Download as pdf or txt
Download as pdf or txt
You are on page 1of 52

The Accounting Review • Issues in Accounting Education • Accounting Horizons

Accounting and the Public Interest • Auditing: A Journal of Practice & Theory
Behavioral Research in Accounting • Current Issues in Auditing
Journal of Emerging Technologies in Accounting • Journal of Information Systems
Journal of International Accounting Research
Journal of Management Accounting Research • The ATA Journal of Legal Tax Research
The Journal of the American Taxation Association

Online Early — Preprint of Accepted Manuscript


This is a PDF file of a manuscript that has been accepted for publication in an American
Accounting Association journal. It is the final version that was uploaded and approved by the
author(s). While the paper has been through the usual rigorous peer review process for AAA
journals, it has not been copyedited, nor have the graphics and tables been modified for final
publication. Also note that the paper may refer to online Appendices and/or Supplements that are
not yet available. The manuscript will undergo copyediting, typesetting and review of page proofs
before it is published in its final form, therefore the published version will look different from this
version and may also have some differences in content.
We have posted this preliminary version of the manuscript as a service to our members and
subscribers in the interest of making the information available for distribution and citation as
quickly as possible following acceptance.

The DOI for this manuscript and the correct format for citing the paper are given at the top
of the online (html) abstract.
Once the final published version of this paper is posted online, it will replace this
preliminary version at the specified DOI.
Noncompliance with Mandatory Disclosure Requirements: The Magnitude and
Determinants of Undisclosed Permanently Reinvested Earnings

Benjamin C. Ayers
University of Georgia
(706) 542-3772
bayers@terry.uga.edu

Casey M. Schwab
University of Georgia
(706) 542-3596
cschwab@terry.uga.edu

Steven Utke
University of Georgia
(706) 542-3633
sutke@uga.edu

April 10, 2014

preprint
Editor’s note: Accepted by Morton Pincus.
Submitted May 2013
Accepted June 2014

accepted
manuscript
Keywords: Permanently Reinvested Earnings, Mandatory Disclosure, Voluntary Disclosure, Tax

JEL codes: M40, M41, H25, K34

Data Availability: Data used in this study are available from public sources identified in the
paper.

_______________
Ayers, Schwab, and Utke gratefully acknowledge the support of the Terry College of Business and the
J.M. Tull School of Accounting. This paper has benefited from helpful comments from Ashley Austin,
Steve Baginski, Andy Bauer, Will Ciconte, Anne Ehinger, Lisa Hinson, Becky Lester, Tom Omer, Mort
Pincus (editor), John Robinson, Bridget Stomberg, Jake Thornock, Erin Towery, Ben Whipple, two
anonymous referees, and workshop participants at the University of Georgia and the University of Illinois
Tax Doctoral Consortium. We also thank Scott Dyreng (Duke University) for providing data on the
location of firms’ foreign subsidiaries.

1
Noncompliance with Mandatory Disclosure Requirements: The Magnitude and
Determinants of Undisclosed Permanently Reinvested Earnings

Abstract
We develop estimates of a firm's foreign earnings designated as permanently reinvested (PRE)
and the unrecorded deferred tax liability (TAX) associated with PRE that are independent of
whether a firm explicitly discloses this information. We then investigate firms’ noncompliance
with ASC 740 provisions requiring financial statement disclosure of PRE and either the tax
associated with PRE or a statement that calculating the tax is not practicable. We find that a
nontrivial portion of firms do not comply with the PRE disclosure requirements and that the
amounts of undisclosed PRE and the related tax are substantial in magnitude. Cross-sectional
evidence suggests managers opportunistically choose when to disclose PRE and TAX and that
compliance with PRE disclosure requirements increased following the American Jobs Creation
Act of 2004 which increased incentives to disclose PRE.

preprint
accepted
manuscript
I. INTRODUCTION

U.S. corporations defer U.S. taxation on earnings of foreign subsidiaries until the

earnings are repatriated to the U.S. Accounting Standards Codification (ASC) 740 requires

companies to accrue this future (deferred) U.S. income tax expense on their financial statements

unless the company declares the foreign earnings as permanently reinvested earnings (PRE)

overseas. Prior studies examining the consequences of PRE provide evidence that being able to

avoid accruing deferred taxes by declaring PRE impacts the location of foreign operations

(Graham, Hanlon, and Shevlin 2011), the repatriation of foreign earnings (Graham et al. 2011;

Blouin, Krull, and Robinson 2012), the amount and location of foreign cash (Blouin et al. 2012;

Blouin, Krull, and Robinson 2013), and foreign investments (Bryant-Kutcher, Eiler, and

Guenther 2008; Edwards, Kravet, and Wilson 2013). Regulators and the financial media have

expressed concerns regarding the effects of PRE. They argue that PRE has contributed to the
preprint
unprecedented $1.9 trillion in earnings held offshore by U.S. multinationals (Levin and Coburn

accepted
2012; Aubin 2013) which results in inefficient capital structures because of cash trapped

manuscript
overseas (Linebaugh 2012) and inflates a firm’s balance sheet and income statement by reporting

lower deferred tax liabilities and tax expense, respectively (Levin 2012).

Regulators and financial commentators are also concerned that the implications of PRE

are not transparent to investors as a result of missing or insufficient financial statement

disclosures (Tully 2011; Whitehouse 2011; Abahoonie and Barbut 2012; FAF 2013).

Specifically, the investment community is concerned that current disclosures make it difficult to

assess how PRE impacts reported earnings (Reilly 2012), domestic and foreign liquidity needs

(Whitehouse 2011), and the composition of cash versus non-cash foreign assets (Hoffelder

2012). The Securities and Exchange Commission (SEC) is also increasingly questioning the

1
adequacy of firms’ PRE-related disclosures, issuing 113 comment letters requesting additional

PRE-related disclosure from 2010 to 2012 versus ten such letters from 2004 to 2009. ASC 740

requires firms designating undistributed foreign earnings as PRE to disclose annually the

cumulative amount of PRE and either the amount of tax due if the earnings were repatriated or a

statement that calculating the tax is not practicable. To shed light on concerns regarding the

sufficiency of PRE disclosures, we investigate firms’ disclosure of PRE and its related tax and

identify factors associated with non-disclosure.

To examine noncompliance with required PRE disclosures, we analyze financial

statement disclosures of S&P 500 firms from 1999 through 2010. We use two financial statement

disclosures to identify firms required to disclose PRE. First, we look for an explicit disclosure of

PRE or the tax due upon repatriation of PRE. If a firm discloses either, we classify the firm-year

as a required PRE disclosure year. Second, we review effective tax rate (ETR) reconciliations for
preprint
disclosure of a tax benefit of foreign earnings, i.e., firms with a lower ETR attributable to foreign

accepted
earnings subject to a current tax rate lower than the U.S. tax rate, where the lower rate is not

manuscript
offset by deferred U.S. taxes due on repatriation. To ensure the foreign tax benefit reported in the

ETR reconciliation does not reflect foreign tax credits unrelated to PRE, we eliminate firm-year

observations in which the tax rate benefit source is ambiguous.1 This process yields a sample of

3,320 firm-year observations subject to the ASC 740 mandatory disclosure requirements.

Within this sample, we estimate the amount of PRE and the related tax due upon

repatriation (TAX) for each firm. Estimating PRE and TAX represents an important departure

from prior studies that assume PRE is disclosed if it exists and TAX is disclosed if it is

1
We confirmed the validity of our method with a Big 4 ASC 740 expert and with a Big 4 accounting partner serving
large multinational S&P 500 clients with required PRE disclosures. Adding further credence that our method allows
us to correctly identify firms with PRE that fail to comply with disclosure requirements, over 85 percent of our
sample firms that do not disclose their PRE amount acknowledge they have PRE. Results are robust to estimating
our analyses after excluding firms that do not acknowledge the existence of PRE.

2
practicable to calculate. Given the material consequences of PRE, undisclosed PRE represents a

failure by firms to disclose mandated information that materially impacts firm behavior.

Estimating PRE allows us to quantify the frequency and magnitude of undisclosed PRE which is

important from an enforcement perspective for the SEC and a policy perspective for those

concerned with the unprecedented level of earnings held overseas by U.S. multinationals.

