in Panel B, by country in Panel C,
and by NACE industry in Panel D. Observations are categorized as affiliates of MNCs that are subject to private CbCR or not.
- 42 -

Table 2: Descriptive Statistics
Panel A. Summary Statistics
Affiliates of Affiliates of
MNCs Subject to MNCs Never
Private CbCR in Subject to
the Sample Private
Global Sample Period CbCR
Variable mean sd p25 p50 p75 mean mean difference
MNC-level variables:
Revenue
(billions) 14.647 30.113 0.517 3.049 15.274 20.056 0.258 19.798***
Size 21.820 2.412 20.095 22.029 23.628 22.940 18.842 4.097***
# of affiliates 101 218 12 35 94 130 23 107***
CbCR 0.727 0.446 0.000 1.000 1.000 - - -
Affiliate-level variables:
lnPTI 13.865 2.176 12.550 13.904 15.228 14.153 13.097 1.056***
lnEBIT 13.880 2.089 12.591 13.908 15.199 14.157 13.143 1.014***
C HL
-0.006 0.074 -0.046 0.000 0.029 -0.007 -0.003 -0.004***
C DKS
-0.005 0.057 -0.032 0.000 0.024 -0.006 -0.003 -0.003***
Post 0.649 0.477 0.000 1.000 1.000 0.635 0.687 -0.052***
lnFixedAssets 13.865 3.986 12.195 14.400 16.317 14.147 13.116 1.031***
lnCompensation 14.826 1.886 13.697 14.856 16.022 15.078 14.154 0.925***
lnAge 2.849 0.895 2.398 2.996 3.401 2.891 2.738 0.153***
Country- and Industry-level variables:
lnGDP 27.636 1.203 26.810 28.011 28.620 27.700 27.467 0.233***
lnGDPperCapita 10.312 0.749 10.185 10.509 10.704 10.298 10.349 -0.051***
ΔGDP 0.017 0.037 0.008 0.019 0.030 0.017 0.016 0.001***
ΔMktSize 6.183 41.424 -0.786 0.202 3.487 6.658 4.971 1.687***
IndustryROA 0.044 0.040 0.024 0.039 0.058 0.044 0.043 0.001***
- 43 -

Panel B. Pearson Correlation Coefficients
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
(1) lnPTI 1
(2) lnEBIT 0.950 1
(3) CHL 0.039 0.034 1
(4) CDKS 0.048 0.043 0.942 1
(5) CbCR 0.216 0.216 -0.025 -0.022 1
(6) Post 0.011 0.008 0.008 -0.006 -0.049 1
(7) lnFixedAssets 0.591 0.597 0.068 0.080 0.115 -0.015 1
(8) lnCompensation 0.655 0.672 0.077 0.081 0.218 -0.007 0.533 1
(9) lnAge 0.297 0.286 0.074 0.082 0.076 0.054 0.282 0.306 1
(10) lnGDP 0.188 0.182 0.376 0.372 0.086 0.009 0.165 0.205 0.201 1
(11) lnGDPperCapita 0.040 0.019 0.077 0.095 -0.031 0.017 -0.060 0.104 0.107 0.192 1
(12) ΔGDP 0.016 0.019 -0.098 -0.112 0.010 0.018 -0.004 -0.023 -0.044 -0.105 -0.142 1
(13) ΔMktSize 0.008 0.009 0.035 0.041 0.018 -0.102 0.003 0.007 0.008 0.064 0.009 -0.028 1
(14) IndustryROA 0.015 0.019 -0.051 -0.055 0.008 0.044 -0.050 0.086 0.012 0.062 0.077 0.066 -0.012 1
Notes: This table provides summary statistics in Panel A for variables to estimate the affiliate-level tax-motivated income shifting models. Panel A also compares
mean values for observations categorized as affiliates of MNCs that are subject to private CbCR or not. Two-sided t-tests are used to test the difference in means.
*, **, and *** indicate statistical significance at the ten, five, and one percent levels, respectively. Panel B provides Pearson correlations for variables to estimate
the affiliate-level tax-motivated income shifting models. Values that are in bold are statistically significant at the five percent level or higher (two-tailed). Variables
are defined in Appendix B.
- 44 -

Table 3: EU MNCs from 2010 – 2018
(1) (2) (3) (4) (5)