Estimating TAX allows us to quantify the economic magnitude of undisclosed TAX and provide

data helpful in assessing the impracticability of calculating TAX.2

We estimate TAX using annual ETR reconciliation data, which provide the tax benefit

associated with not recording U.S. deferred tax expense on earnings designated as PRE. We

estimate TAX as the three-year sum of the net accumulated tax benefits associated with earnings

designated as PRE during this period. While estimating TAX using three years of data generally

understates the actual tax due on repatriation of cumulative PRE, this estimate should be
preprint
sufficient to evaluate cross-sectionally how the magnitude of TAX influences disclosure and the

accepted
practicability of estimating TAX. We estimate PRE by first grossing up the estimated TAX by

manuscript
the difference between the U.S. statutory rate and the firm’s foreign effective tax rate to produce

an estimate of foreign pre-tax earnings classified as PRE. Because PRE is an after-tax number,

we multiply the estimated foreign pre-tax earnings by one minus the foreign effective tax rate to

estimate the firm’s PRE. Correlation analyses indicate a significant positive correlation between

our estimates of PRE and TAX and reported PRE and TAX for firms disclosing these amounts.

Specifically, after scaling each measure by three-year pre-tax income, the correlation between

2
Even though disclosing that TAX is not practicable is permitted under ASC 740, the SEC has begun to question
firms why it is “not practicable” to disclose TAX. This suggests the SEC is uncertain why TAX is not practicable to
calculate, particularly when firms simultaneously record deferred tax liabilities on foreign earnings not designated as
PRE. We discuss SEC comment letters regarding PRE and TAX in more detail in Section 5.

3
estimated and reported PRE (TAX) equals 0.34 (0.44).3

Annually between 9.9 and 16.7 percent of our sample firms do not disclose PRE despite

being required to do so. We estimate annual aggregate undisclosed PRE ranges from $5.2 to

$99.0 billion and annual average undisclosed PRE per non-discloser ranges from $150 million to

$2.61 billion. These results suggest the magnitude of undisclosed PRE is substantial and that

some of the largest, most sophisticated firms in the U.S. capital markets do not comply with the

mandatory disclosure requirements for PRE over a long window of time.

Annually between 71.6 and 83.1 percent of our sample firms disclose TAX is not

practicable to calculate or do not mention TAX at all. These large percentages are interesting

given that 74.9 percent of these firms calculate and disclose the tax benefit associated with

current year foreign earnings deemed permanently reinvested in their ETR reconciliation and our

estimates of TAX significantly correlate with TAX for firms disclosing TAX. We estimate
preprint
annual aggregate undisclosed TAX ranges from $10.20 to $76.88 billion and annual average

undisclosed TAX per non-discloser ranges from $70 to $280 million.4


accepted
manuscript
Next we examine the determinants of a firm’s choice to comply with the PRE disclosure

requirement and a firm's choice to disclose TAX versus stating TAX is not practicable to

calculate or not mentioning TAX at all. Given PRE and TAX disclosures reveal strategic

information regarding foreign investments, we view the related disclosures as costly and

anticipate that compliance with the PRE disclosure requirements and the disclosure of TAX

varies cross-sectionally with the costs and benefits of disclosure.

We find a positive association between PRE disclosure and estimated TAX and a

3
The correlation between unscaled estimated PRE (TAX) and reported PRE (TAX) is 0.69 (0.87).
4
The range of the annual estimated aggregate undisclosed TAX is similar in magnitude to that of annual estimated
aggregate undisclosed PRE because the vast majority of firms do not disclose TAX, while a smaller portion of firms
do not disclose PRE. However, the annual average undisclosed TAX per non-disclosing TAX firm is much lower
than the annual average undisclosed PRE per non-disclosing PRE firm.

4
negative association between PRE disclosure and estimated PRE. These results suggest that the

unrecorded deferred tax liability associated with PRE motivates PRE disclosure as opposed to

the magnitude of the PRE itself. The negative association between PRE disclosure and estimated

PRE is consistent with higher political costs of disclosing large amounts of earnings permanently

reinvested abroad.5 Firms with more leverage, a greater percentage of income earned abroad,

more geographic segments, and larger absolute discretionary accruals are also less likely to

disclose PRE. These results are consistent with increased noncompliance when firms benefit

more from eliminating the TAX on PRE as a recorded liability, have more material foreign

operations, and exhibit more financial reporting discretion.

For TAX, we find firms with more material estimated PRE are less likely to disclose

TAX. Interestingly, after controlling for estimated PRE, we find no association between

estimated TAX and TAX disclosure. Explanations for the negative association between
preprint
estimated PRE and TAX disclosure include TAX being less practicable to calculate for firms

accepted
with larger PRE or firms being less likely to disclose TAX if the disclosure reflects negatively on

manuscript
the firm. In supplemental analyses, we include a number of proxies for the complexity in

calculating TAX and find the negative relation between TAX disclosure and estimated PRE

continues to hold. We also find that firms are more likely to disclose TAX if they are smaller,

have lower institutional ownership, or are more profitable. These results are consistent with

disclosure increasing when political costs are lower, information asymmetry is higher, or

operating performance is stronger.

In supplemental analyses, we examine firms’ responses to increased disclosure incentives

5
Business press articles highlighting concerns regarding PRE include: Fleischer, V. “Overseas cash and the tax
games multinationals play,” The New York Times, October 3, 2012; Fisher, D. “Foreign tax reserves are ‘crack
cocaine’ for earnings manipulation, study says,” Forbes.com, October 25, 2012; Cohn, M. “Accounting option
facilitates multinational earnings manipulation,” Accountingtoday.com, October 12, 2012.

5
created by the American Jobs Creation Act of 2004 (AJCA). The AJCA allowed firms to

repatriate foreign earnings at a reduced tax cost. To take full advantage of the AJCA provisions,

firms must have disclosed the amount of their PRE in the years leading up to the AJCA.6 By

linking the repatriation tax benefits to PRE disclosure, the AJCA gave firms incentives to

disclose PRE, but not TAX, in anticipation of future repatriation holidays similar to the AJCA.

Consistent with the AJCA providing incentives for PRE disclosure, we find an increase in PRE

disclosure but not TAX disclosure following the AJCA. We recognize that various other factors

could have improved PRE and TAX reporting in years following the AJCA, including the

issuance of FIN 48 and increased media and political scrutiny on PRE and foreign cash balances.

These factors make it difficult to attribute any long-run response solely to anticipation of a future

tax holiday.

This study makes three primary contributions. First, we develop estimates of PRE and
preprint
TAX that are independent of whether firms explicitly disclose such information. These estimates

accepted
should be useful to regulators, financial statement users, and academics in their efforts to identify

manuscript
noncompliance with PRE disclosures, estimate the impact of PRE and TAX on firms’ financial

statements, and examine how PRE impacts firm behavior.

Second, we address concerns that the implications of PRE are not transparent to investors

due to noncompliance with the mandated disclosure requirements. A nontrivial percentage of

firms fail to disclose PRE, and most firms disclosing PRE state that TAX is not practicable to

calculate or provide no mention of TAX. We acknowledge the inherent complexity in estimating

the deferred tax on foreign earnings that are or are not designated as PRE. However, our ability

6
The AJCA repatriation tax holiday allowed firms to receive an 85 percent dividends received deduction on cash
dividends from controlled foreign corporations. The preferential dividend received deduction was limited to the
greater of $500 million or the PRE reported in financial statements issued on or before June 30, 2003. As such, firms
reporting PRE generally derived a more substantial benefit than firms with undisclosed PRE.

6
to generate proxies for the undisclosed tax that significantly correlate with disclosed TAX is

consistent with the SEC questioning that disclosures are “not practicable.” Moreover, our cross-

sectional evidence with respect to disclosure incentives suggests managers opportunistically

choose when to disclose PRE and TAX.

Third, this study contributes to the growing literature investigating noncompliance with

mandatory disclosure requirements. We provide evidence that firms strategically choose not to

comply with mandatory disclosure requirements in a setting involving a long-standing

mandatory disclosure requirement where (1) there is no ambiguity in the requirement or nature of

the disclosure or underlying transactions requiring disclosure, (2) firms are heavily followed by

market participants, and (3) the required disclosure for PRE requires minimal effort. Thus, this

study provides evidence regarding mandatory disclosure compliance in a setting not subject to a

number of concerns faced by prior studies. Among other findings, evidence of increased PRE
preprint
disclosure following the AJCA suggests mandatory disclosure compliance is not only influenced

accepted
by the current regulatory environment but also by managers’ anticipation of future regulatory

manuscript
action that affects the net cost of compliance. This result suggests regulators need to consider

both the immediate impact of temporary regulations and how temporary regulations affect future

expectations and managerial behavior.

II. DISCLOSURE REQUIREMENTS AND PRIOR LITERATURE

Disclosure Requirements

Since 1972, APB No. 23 (now ASC 740-30-25-17) has permitted firms to avoid

recording deferred taxes on a foreign subsidiary’s earnings that are designated permanently

reinvested abroad. To designate earnings as PRE, the parent company should have specific plans

for reinvestment of the PRE. Firms taking advantage of the PRE exception must disclose

7
annually the amount of PRE and an estimate of the unrecorded deferred tax liability on the PRE

(TAX) or, alternatively, disclose that an estimate of TAX is not practicable (ASC 740-30-50-2).7

Despite differences in the expected timing of repatriation for PRE and non-PRE, the

process for calculating the TAX associated with PRE is no different than the process for

calculating a recognized deferred tax liability (DTL) for non-PRE. Both the TAX and DTL must

be based on current tax law without speculation regarding future tax law changes, current foreign

tax credits, etc. This process is inherently complex for firms with PRE and/or non-PRE,

especially for firms without accounting systems in place to estimate tax repatriation costs. In

SFAS 109’s “Basis for Conclusions” (FASB 1992), the Board states the “calculation of a

deferred tax liability for undistributed earnings that are or will be invested in a foreign subsidiary

indefinitely may sometimes be extremely complex. The hypothetical nature of those calculations

introduces significant implementation issues and other complexities that occur less frequently in
preprint
calculations of a deferred tax liability for an expected remittance of earnings from a foreign

accepted
entity. For that reason, the Exposure Draft proposed to not require recognition of a deferred tax

manuscript
liability for undistributed earnings that are or will be invested in a foreign entity indefinitely.”8

Review of Related Literature

Disclosure Literature

Within the disclosure literature, firm compliance with mandatory disclosure requirements
7
ASC 740-30-50-2 requires disclosure irrespective of PRE materiality. However, SEC rules generally only require
public firms to disclose material information (e.g., Regulation S-X, Rule 4-02 and Rule 4-08(h)(2)). Because we
identify required disclosers based on PRE-related disclosure in SEC filings, by construction, our methodology
identifies firms with material PRE and TAX.
8
We highlight that the Board concluded the calculation of TAX “may sometimes be” extremely complex. While our
method cannot prove that estimating TAX is practicable for non-disclosers, we provide evidence consistent with the
SEC questioning whether calculating TAX is impracticable. In addition, discussions with two international tax
directors indicated that their firms maintain comprehensive models tracking foreign earnings and repatriation costs
despite disclosing that TAX is not practicable to calculate in their financial statements. Although we exercise
caution in drawing conclusions from anecdotal evidence, this evidence suggests that some firms track tax
repatriation costs internally despite disclosing that it is impracticable to calculate TAX. Finally, our review of SEC
comment letters in Section 5 suggests that some non-disclosing firms disclose TAX in response to SEC comment
letters requesting disclosure, further suggesting that calculating TAX is practicable for some non-disclosers.

8
has garnered substantially less attention. Although most studies assume that firms comply with

mandatory disclosure requirements (e.g., Botosan 1997; Collins, Hand, and Shackelford 2001;

Francis, Nanda, and Olsson 2008), other research documents noncompliance with required

disclosures relating to environmental liabilities (Barth, McNichols, and Wilson 1997; Li,

Richardson, and Thornton 1997; Peters and Romi 2013), legal liabilities (Chen, Hou,

Richardson, and Ye 2013), tax contingencies (Gleason and Mills 2002; Robinson and Schmidt

2013), executive compensation (Robinson, Xue, and Yu 2011), and internal control weaknesses

(Rice and Weber 2012). Consistent with voluntary disclosure incentives influencing compliance

with mandatory disclosure requirements, prior studies provide evidence that compliance is

increasing in the size and estimated materiality of the item (Barth et al. 1997; Gleason and Mills

2002), regulatory oversight (Barth et al. 1997), access to capital markets (Barth et al. 1997; Chen

et al. 2013), litigation risk and publicity (Li et al. 1997; Gleason and Mills 2002), the absence of
preprint
specific managerial incentives for noncompliance (Robinson et al. 2011), and SEC enforcement

accepted
action (Robinson et al. 2011).

manuscript
Prior evidence should be interpreted in light of the following limitations. First, in some

settings, such as loss contingencies, the ambiguous nature of the standards or underlying

transactions is likely the primary source of non-disclosure. Second, the difficulty in estimating

information possessed by management, both in terms of dollar amount and materiality, limits the

ability to conclude management is willfully noncompliant. Finally, in settings examining

recently enacted standards (Robinson et al. 2011; Robinson and Schmidt 2013), management

may be slow to comply with the new standards due to uncertainty surrounding the requirements,

resource constraints, or lack of regulatory enforcement during the early period of a standard,

making it difficult to generalize evidence to more “steady state” disclosure requirements.

9
Several characteristics of our setting make it advantageous to investigate noncompliance

with mandatory disclosure requirements. First, the disclosure requirements are clear and

unambiguous under ASC 740 – firms designating undistributed foreign earnings as PRE are

required to disclose annually the cumulative amount of PRE and either the tax due upon PRE

repatriation or a statement that calculating the tax is not practicable. Moreover, these disclosures

have been required for decades. Second, our PRE and TAX estimates are based on the firm’s line

item disclosures of material differences between the U.S. statutory tax rate and the firm’s

effective tax rate on foreign operations. Thus, by construction, the PRE and TAX amounts are

material for our sample firms. Third, because our estimates are based on disclosed ETR

reconciliation items, we can identify that managers have declared PRE regardless of whether

they comply with the ASC 740 disclosure requirements. Fourth, PRE is based on historical,

verifiable information, and firms must have a specific plan for reinvestment of earnings to
preprint
qualify for the PRE designation. Accordingly, disclosing solely PRE requires minimal effort.

accepted
Permanently Reinvested Earnings Literature

manuscript
As discussed earlier, several studies examine the consequences of the PRE designation.

These studies provide evidence that the ability to declare PRE impacts decisions regarding the

location of foreign operations (Graham et al. 2011), the repatriation of foreign earnings (Graham

et al. 2011; Blouin et al. 2012), the amount and location of foreign cash (Blouin et al. 2012,

2013), and investments (Bryant-Kutcher et al. 2008; De Waegenaere and Sansing 2008; Hanlon,

Lester, and Verdi 2012; Edwards et al. 2013). Other studies examine the market valuation of

TAX and find that investors capitalize TAX that is disclosed (Collins et al. 2001) but do not

capitalize undisclosed TAX (Bauman and Shaw 2008).

These studies assume PRE is disclosed if it exists and TAX is disclosed if it is practicable

10
to calculate. As such, these studies do not examine firms’ compliance with the reporting

requirements of ASC 740. In a contemporaneous paper, Eiler and Kutcher (2013, hereafter EK)

examine a firm’s decision to disclose TAX. EK assume firms disclose PRE when required and

do not examine potential noncompliance. EK also do not estimate PRE, quantify undisclosed

PRE, or quantify undisclosed TAX. Finally, EK do not examine the effect of the AJCA on PRE

or TAX disclosures. Accordingly, our study makes key contributions distinct from EK.

III. RESEARCH DESIGN

Identification of Firms Required to Disclose PRE and TAX

We review 10-K filings to identify firms required to disclose PRE. We review each

firm’s income tax disclosures and identify firm-years that include an explicit disclosure of PRE,

TAX, or a statement that calculating TAX is not practicable. We classify firm-years with any of

these disclosures as required to disclose PRE. We then review each firm’s ETR reconciliation,
preprint
which reconciles a firm’s ETR to the domestic federal statutory rate and, as mandated under

accepted
ASC 740-10-50-12, lists reconciling items that result in a material difference between a firm’s

manuscript
ETR and the domestic federal tax rate.9 We collect data on reconciling line items related to

foreign operations. Reconciling items related to foreign tax rate differentials that reduce the

firm’s ETR suggest the firm has designated at least some of its foreign earnings as permanently

reinvested and is required to disclose PRE. Collecting data on reconciling items allows us to

identify firm-years required to disclose PRE that fail to do so and estimate PRE and TAX for

firm-years regardless of whether they disclose PRE or TAX.

Donohoe, McGill, and Outslay (2012) indicate firms often put other items not related to

9
SEC Regulation S-X, Rule 4-08(h)(2) states that any reconciling item that is individually less than five percent of
the product of pre-tax income multiplied by the applicable federal statutory income tax rate can be aggregated in the
ETR reconciliation. Assuming a 35 percent federal statutory income rate, this rule implies that ETR differences less
than 1.75 percent of pre-tax income can be aggregated.

11
foreign tax rate differentials into the foreign ETR line item. They specifically identify foreign tax

credits, repatriations, subpart F income, foreign export incentives, and audit settlements as being

misclassified in their example disclosures. Repatriations, subpart F income, and audit settlements

generally increase the effective tax rate, which would result in us improperly excluding firms

that actually have foreign rate benefits from PRE. Foreign tax credits and export incentives can

decrease the effective tax rate, potentially resulting in improperly including firms in our sample.