Joshi [2020] Replication of Replication of Huizinga & De Simone et
Table 8 Panel A Joshi [2020] Joshi [2020] with Laeven [2008] al. [2017]
column 2 no ISIN restriction model model
Variables lnPTI lnPTI lnPTI lnEBIT lnEBIT
C -0.888** -1.731*** -1.263*** -1.328*** -1.164***
(-2.099) (-2.840) (-4.208) (-5.767) (-3.128)
C×CbCR 0.513 0.570 0.072 0.297 0.477
(1.145) (0.943) (0.223) (1.043) (1.076)
C×Post 0.349 0.179 0.752*** 0.717*** 0.946***
(0.709) (0.439) (3.308) (3.632) (3.000)
C×CbCR×Post -0.140 -0.086 -0.553** -0.465** -0.803**
(-0.273) (-0.195) (-2.068) (-2.001) (-2.155)
lnFixedAssets 0.186*** 0.218*** 0.188*** 0.172*** 0.184***
(37.200) (29.522) (32.170) (30.575) (29.844)
lnCompensation 0.547*** 0.483*** 0.530*** 0.533*** 0.507***
(60.778) (41.025) (53.699) (48.718) (45.641)
lnGDP 0.058*** 0.091*** 0.069***
(5.800) (4.262) (4.583)
lnGDPperCapita 0.058***
(2.940)
ΔGDP 4.158***
(7.138)
lnAge 0.118***
(9.658)
ΔMktSize 0.000
(1.535)
IndustryROA -0.340
(-0.971)
CbCR -0.024 0.280*** 0.389*** 0.309*** 0.349***
(-1.143) (3.620) (8.888) (9.618) (8.472)
CbCR×Post -0.078*** -0.028 -0.041* -0.005 -0.017
(-4.588) (-0.837) (-1.953) (-0.253) (-0.823)
Fixed Effects Year Year Year Year, Industry Year
Observations 123,920 134,070 192,595 192,595 183,860
R2 - 0.511 0.520 0.558 0.522
Adjusted R2 0.586 0.511 0.520 0.558 0.522
Notes: This table provides regression results for EU MNCs’ tax-motivated income shifting around the implementation
of private CbCR, where the dependent variable is lnPTI in columns 1-3 and lnEBIT in columns 4-5. The regressions
are estimated using OLS regression. Standard errors have been adjusted for clustering at the MNC level and the related
t-statistics are reported in parentheses. *, **, and *** indicate two-tailed statistical significance at the ten, five, and
one percent levels, respectively. Variables are defined in Appendix B.
- 45 -

Table 4: Impact of the ISIN Sample Restriction
Panel A. Summary Statistics
Affiliates of MNCs with Affiliates of MNCs without
an ISIN listed in Orbis an ISIN listed in Orbis
Variable mean mean difference
MNC-level variables:
Revenue (billions) 19.129 1.439 17.691***
Size 22.365 19.411 2.954***
# of affiliates 193 29 164***
CbCR 0.828 0.291 0.537***
Affiliate-level variables:
lnPTI 13.912 13.153 0.759***
lnEBIT 13.930 13.211 0.719***
CHL -0.006 -0.006 0.001
Post 0.466 0.506 -0.039***
lnFixedAssets 13.823 13.483 0.340***
lnCompensation 14.899 14.318 0.581***
Country-level variables:
lnGDP 27.501 27.458 0.042***
lnGDPperCapita 10.334 10.321 0.014***
Panel B. Sample Composition by MNC Headquarter Country

Replication of Joshi Replication of Joshi [2020] Difference in
Country [2020] with no ISIN restriction Sample Composition
DE (Germany) 14.8% 17.9% -3.1%
ES (Spain) 4.0% 6.8% -2.8%
FR (France) 24.8% 19.5% 5.3%
GB (UK) 17.6% 14.0% 3.6%
IT (Italy) 2.4% 7.3% -4.9%
SE (Sweden) 13.7% 12.4% 1.3%
Other countries 22.7% 22.1% 0.6%
- 46 -

Panel C. Overrepresentation of MNCs Headquartered in France and the UK
(1) (2)
Huizinga & Laeven [2008] model De Simone et al. [2017] model
Variables lnEBIT lnEBIT
C -1.363*** -0.713
(-3.417) (-1.350)
C×CbCR 0.219 0.089
(0.466) (0.135)
C×Post 0.376 -0.244
(1.061) (-0.411)
C×CbCR×Post -0.004 0.662
(-0.010) (1.008)
lnFixedAssets 0.150*** 0.166***
(15.606) (16.676)
lnCompensation 0.529*** 0.501***
(24.707) (25.181)
lnGDPperCapita 0.095***
(3.535)
ΔGDP 5.436***
(7.130)
lnAge 0.147***
(6.755)
ΔMktSize 0.000
(1.268)
IndustryROA -0.661
(-0.964)
CbCR 0.341*** 0.398***
(5.753) (4.743)
CbCR×Post -0.060 -0.149***
(-1.394) (-2.809)
Fixed Effects Year, Industry Year
Observations 93,954 81,035
R2 0.534 0.505
Adjusted R2 0.534 0.505
- 47 -