Most firms in our sample disclose foreign tax credits and export benefits as separate line items or

discuss them in the footnote. We exclude firm-years if the firm discussed foreign tax credits or

export benefits without a separate ETR reconciliation line item and without clearly specifying

the magnitude of the foreign tax credits or export benefits relative to the rate benefit. As such, we

believe we correctly determine whether a firm is required to disclose PRE. Consistent with our

method correctly identifying firms with PRE that fail to disclose PRE, over 85 percent of our
preprint
sample non-disclosers acknowledge the existence of PRE despite not disclosing the amount in

accepted
their financial statements. Nonetheless, our PRE and TAX estimates are measured with error to

manuscript
the extent the identified ETR reconciliation items include non-PRE related items.

Estimation of PRE and TAX

For firm-years identified as required to disclose PRE, we collect ETR reconciling items

related to tax rate differentials due to current year foreign operations (ReconcilingForeign), prior

year earnings designated as PRE in the current year (ReconcilingNon-CurrentPRE), PRE repatriations

(ReconcilingRepat), and reversal of earnings previously designated as PRE (ReconcilingReversal).10

We then estimate the accumulated TAX in year t associated with the accumulated PRE

10
See Appendix B for details on our data collection process and examples of footnote disclosures. We use
ReconcilingForeign and ReconcilingNon-CurrentPRE to identify firms required to disclose PRE because these items indicate
the presence of accumulated PRE at year end. In our sample, 72.9, 24.6, 1.8, and 0.7 percent of firms are included
due to disclosing a ReconcilingForeign item, a PRE amount, TAX is “not practicable” to calculate, and a TAX amount,
respectively. All firms disclosing ReconcilingNon-CurrentPRE also disclose one of the other four items.

12
(TAXEstimated) as follows:

TAXEstimated = ∑ +∑ -
∑ -∑

If TAXEstimated is negative, we set it equal to zero. Because TAXEstimated is based on three

years of data, it could be less than zero, for example, if a firm repatriates ten years’ worth of

PRE. We also set TAXEstimated equal to zero for firms that disclose PRE but do not disclose a

material ETR reconciling item related to PRE in the current and two preceding years.11

Accordingly, TAXEstimated is a conservative measure because it only captures the taxes on the prior

three years of PRE and ETR reconciling items that were material enough to warrant disclosure.12

We estimate a firm’s PRE (PREEstimated) by first calculating TAX on an annual basis using

only current year data. In year t, we label this single year measure CurrentTAXt. We then

estimate the earnings a firm designates as PRE in year t using the following formula:
preprint
CurrentPREt = (CurrentTAXt / (35% - FETRt)) * (1 – FETRt). In this calculation, we gross up

accepted
CurrentTAXt by the difference between the U.S. statutory rate and the firm’s foreign ETR

manuscript
(FETRt) to produce an estimate of the foreign pre-tax earnings classified as PRE in year t. Since

PRE is an after-tax number, we multiply estimated foreign pre-tax earnings by one minus FETRt

to estimate the firm’s current addition to PRE in year t.13 We then sum CurrentPRE over years t-

11
While TAX is likely greater than zero for these firms, the absence of material ETR reconciliation items related to
PRE suggests a level of TAX that is less than the TAX for firms disclosing material ETR reconciliation items. As
such, TAXEstimated for firms without reconciling items should identify firms with lower amounts of TAX.
12
The intensive data collection required to estimate PRE and TAX limit our ability to base PRE and TAX estimates
on all historical footnote data available. We estimate PRE and TAX based on three years of data to ensure a
consistent estimation approach across all firm-year observations. To the extent identified ETR reconciliation items
include non-PRE related items, TAXEstimated could be over- or understated.
13
See Appendix A for details on our calculation of FETR. If FETRt is greater than 35 percent, CurrentTAXt is set
equal to zero because it is unlikely tax would be due on repatriation. We generate an estimated PRE of zero for firms
that disclose PRE, TAX, or that calculating TAX is not practicable but do not disclose a foreign ETR benefit in their
rate reconciliation over the past 3 years, have a foreign ETR that exceeds the 35 percent U.S. statutory tax rate, or
have repatriations that exceed the related tax benefit of PRE designated during our three-year measurement window.
While actual PRE for these firms is greater than zero, we anticipate their PRE will be lower than other sample firms
because their ETR benefit from PRE was not sufficient for disclosure in the ETR reconciliation during the three-year

13
2, t-1, and t to derive our three-year measure of PRE, PREEstimated.14

Empirical Design

Determinants of PRE Disclosure

We use the following logit model to investigate the determinants of PRE disclosure:

β0 + β1PREEstimated/PIit + β2TAXEstimated/PIit + β3E/Pit + β4Leverageit +


Disclosureit = (1)
β5PROAit + β6NAnalystit + β7Instit + β8LnSalesit + β9Foreign%it +
β10HHIit + β11Segmentsit + β12AbsDiscAccrit + γkIndustryk + αtYeart + εit

The dependent variable, Disclosure, is an indicator variable equal to one if a firm

discloses the amount of its PRE and zero otherwise. Equation (1) includes six categories of

variables that proxy for factors that likely influence compliance with the PRE disclosure

requirement: (1) materiality of PRE and TAX amounts (PREEstimated/PI and TAXEstimated/PI), (2)

market pressure (E/P and Leverage), (3) firm performance (PROA), (4) external information

preprint
environment (NAnalyst, Inst, and LnSales), (5) firm operating environment (Foreign%, HHI, and

Segments), and (6) reporting discretion (AbsDiscAccr).

accepted
We include PREEstimated/PI and TAXEstimated/PI to assess whether the magnitudes of PRE

manuscript
and TAX influence mandatory disclosure compliance. PRE /PI and TAX Estimated Estimated/PI equal

the ratios of PREEstimated and TAXEstimated, respectively, to the three-year sum of the absolute value

of pre-tax income (PI). Deflating by the three-year sum of the absolute value of pre-tax income

window, they have smaller incentive to elect PRE given a foreign ETR that exceeds the U.S. tax rate, or they had
repatriations that result in no additional net PRE during the three-year window. In our PRE regression sample, we
estimate TAX to be greater than zero but PRE equal to zero for 8.8 percent of firms. Inferences are similar when we
eliminate firms for which PREEstimated equals zero.
14
Alternatively, PRE could be estimated using the following formula: PREEstimated = TAXEstimated / (35%-FETR)*(1-
FETR), where TAXEstimated and FETR are based on three years of data. Although this estimation method is more
similar to the method used to calculate TAXEstimated, this estimation method is affected more by unusual one year
items, such as large tax refunds or settlements, large book losses not deductible for tax purposes, etc., that occur
during the three-year estimation window. These unusual items can result in a three-year FETR greater than 35
percent which results in values of estimated PRE that equal zero, are undefined, or are negative. By computing PRE
annually and then summing current PRE over the three-year estimation window, unusual one year items result in
only one of the three years during the estimation window having a zero PRE estimate while the cumulative sum of
three years of data still results in a positive PRE balance.

14
captures the materiality of PRE and TAX relative to the firm’s profits. Since by construction,

sample firms are required to disclose PRE, there is no reason pursuant to ASC 740 to expect that

the materiality of PRE or TAX affects the required disclosure of PRE. However, prior research

(Gleason and Mills 2002) finds in certain circumstances that mandatory disclosure compliance

increases with the materiality of the disclosure item, suggesting that firms with larger

PREEstimated/PI and TAXEstimated/PI would be more likely to disclose PRE. In contrast, it is possible

that PREEstimated/PI and TAXEstimated/PI proxy for political costs (Watts and Zimmerman 1986;

Wagenhofer 1990; Healy and Palepu 2001; Frischmann, Shevlin, and Wilson 2008; Mills,

Nutter, and Schwab 2013), which might lead to less disclosure if firms with large PRE or TAX

are viewed as not paying their “fair share” of taxes. In this setting, political costs might include

costs associated with public scrutiny of amounts permanently reinvested abroad, IRS audits,

additional taxes due upon audit, or the imposition of new taxes.15 Accordingly, we do not make a
preprint
directional prediction for the coefficients on PREEstimated/PI or TAXEstimated/PI in our PRE model.

accepted
E/P proxies for decreased market pressure on managers to hide items that negatively

manuscript
impact earnings (Dechow, Ge, Larson, and Sloan 2011) as well as growth opportunities and

information asymmetry (Smith and Watts 1992). Because E/P proxies for a variety of opposing

effects, we make no directional prediction. Leverage represents a firm’s debt recorded on its

balance sheet. It is likely more costly for firms with higher leverage to disclose PRE which

suggests the presence of unrecorded liabilities. This suggests a negative association between

Leverage and Disclosure. We include a proxy for profitability (PROA) to control for any impact

performance has on the disclosure decision. Because firms are more likely to disclose costly

information when firm performance is high (Miller 2002), we anticipate a positive association

15
Consistent with a link between PRE disclosures public scrutiny, Microsoft and Hewlett-Packard, both PRE
disclosers throughout our sample period, have been subject to Congressional investigations focused on PRE and
other offshoring issues (Levin and Coburn 2012; Dixon 2012).

15
between PROA and Disclosure.

We include three proxies for a firm’s information environment (NAnalyst, Inst, and

LnSales). We use the number of analysts, NAnalyst, to proxy for information demand. Because

information demand is increasing in the number of analysts, we expect a positive association

between Disclosure and NAnalyst (Francis et al. 2008; Zechman 2010). We use the percentage of

shares owned by institutions, Inst, to proxy for information asymmetry. The impact of higher

institutional ownership is ambiguous because it can represent lower information asymmetry and

less need for disclosure (Bushee, Matsumoto, and Miller 2003) or higher information demand

and thus more disclosure (Ajinkya, Bhojraj, and Sengupta 2005). Hence, we make no directional

prediction for Inst. We use the natural log of sales (LnSales) to control for firm size. Because

firm size proxies for a variety of factors, we make no directional prediction for LnSales.

We include three variables, Foreign%, HHI, and Segments, to proxy for a firm’s
preprint
operating environment. We use the percentage of pre-tax income earned abroad, Foreign%, and

accepted
the number of disclosed geographical segments, Segments, to proxy for the importance of a

manuscript
firm’s foreign operations and potential propriety costs associated with disclosures related to

foreign operations.16 As such, we expect a negative association between Disclosure and both

Foreign% and Segments. We use the ranked Herfindahl index, HHI, to proxy for industry-level

proprietary costs (Verrecchia and Weber 2006). Higher values of HHI indicate the firm operates

in a less competitive industry and, thus, has a lower cost associated with disclosing proprietary

information. This suggests a positive association between Disclosure and HHI.

We include the absolute value of performance-matched discretionary accruals

(AbsDiscAccr) to proxy for financial reporting discretion and expect that firms exercising greater

16
Proprietary costs are costs of disclosing sensitive information to competitors, investors, and creditors (Verrecchia
1983). In our setting, proprietary costs arise from disclosing the amount and duration of earnings invested abroad.

16
financial reporting discretion are less transparent in their disclosures. This is particularly likely in

the PRE setting given Krull (2004) finds that firms use the PRE designation to manage earnings

to beat analysts’ forecasts. Finally, we include year and Fama and French 17 industry fixed

effects to control for differences in disclosure practices across time and industry.

Determinants of TAX Disclosure

To investigate the determinants of TAX disclosure, we make two modifications to

equation (1). First, because TAX disclosure is uncommon when PRE is not disclosed, we

estimate our TAX model using only firms that disclose PRE.17 Second, we set Disclosure equal

to one if a firm discloses the actual TAX amount and zero if the firm does not disclose TAX or

discloses that TAX is not practicable to determine.

Unlike PRE disclosure which is mandatory for firms in our sample, we anticipate

managers view TAX as a voluntary disclosure because ASC 740 allows managers to disclose
preprint
that TAX is not practicable to determine. Because our control variables proxy for benefits or

accepted
costs of disclosure, most predictions are the same when considering TAX disclosure. Three

manuscript
notable exceptions relate to the expected coefficients on TAXEstimated/PI, PREEstimated/PI, and

Segments. Since TAX disclosure is more voluntary in nature and TAXEstimated/PI represents the

cash taxes and financial reporting expenses associated with the repatriation of PRE, we anticipate

a negative association between TAX Disclosure and TAXEstimated/PI. This prediction is consistent

with prior studies that find managers are less likely to voluntarily disclose information that

adversely affects firm value (Robinson et al. 2011) and that TAX disclosure adversely affects

firm value for firms that disclose PRE (Bauman and Shaw 2008). We also expect a negative

association between TAX Disclosure and TAXEstimated/PI to the extent the complexity of

calculating TAX increases with TAXEstimated/PI. When considering TAX disclosure, larger values
17
If we include firms that do not disclose PRE in our TAX sample and estimate our TAX regression, all results hold.

17
of PREEstimated/PI likely attract additional scrutiny of TAX and, thus, increase political costs.

Also, larger values of PREEstimated/PI potentially indicate more complex TAX calculations. As

such, we expect a negative relation between PREEstimated/PI and TAX Disclosure. We make no

prediction on the relation between Segments and TAX Disclosure. More Segments likely

increase the complexity of calculating TAX, suggesting a negative relation with TAX

Disclosure. On the other hand, the presence of more Segments indicates that a firm considers its

various geographic segments to be an important way of tracking firm activity and suggests firms

have the information systems in place to track complex foreign information.

IV. SAMPLE SELECTION AND DESCRIPTIVE STATISTICS

Sample Selection

Table 1 outlines our sample selection process. We limit our sample to firms in the S&P

500 at any time between 1999 and 2010.18 Focusing on S&P 500 firms results in a manageable
preprint
sample for hand-collection of data while ensuring our analyses use an economically significant

accepted
component of the population of firms. We eliminate firm-years not subject to corporate tax,

manuscript
foreign firm-years, and firm-years missing Form 10-K. From the hand-collected sample, we limit

our sample to firms required to disclose PRE as defined in Section 3, yielding 3,320 firm-years

for which we estimate PRE and TAX. After requiring Compustat, I/B/E/S, and Thomson Reuters

data necessary for our regression analyses, our sample consists of 2,977 firm-year observations.

INSERT TABLE 1 HERE

Descriptive Statistics

PRE and TAX Estimates

Table 2, Panel A presents descriptive statistics on unscaled and scaled PRE (PREReported,

18
These disclosures are also relevant to non-S&P 500 firms. In 2010, 19.1 percent of Compustat firms, representing
44.5 percent of the market value of Compustat firms, report positive foreign source income and thus have the
possibility of designating current year earnings as PRE.

18
PREEstimated, PREReported/PI, and PREEstimated/PI), variables underlying our PRE estimate

(TAXEstimated and FETR), and PRE estimation errors (PRE/PIEstimationError). As expected, reported

PRE, both scaled and unscaled, exceeds estimated PRE for PRE disclosers. To quantify the

magnitude of estimation error, we calculate PRE/PIEstimationError which equals the absolute value

of the difference between PREReported/PI and PREEstimated/PI. The mean (median) value of

PRE/PIEstimationError equals 0.272 (0.176). To identify sources of PRE/PIEstimationError, Panel B

presents the results from regressing PRE/PIEstimationError on factors potentially associated with

estimation errors. The first three variables, PRE=0: FETR>=35%, PRE=0: Repatriation, and

PRE=0: No ETR Rec Item, are indicator variables equal to one if estimated PRE in year t equals

zero because FETR is greater than 35 percent, the firm repatriates foreign earnings that

completely offsets PRE during the estimation window, or the firm has no PRE-related ETR

reconciliation item during the estimation window, respectively. Given that PRE is unlikely equal
preprint
to zero for these firms, we expect a positive association between PRE/PIEstimationError and these

accepted
measures. Other Repatriation is an indicator variable equal to one if the firm repatriated foreign

manuscript
earnings during the estimation window, but the repatriation was not so large that it reduced

estimated PRE to zero. To the extent these repatriations consist, at least in part, of earnings

designated as PRE prior to the estimation window, repatriations likely increase estimation errors

suggesting a positive association between PRE/PIEstimationError and Other Repatriation. FETR

Volatility is an indicator variable equal to one for firm-years with above median volatility of

foreign ETR reconciliation items over the estimation window. Estimated TAX is a key input into

our PRE estimate and is based on foreign ETR reconciliation items. Volatile foreign ETR

reconciliation items likely represent inconsistent disclosure of ETR reconciliation items,

variation in the foreign tax rate, or variation in the extent of foreign earnings declared annually

19
as PRE. These possibilities potentially decrease the quality of our PRE estimate. As such, we

expect a positive association between PRE/PIEstimationError and FETR Volatility. The remaining

measures, LnSales, Segments, Haven, and #Exhibit21Countries, capture firm size or other firm

characteristics associated with foreign operations. We make no predictions for these variables.

INSERT TABLE 2 HERE

We find PRE estimation errors are larger for firms with zero estimated PRE due to high

foreign ETRs (PRE=0: FETR>=35%) and our inability to estimate PRE (PRE=0: No ETR Rec

Item). We also find larger estimation errors for firms with more volatile foreign ETR

reconciliation items (FETR Volatility) and material operations in more foreign countries

(#Exhibit21Countries). Interestingly, we find estimation error decreases for firms with a material

subsidiary in a tax haven country (Haven).