Panel D. 2x2 Matrix of France and UK MNCs’ C Coefficients from Column 1
Notes: This table provides summary statistics in Panel A and sample composition by headquarter country in Panel B
for variables to estimate the affiliate-level tax-motivated income shifting models based on whether the MNC has an
ISIN listed in Orbis or not. Panel A also compares mean values for observations categorized as affiliates of MNCs
that have an ISIN listed in Orbis or not. Panel C provides regression results for France and UK tax-motivated income
shifting around the implementation of private CbCR, where the dependent variable is lnEBIT. The regressions are
estimated using OLS regression. Standard errors have been adjusted for clustering at the MNC level and the related t-
statistics are reported in parentheses. *, **, and *** indicate two-tailed statistical significance at the ten, five, and one
percent levels, respectively. Panel D provides a 2x2 matrix of the implied coefficients on C from Panel C column 1.
Variables are defined in Appendix B.
- 48 -

Table 5: Test of Hypothesis 1
Panel A. Regression Results

EU Sample from 2010-2021 Global Sample from 2010-2021
(1) (2) (3) (4)
Huizinga & Laeven De Simone et al. Huizinga & Laeven De Simone et al.
[2008] model [2017] model [2008] model [2017] model
Variables lnEBIT lnEBIT lnEBIT lnEBIT
C -1.344*** -1.207*** -1.155*** -1.328***
(-5.731) (-3.286) (-6.042) (-4.395)
C×CbCR 0.308 0.503 0.369* 0.712**
(1.071) (1.137) (1.710) (2.091)
C×Post 0.661*** 0.678** 0.484*** 0.645**
(3.073) (1.988) (2.675) (2.267)
C×CbCR×Post -0.313 -0.398 -0.033 -0.061
(-1.259) (-1.021) (-0.164) (-0.197)
lnFixedAssets 0.168*** 0.173*** 0.168*** 0.174***
(32.034) (30.388) (45.204) (43.617)
lnCompensation 0.525*** 0.503*** 0.537*** 0.511***
(49.848) (47.122) (72.624) (67.027)
lnGDPperCapita 0.071*** -0.001
(3.824) (-0.108)
ΔGDP 2.979*** 4.070***
(7.691) (15.266)
lnAge 0.135*** 0.126***
(11.962) (16.077)
ΔMktSize 0.000** 0.000
(1.989) (1.188)
IndustryROA -0.562* -0.476**
(-1.796) (-2.076)
CbCR 0.324*** 0.361*** 0.301*** 0.363***
(10.276) (8.691) (13.023) (13.125)
CbCR×Post -0.027 -0.036* -0.017 -0.027
(-1.347) (-1.658) (-1.070) (-1.443)
Fixed Effects Year, Industry Year Year, Industry Year
Observations 289,231 278,438 510,324 481,279
R2 0.550 0.512 0.548 0.519
Adjusted R2 0.549 0.512 0.548 0.519
- 49 -

Panel B. 2x2 Matrix of Global MNCs’ C Coefficients from Column 3
Notes: Panel A provides regression results for EU and global MNCs’ tax-motivated income shifting around the
implementation of private CbCR, where the dependent variable is lnEBIT. The regressions are estimated using OLS
regression. Standard errors have been adjusted for clustering at the MNC level and the related t-statistics are reported
in parentheses. *, **, and *** indicate two-tailed statistical significance at the ten, five, and one percent levels,
respectively. Panel B provides a 2x2 matrix of the implied coefficients on C from Panel A column 3. Variables are
defined in Appendix B.
- 50 -

Table 6: Finer Gradations on Post and CbCR
Panel A. Post
(1) (2)
C -1.206*** -1.588***
(-5.351) (-4.494)
C×CbCR 0.448* 0.917**
(1.754) (2.307)
C×2014-2015 0.055 0.447*
(0.310) (1.749)
C×2016-2017 0.498** 1.095***
(2.291) (3.252)
C×2018-2019 0.718*** 1.037***
(2.864) (2.719)
C×2020-2021 0.373 0.513
(1.300) (1.167)
C×CbCR×2014-2015 -0.137 -0.374
(-0.699) (-1.293)
C×CbCR×2016-2017 -0.263 -0.748**
(-1.090) (-1.983)
C×CbCR×2018-2019 -0.336 -0.424
(-1.215) (-1.016)
C×CbCR×2020-2021 0.413 0.534
(1.301) (1.090)
CbCR 0.302*** 0.364***
(13.041) (13.132)
CbCR×Post -0.017 -0.026
(-1.066) (-1.421)
Controls Y Y
R2 0.548 0.519
Adjusted R2 0.548 0.519
- 51 -

Panel B. CbCR
(1) (2) (3) (4) (5) (6) (7) (8)
Huizinga and Laeven [2008] model De Simone et al. [2017] model
CbCR=0 CbCR=1 CbCR=0 CbCR=1
Smaller Larger Smaller Larger Smaller Larger Smaller Larger
Variables lnEBIT lnEBIT lnEBIT lnEBIT lnEBIT lnEBIT lnEBIT lnEBIT
C -1.101*** -0.827*** -0.886*** -0.724*** -1.437*** -1.030*** -0.747*** -0.484*
(-3.648) (-3.799) (-7.563) (-4.989) (-3.131) (-3.369) (-4.039) (-1.782)
C×Post 0.587* 0.322 0.422*** 0.484*** 0.734 0.599* 0.521*** 0.452**
(1.914) (1.542) (3.867) (4.032) (1.627) (1.932) (3.142) (2.058)
Controls Y Y Y Y Y Y Y Y
Fixed Effects Year, Year, Year, Year, Year Year Year Year
Industry Industry Industry Industry
Observations 69,693 69,746 185,448 185,436 65,175 64,740 166,117 166,716
R2 0.462 0.506 0.525 0.546 0.456 0.498 0.515 0.535
Adjusted R2 0.461 0.506 0.524 0.546 0.456 0.498 0.515 0.535
Notes: This table provides regression results for global MNCs’ tax-motivated income shifting around the implementation of
private CbCR, where the dependent variable is lnEBIT. In Panel A (B), the Post (CbCR) indicator variable is split into finer
gradations. The regressions are estimated using OLS regression. Standard errors have been adjusted for clustering at the MNC
level and the related t-statistics are reported in parentheses. *, **, and *** indicate two-tailed statistical significance at the ten,
five, and one percent levels, respectively. Variables are defined in Appendix B.
- 52 -

Table 7: Additional Analyses
Panel A. High Regulation Countries
(1) (2)
C -0.939*** -0.731***
(-8.474) (-4.248)
C×HighRegn -0.061 -0.198
(-0.387) (-0.797)
C×2016-2017 0.087 0.128
(0.655) (0.587)
C×2018-2019 0.250* 0.348*
(1.794) (1.681)
C×2020-2021 0.356** 0.436*
(2.465) (1.922)
C×HighRegn×2016-2017 0.384** 0.539**
(2.286) (2.045)
C×HighRegn×2018-2019 0.408* 0.376
(1.876) (1.109)
C×HighRegn×2020-2021 0.942*** 0.729*
(2.964) (1.646)
Controls Y Y
R2 0.545 0.535
Adjusted R2 0.545 0.535
- 53 -

Panel B. Countries with BEPS Action 4 or Action 5 “Issues”
(1) (2)
C -0.907*** -0.815***
(-9.880) (-5.421)
C×Action4or5issue -0.206 -0.290
(-0.832) (-0.893)
C×2016-2017 0.346*** 0.494***
(4.417) (4.111)
C×2018-2019 0.483*** 0.553***
(4.896) (3.719)
C×2020-2021 0.772*** 0.802***
(6.279) (4.315)
C×Action4or5issue×2016-2017 -0.299* -0.464*
(-1.540) (-1.500)
C×Action4or5issue×2018-2019 -0.227 -0.287
(-1.116) (-1.000)
C×Action4or5issue×2020-2021 -0.350* -0.246
(-1.312) (-0.625)
Controls Y Y
R2 0.546 0.538
Adjusted R2 0.546 0.538
Notes: This table provides regression results for global MNCs’ tax-motivated income shifting around the
implementation of private CbCR, where the dependent variable is lnEBIT. In Panel A (B), an indicator variable is
included representing the strength of the regulatory environment of the headquarter-countries (whether the
headquarter-country has an “issue” with BEPS Action 4 or Action 5). The regressions are estimated using OLS
regression. Standard errors have been adjusted for clustering at the MNC level and the related t-statistics are reported
in parentheses. *, **, and *** indicate two-tailed statistical significance at the ten, five, and one percent levels,
respectively, except for Panel B which uses one-tailed statistical significance. Variables are defined in Appendix B.
- 54 -