Table 2, Panel C presents correlations between measures of reported and estimated PRE,
preprint
which suggest that our PRE estimates significantly correlate with reported PRE amounts. The

accepted
Pearson (Spearman) correlation between PREEstimated/PI and PREReported/PI is 0.337 (0.443); the

manuscript
Pearson (Spearman) correlation between PREEstimated and PREReported, is 0.687 (0.573).

INSERT TABLE 3 HERE

Table 3 provides similar descriptive statistics and analyses for TAX. As expected,

reported TAX, both scaled and unscaled, exceeds estimated TAX for TAX disclosers. To

quantify the magnitude of estimation error, we calculate TAX/PIEstimationError which equals the

absolute value of the difference between TAXReported/PI and TAXEstimated/PI. The mean (median)

value of TAX/PIEstimationError equals 0.053 (0.015). To identify sources of TAX/PIEstimationError,

Panel B presents the results from regressing TAX/PIEstimationError on factors potentially associated

with estimation errors. Similar to PRE estimation errors, we anticipate larger errors when our

20
methodology produces a TAX estimate equal to zero (TAX=0: Repatriation; TAX=0: No Rate

Rec Item), when the firm repatriates during the estimation window (Other Repatriation), and

when the firm has volatile foreign ETR reconciliation items (FETR Volatility). Consistent with

expectations, we find TAX estimation errors are larger for firms with zero estimated TAX due to

repatriations and for firms with volatile foreign ETR reconciliation items. We also find that

estimation errors are larger for firms with a material subsidiary in a tax haven country (Haven).

The relatively low adjusted R2 in the PRE and TAX estimation error regressions indicates that

the estimation errors are largely noise and not explained by factors expected to generate errors.

Table 3, Panel C presents correlations between measures of reported and estimated TAX,

which suggest that our TAX estimates significantly correlate with reported TAX amounts. The

Pearson (Spearman) correlation between TAXEstimated/PI and TAXReported/PI is 0.437 (0.541); the

Pearson (Spearman) correlation between TAXEstimated and TAXReported, is 0.872 (0.587).19


preprint
In sum, descriptive data suggest that estimates for PRE and TAX are sufficient to

accepted
evaluate cross-sectionally how the magnitude of PRE and TAX influences disclosure and assess

manuscript
the practicability of estimating TAX. Indeed, our ability to generate proxies for TAX based on

three years of ETR reconciliation data that exhibit significant correlations with reported TAX,

19
TAXEstimated performs reasonably well relative to the measure developed by Bauman and Shaw (2008). The Pearson
(Spearman) correlation between TAXEstimated and TAXReported is 0.872 (0.587), relative to an analogous correlation of
0.778 (0.459) for the Bauman and Shaw measure in our sample. The Pearson (Spearman) correlation between our
scaled measure of TAX (TAXEstimated/PI) and a scaled measure of reported TAX (TAXReported/PI) is 0.437 (0.541),
relative to an analogous correlation of 0.594 (0.546) for the Bauman and Shaw measure. Because the Bauman and
Shaw measure is based on the cumulative disclosed PRE balance rather than three years of ETR reconciliations, it is
not surprising their estimate of cumulative tax performs better than our measure on some dimensions. Using five
years of data to calculate TAXEstimated yields correlations between unscaled measures of TAXEstimated with TAXReported
that continue to be higher than analogous correlations for Bauman and Shaw’s measure and correlations of our five-
year scaled measure of TAXEstimated with TAXReported that move closer to that of Bauman and Shaw. In paired t-test
comparisons of scaled absolute estimation errors, we find that the estimation error is larger using our methodology
over three years (p-value < 0.01) and over five years (p-value < 0.06) than the corresponding Bauman and Shaw
methodology. One advantage of our method is that it allows us to estimate TAX for firms that do not disclose PRE.
As a robustness test, we re-estimate our TAX regression using actual PRE and the Bauman and Shaw TAX estimate;
all results hold. Additionally, we estimate both the PRE and TAX regressions using our estimates based on five
years of data; all results hold except Segments is negative but insignificant in the PRE regression (p-value = 0.12).

21
when disclosed, is consistent with the SEC questioning the “not practicable” disclosure.

Table 4 presents PRE descriptive statistics partitioned by year. Panel A indicates 12.8

percent of our sample does not disclose PRE when required. Non-disclosure declines following

the enactment of the AJCA in 2004, with approximately 15.3 (11.4) percent of firm-year

observations not disclosing PRE prior to (following) the AJCA. The magnitude of PRE generally

increases over time. For disclosers, PREReported (PREEstimated) increased from $202.0 ($110.7)

billion in 1999 to $994.9 ($616.8) billion in 2010. A notable exception to this trend occurs in

2005 when PREReported declined as firms repatriated foreign earnings during the AJCA

repatriation tax holiday. For non-disclosers, PREEstimated increased from $5.8 billion in 1999 to

$99.0 billion in 2010. The only noticeable decline occurs in 2007, which likely results from

lower levels of foreign earnings declared as PRE at the beginning of the financial crisis.20

INSERT TABLE 4 HERE


preprint
Table 5 presents TAX descriptive statistics by year. Panel A indicates that 77.4 percent of

accepted
our sample either does not disclose TAX or discloses that TAX is not practicable to calculate. In

manuscript
65.5 percent of non-discloser firm-years, firms state that is not practicable to estimate TAX with

the remaining 34.5 percent of non-disclosers silent with respect to TAX.21 Unlike the increase of

PRE disclosure over time, the percentage of firms disclosing TAX generally decreases over time,

with 26.8 percent of firms disclosing TAX in 1999 and 16.9 percent of firms disclosing TAX in

2010. The only increase occurs in 2004 when the AJCA was enacted. Table 5 also indicates an

20
Non-disclosure of PRE occurs among even the largest, most well-known firms. Dell, GE, and Pepsi changed from
non-disclosers in 2001 to disclosing over $25 billion (combined) in PRE in 2002. Their ETR line items were not
substantially different in 2002 from prior years, suggesting the majority of PRE disclosed in 2002 existed in 2001.
21
Relative to firms that disclose estimating TAX is not practicable, firms that remain silent are significantly smaller,
more profitable, in more competitive industries, and exercise more financial reporting discretion. Although there is
little informational difference between disclosing that TAX is not practicable relative to remaining silent, these
differences yield inferences similar to those from our primary analysis discussed later—firms disclose less when
they are likely to have a weaker information environment, i.e., are smaller, operate in a more competitive industry,
and exercise more financial reporting discretion.

22
increase in the magnitude of TAX over time. For disclosers, TAXReported (TAXEstimated) increased

from $7.55 ($3.07) billion in 1999 to $46.11 ($25.52) billion in 2010. For non-disclosers,

TAXEstimated increased from $10.20 billion in 1999 to $76.88 billion in 2010 suggesting the

amount of the undisclosed TAX has risen dramatically during the sample period.

INSERT TABLE 5 HERE

Regression Variable Descriptive Statistics

Table 6, Panel A presents univariate statistics for our PRE regression sample partitioned

by PRE disclosure status. The mean PREEstimated/PI is not statistically different for disclosers and

non-disclosers, although the median PREEstimated/PI is statistically larger for disclosers. In

contrast, the mean and median TAXEstimated/PI are significantly larger for disclosers than non-

disclosers. In combination, these univariate data provide preliminary evidence suggesting the

materiality of PRE or TAX likely influences PRE disclosure. PRE disclosers also have less
preprint
leverage, are more profitable, are larger, derive less of their income from foreign sources, have

accepted
fewer geographic segments, and exhibit less financial reporting discretion.

manuscript
INSERT TABLE 6 HERE

Table 6, Panel B presents univariate statistics for the TAX regression sample partitioned

by TAX disclosure status. In contrast to the data in Panel A, TAX disclosers have less material

levels of PRE (PREEstimated/PI) and TAX (TAXEstimated/PI). Collectively, the data in Panels A and

B suggest firms with material levels of TAX or PRE comply with the minimum disclosure

requirements by being more likely to disclose PRE, but do not disclose TAX. TAX disclosure is

important because TAX impacts the tax and financial reporting effects of repatriation and the

repatriation decision. TAX disclosers also have less leverage, are more profitable, have a greater

analyst following, have lower institutional ownership, are smaller, operate in less competitive

23
industries, report more geographic segments, and exhibit more financial reporting discretion.

V. MULTIVARIATE RESULTS

Determinants of PRE Disclosure

Table 7, Column 1 presents the results from estimating equation (1) with Disclosure

equal to one for firm-years in which a firm discloses PRE, and zero otherwise.22 We find a

negative and significant coefficient on PREEstimated/PI (p-value = 0.003) and positive and

significant coefficient on TAXEstimated/PI (p-value = 0.006).23 These results suggest that concern

over the materiality of the unrecorded tax liability associated with PRE motivates PRE disclosure

more so than the materiality of PRE itself. The negative association between PRE disclosure and

PREEstimated/PI is consistent with higher political costs of disclosing large amounts of earnings

permanently reinvested abroad. The marginal effects indicate that a one standard deviation

increase in PREEstimated/PI (TAXEstimated/PI) decreases (increases) the probability of disclosure by

2.4 percent (3.3 percent).24 preprint


accepted
INSERT TABLE 7 HERE

manuscript
We find a negative and significant association between PRE disclosure and Leverage (p-

value = 0.013), suggesting that firms benefiting more from eliminating the TAX on PRE as a

recorded liability are less likely to disclose PRE. We find a negative and significant association

22
VIFs and condition indices do not exceed 7 in any of our regressions, which are below cutoffs commonly used to
assess multicollinearity (Belsley, Kuh, and Welsh 1980; Mendenhall and Sincich 2003).
23
Scaling by PI is intuitively appealing because TAXEstimated/PI captures the extent to which a firm increases its
earnings by designating foreign earnings as PRE. If we scale PREEstimated and TAXEstimated by the three-year sum of
gross profits or sales, all results hold. Also, all results hold using the log of PREEstimated/PI and TAXEstimated/PI.
24
We compute the marginal effect as the effect of a one standard deviation change across the mean of a variable,
from ½ standard deviation below the mean to ½ standard deviation above the mean (or from zero to one for
categorical variables), with all other variables held constant at their means (Caramanis and Lennox 2008). For
example, the effect of a one standard deviation increase in TAXEstimated/PI on PRE disclosure is calculated by
estimating the expected probability of PRE disclosure when TAXEstimated/PI equals 0.014 [0.042 (the sample mean
of TAXEstimated/PI) - 0.055/2 (½ of the standard deviation of TAXEstimated/PI)], and all other variables are held at
their mean, and again estimating the expected probability of PRE disclosure when TAXEstimated/PI equals 0.069
(0.042 + 0.055/2), and all other variables are held at their mean. The marginal effect represents the difference
between these two probabilities. See Long and Freese (2005).

24
between PRE disclosure and both Foreign% (p-value = 0.005) and Segments (p-value = 0.097),

suggesting that firms with more material foreign operations are less likely to disclose PRE.

Finally, we find a negative coefficient on AbsDiscAccr (p-value = 0.009), consistent with firms

exhibiting greater financial reporting discretion being less transparent in their PRE disclosure.

Determinants of TAX Disclosure

Table 7, Column 2 presents the results from estimating equation (1) in the subsample of

firms that disclose PRE with Disclosure equal to one for firm-years in which a firm discloses

TAX, and zero otherwise. We find a negative and significant coefficient on PREEstimated/PI (p-

value = 0.020) and an insignificant coefficient on TAXEstimated/PI (p-value = 0.378). The results in

Columns (1) and (2) suggest firms with more material PRE are less likely to disclose their PRE

and the TAX associated with their PRE. Possible explanations include TAX being less

practicable to calculate for firms with larger PRE or firms being less likely to disclose TAX if
preprint
the disclosure reflects negatively on the firm. We investigate the association between TAX

accepted
disclosure and complexity in supplemental analyses below.

manuscript
We also find that larger firms (LnSales, p-value = 0.029) and firms with a larger

percentage of institutional owners (Inst, p-value = 0.022) are less likely to disclose TAX. The

institutional ownership result is consistent with findings in prior literature that institutions prefer

less disclosure to maintain an informational advantage. The size result is counterintuitive if it

represents sophistication of accounting systems to estimate TAX; it is intuitive if it captures the

complexity, and thus difficulty of calculating TAX, or political visibility, i.e., larger firms avoid

scrutiny associated with TAX by not disclosing TAX. We find that firms with better operating

performance (PROA, p-value = 0.035) are more likely to disclose TAX, consistent with more

profitable firms being less concerned with the disclosing the unrecorded liability. The

25
coefficients on the remaining variables are not significantly associated with TAX disclosure.25

Interestingly, we do not find a significant association between TAX disclosure and Segments, a

common proxy for firm complexity, which could factor into the firm’s TAX calculation.26

Supplemental Analyses

The Impact of the AJCA on PRE and TAX Disclosure

The AJCA allowed firms to repatriate foreign earnings at a reduced tax cost during 2004

and 2005. To take full advantage of this provision, firms must have disclosed the amount of their

PRE in years leading up to the AJCA. By linking the repatriation tax benefits to PRE disclosure,

the AJCA incentivized firms to disclose PRE, but not TAX, in anticipation of future repatriation

holidays similar to the AJCA.27 To test whether the AJCA increased PRE disclosure, we test

whether the average year fixed effects from estimating equation (1) (untabulated) during the

AJCA (2004 – 2005) and post-AJCA (2006 – 2010) periods are greater than the average year

fixed effect during the pre-AJCA (1999 – 2003) period. preprint


accepted
INSERT TABLE 8 HERE

manuscript
Table 8, Panel A presents results from testing whether the average year fixed effects

during the AJCA and post-AJCA periods are greater than the average year fixed effects during

the pre-AJCA period when estimating equation (1) within the full sample. Consistent with

expectations, the probability of PRE disclosure is significantly greater in the AJCA (p-value =

0.027) and post-AJCA (p-value = 0.057) periods relative to the pre-AJCA period. However, the

probability of TAX disclosure is not significantly different in the AJCA (p-value = 0.206) or

25
HHI is positive and significant (p-value = 0.039) when we exclude industry fixed effects. This suggests the
disclosure effect of proprietary costs, proxied for by industry competition, is captured by the industry fixed effects.
26
If we scale PREEstimated and TAXEstimated by the three-year sum of gross profits or sales, all results hold. Also, all
results hold using the log of PREEstimated/PI and TAXEstimated/PI.
27
Tax executives working at two of our PRE non-disclosing firms at the time of the AJCA indicated that the AJCA
was one of the main factors they considered in their PRE disclosure decision. Also, firms likely anticipated future
repatriation tax holidays following the AJCA due to a variety of lobbying and legislative efforts aimed at enacting a
second repatriation tax holiday (Kocieniewski 2011; Rauf 2012).

26
post-AJCA (p-value = 0.989) periods. Figure 1, Panels A and B plot the annual marginal effects

of the year fixed effects for PRE and TAX disclosure, respectively, and show a distinct increase

in the year marginal effects for PRE disclosure around the AJCA and continuing into later years.

Collectively, these results are interesting because they suggest an impact of a tax law change on

financial reporting behavior, and more specifically, mandatory disclosure compliance.

Nonetheless, we recognize that various other factors that could improve PRE and TAX reporting

occurred during the post-AJCA period, such as the issuance of FIN 48 and increased media and

political scrutiny on PRE and foreign cash balances. Those factors make it difficult to attribute

any long-run response solely to anticipation of a future tax holiday.

INSERT FIGURE 1 HERE

Table 8, Panels B and C present similar analyses based on estimating equation (1) within

subsamples of firms most likely to benefit from a future repatriation tax holiday—firms that
preprint
repatriated at any point in our sample and firms with estimated PRE exceeding $250 million.28

accepted
Within these subsamples, relative to the pre-AJCA period, we find a significant increase in PRE

manuscript
disclosure in the AJCA and post-AJCA periods (p-values < 0.025) but no significant increase in

TAX disclosure (p-values > 0.28). These results corroborate results presented in Panel A and

suggest the expected benefits of PRE disclosure play an important role in the disclosure decision.

SEC Comments Regarding PRE Disclosure

To better understand SEC enforcement and concerns related to PRE and TAX disclosure,

we analyze comment letters from fiscal year 2004, the initial year comment letters are publicly

available, to 2012. During this period, the SEC issued, on average, 4,288 comment letters

annually related to Form 10-K. On average, 600 of these letters per year mentioned “taxes.” The

SEC issued ten comment letters related to the disclosure of PRE or TAX from fiscal year 2004 to
28
We use $250 million instead of $500 million because our estimation methodology underestimates PRE.

27
2009 compared to 113 comment letters from 2010 to 2012. Of these letters, 44 request firms

disclose their PRE, 27 request firms disclose TAX or that its calculation is not practicable, 25

request both PRE and TAX disclosure, and 21 request firms explain why it is not practicable to

calculate TAX. The remaining six letters request clarification of the disclosure wording.