Mandatory CBC Report Affect MNE BEPS (2024)

Uploaded by

Copyright:

Available Formats

Mandatory CBC Report Affect MNE BEPS (2024)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mandatory CBC Report Affect MNE BEPS (2024)

Uploaded by

Copyright:

Available Formats

Did Private Country-by-Country Reporting Really Not Affect Tax-Motivated International

Keywords: country-by-country reporting, private disclosure, mandatory disclosure, disclosure

Electronic copy available at: https://ssrn.com/abstract=4840684

Electronic copy available at: https://ssrn.com/abstract=4840684

following research question: Did mandatory private country-by-country reporting (CbCR)

shifting of multinational corporations (MNCs)?

Electronic copy available at: https://ssrn.com/abstract=4840684

reported in these countries.

Electronic copy available at: https://ssrn.com/abstract=4840684

there may be country-level differences in MNCs’ income shifting responses, as evidenced by

shifting would be seen in the expansion to a global sample of MNCs.

Electronic copy available at: https://ssrn.com/abstract=4840684

and the De Simone et al. [2017] affiliate-level income shifting models.

decreases are monotonic across MNC size.

Electronic copy available at: https://ssrn.com/abstract=4840684

they apply to the non-treated MNCs in my sample.

In additional analysis, I explore cross-sectional differences in the regulatory environments

headquartered in countries with stronger regulatory environments exhibit greater decreases in

My results are important to governments and policymakers worldwide, particularly the

Electronic copy available at: https://ssrn.com/abstract=4840684

provisions may yield important insights.

2. Background and Hypothesis Development

2.1 Mandatory Tax Disclosure

Electronic copy available at: https://ssrn.com/abstract=4840684

by Hoopes et al. [2023]).

2.2 Mandatory Private CbCR and Tax-Motivated International Income Shifting

Electronic copy available at: https://ssrn.com/abstract=4840684

in those countries (Hanlon [2018]).

significant decrease in MNCs’ income shifting. Using a sample of affiliate-years belonging to

implementation of private CbCR (coefficient of 2.122 on C×CbCR×Post, t-stat of 1.801 in her

Table 9 Panel A column 1).

Electronic copy available at: https://ssrn.com/abstract=4840684

decreased income shifting may be seen in a longer sample period.

seen in the expansion to a global sample of MNCs.

I state my hypothesis as follows:

Electronic copy available at: https://ssrn.com/abstract=4840684

Unlike Action 13 Country-by-Country Reporting, Action 4 generally applies to all MNCs by

the following in our discussion about my study:6

Electronic copy available at: https://ssrn.com/abstract=4840684

decrease their income shifting.

3. Sample and Research Design

3.1 Sample Selection

my EU sample has 289,231 affiliate-years from 2010 to 2021.

Electronic copy available at: https://ssrn.com/abstract=4840684

can be found in Table 1 Panels B, C, and D, respectively.

3.2 Research Design

affiliate-level income shifting models modified for a difference-in-differences design as follows:

𝑙𝑛𝐸𝐵𝐼𝑇𝑖,𝑡 = 𝛽0 + 𝛽1 𝐶𝑖,𝑡 + 𝛽2 𝐶𝑏𝐶𝑅𝑀 + 𝛽3 𝐶𝑏𝐶𝑅𝑀 × 𝑃𝑜𝑠𝑡𝑡 + 𝛽4 𝐶𝑖,𝑡 × 𝐶𝑏𝐶𝑅𝑀 + 𝛽5 𝐶𝑖,𝑡 × 𝑃𝑜𝑠𝑡𝑡

or the De Simone et al. [2017] capital-weighted tax rate differential:

equation 2a or 2b, which is alternatively the revenue-weighted or capital-weighted tax rate

Electronic copy available at: https://ssrn.com/abstract=4840684

included in equation 1 as it is subsumed by year fixed effects.

is the standard difference-in-differences coefficient. β6 is an empirical estimate of the differential

to be generally statistically indistinguishable from zero. Because tax-motivated income shifting

extent greater than MNCs not subject to private CbCR.

Electronic copy available at: https://ssrn.com/abstract=4840684

4.1 Summary Statistics

Electronic copy available at: https://ssrn.com/abstract=4840684

countries with greater GDP but lesser GDP per capita.