This analysis provides several insights.29 First, SEC focus on PRE and TAX disclosure

has increased substantially in recent years. Second, comment letters result in a sizable disclosure

of previously undisclosed PRE. Specifically, firms receiving SEC comment letters disclosed $8.5

billion of PRE. For the median firm, the previously undisclosed PRE represents 25.5 percent of

cumulative three-year pre-tax income and 20.7 percent of book value of equity. Third, the SEC

has begun to question firms why it is “not practicable” to disclose TAX. In response to the 52

comment letters requesting firms disclose TAX or that it is not practicable, 15 firms either

disclose TAX or state that they will disclose TAX going forward, with the remaining firms
preprint
disclosing that calculating TAX is not practicable. This evidence suggests that some firms do not

accepted
disclose TAX when calculating TAX is practicable.

manuscript
Further Examination of a Firm's Ability to Estimate TAX

In our primary TAX disclosure analysis, we include the number of geographic segments

as a proxy for the complexity in calculating TAX but find no association between TAX

disclosure and Segments. However, we do find that TAX disclosure is negatively associated with

PREEstimated/PI and LnSales, both of which could be explained by the increased complexity of

calculating TAX for firms with larger PRE or for larger firms. In sensitivity analyses, we

29
We use data from SEC comment letters to further validate our PRE estimates. These tests are based on four (28)
firms currently included (not included) in our sample that do not originally disclose PRE in a given year,
subsequently disclose PRE for that year in response to an SEC comment letter, and have sufficient data to estimate
PRE using our methodology. For the four firms included in our sample, the correlation between scaled (unscaled)
estimated PRE and PRE disclosed in response to the SEC is 0.92 (0.95). For the 28 firms not included in our sample,
the correlation between scaled (unscaled) estimated PRE and PRE disclosed in response to the SEC is 0.70 (0.88)
These correlations suggest our estimates are reasonable even for firms not disclosing PRE in their financial
statements.

28
consider five additional proxies for the complexity in calculating TAX—the volatility of pre-tax

foreign income, the volatility of the ETR reconciling items associated with PRE, the number of

Exhibit 21 subsidiaries, the number of Exhibit 21 countries, or whether a firm has a tax internal

control material weakness (ICMW). Volatility in a firm’s foreign earnings and in the extent to

which foreign earnings are permanently reinvested likely increases TAX complexity. Similarly,

TAX complexity likely increases with the number of subsidiaries and the number of countries in

which a firm operates. Finally, a tax ICMW indicates a potential lack of sophistication or

resources in the tax department, which decreases the likelihood the firm can compute TAX. We

find no significant association between TAX disclosure and these additional measures

(untabulated), suggesting that cross-sectional differences in the complexity of estimating TAX as

captured by these proxies are not significant determinants of TAX disclosure for our sample

firms.
preprint
VI. CONCLUSION

accepted
In this study, we develop an estimate of PRE and TAX that is independent of whether a

manuscript
firm explicitly discloses such information. Based on our estimates, we find that (1) annually

between 9.9 and 16.7 percent of our sample S&P 500 firms required to disclose PRE fail to do

so, (2) PRE disclosure varies with political costs, leverage, the importance of foreign operations,

financial reporting discretion, and increased incentives associated with the AJCA, and (3)

annually between 71.6 and 83.1 percent of our sample firms disclose TAX is not practicable to

calculate or provide no mention of TAX. Though we acknowledge the inherent complexity in

estimating the unrecorded deferred tax on foreign earnings designated as PRE, our ability to

generate proxies for the undisclosed tax based on three years of ETR reconciliation data that

significantly correlate with disclosed TAX is consistent with the SEC questioning the “not

29
practicable” disclosure. Collectively, this evidence is important as regulators assess the

sufficiency of current PRE and TAX disclosures, especially in light of the effects of PRE and

TAX on repatriations, the location of foreign operations, and investment of foreign funds.

This study contributes to the literature investigating noncompliance with mandatory

disclosure requirements. We provide evidence that firms appear to strategically choose not to

comply with mandatory PRE disclosure requirements in a setting not subject to a number of

concerns faced by prior studies. Moreover, our evidence of increased PRE disclosure following

the AJCA suggests mandatory disclosure compliance is influenced by managers’ anticipation of

future regulatory action that would affect the net cost of compliance. This result suggests

regulators need to consider not only the immediate impact of temporary regulations but also how

temporary regulations likely affect future expectations and managerial behavior.

Our study is subject to the following limitations. First, because we base our PRE and
preprint
TAX estimates on three years of ETR reconciliation data, our estimates are measured with error.

accepted
Second, we cannot rule out that non-disclosing firms differ in ways that makes their TAX

manuscript
calculations impracticable. However, empirical evidence that the TAX disclosure decision varies

with disclosure incentives and anecdotal evidence that some non-disclosing firms internally track

the costs of repatriation while others disclose TAX upon SEC inquiry suggests questioning the

impracticability of TAX has some merit. Third, because various other factors could improve

PRE and TAX reporting during the post-AJCA period, such as the issuance of FIN 48 and

increased media and political scrutiny on PRE and foreign cash balances, it is difficult to

attribute any long-run response solely to anticipation of a future tax holiday.

30
APPENDIX A
Variable Definitions
Variable Definition
Disclosureit (PRE) an indicator variable equal to one if firm i discloses the amount of its PRE in year t,
and zero otherwise.
Disclosureit (TAX) an indicator variable equal to one if firm i discloses the amount of its TAX in year t,
and zero otherwise.
PREReportedit equals firm i’s cumulative reported PRE balance in its year t SEC filings.
TAXReportedit equals firm i’s cumulative reported TAX balance in its year t SEC filings.
TAXEstimatedit equals firm i’s cumulative TAX balance from year t-2 to year t. Algebraically,
TAXEstimated for firm i in year t equals ∑ +
∑ -∑ -
∑ , where ReconcilingForeign equals the current year ETR
benefit from designating current year foreign earnings as PRE, ReconcilingNon-
CurrentPRE equals the current year ETR benefit from designating prior year foreign
earnings as PRE, ReconcilingRepat equals the ETR effect from repatriating PRE, and
ReconcilingReversal equals the current year ETR effect resulting from removing the
PRE designation from prior year foreign earnings originally designated PRE. If
TAXEstimatedit is negative, TAXEstimatedit is set equal to zero.
FETRit equals firm i’s foreign effective tax rate in year t. FETRit equals the ratio of the
current foreign taxes in year t (txfo-txdfo) to pre-tax foreign income (pifo) in year t.
FETRit is set to zero if the current foreign taxes are negative and pre-tax foreign

preprint
income is positive. FETRit is coded as missing if pre-tax foreign income is negative;
this has no effect on our estimation because we cannot obtain ETR items in those
years. Pre-tax foreign income in a given year is set equal to zero if a firm-year is
missing pre-tax foreign income and reports zero current foreign taxes. Current

accepted
foreign tax in a given year is set equal to zero if the firm-year is missing current
foreign taxes and reports zero pre-tax foreign income. If a firm-year is missing both
manuscript
pre-tax foreign income and current foreign taxes, both are set to zero only if the firm
also reports no ETR effect from foreign operations.
PREEstimatedit equals firm i’s cumulative PRE from year t-2 to year t. If TAXEstimatedit equals zero,
we set PREEstimatedit equal to zero. Otherwise, this is calculated by first estimating the
TAX on earnings designated PRE in year t based on only ETR reconciling items
from year t (CurrentTAXt), i.e., using the TAXEstimated formula above using only data
from year t. For each year, CurrentPREt = (CurrentTAXt / (35% - FETRt)) * (1 –
FETRt). If FETRt exceeds 35 percent, CurrentPREt is set equal to zero. We sum
CurrentPRE over years t-2, t-1, and t to derive our final PREEstimatedit.
PIit equals the sum of the absolute value of pre-tax income (pi) for firm i from year t-2
to t (∑ | |).
PREReported/PIit equals the ratio of firm i’s PREReportedit to PIit.
TAXReported/PIit equals the ratio of firm i’s TAXReportedit to PIit.
PREEstimated/PIit equals the ratio of firm i’s PREEstimatedit to PIit.
TAXEstimated/PIit equals the ratio of firm i’s TAXEstimatedit to PIit.

31
APPENDIX A (continued)
Variable Definitions
Variable Definition
PRE/PIEstimationErrorit equals the absolute value of the difference between PREReported/PIit and
PREEstimated/PIit
TAX/PIEstimationErrorit equals the absolute value of the difference between TAXReported/PIit and
TAXEstimated/PIit
E/Pit equals the ratio of firm i’s EPS to share price (epsfx/prcc_f) at the end of year t.
Leverageit equals the ratio firm i’s total debt to assets ((dltt+dlc-txndb)/at) at the end of year t.
PROAit equals firm i’s ratio of pre-tax income (pi) to total assets (at) at the end of year t.
NAnalystit equals the number of analysts issuing earnings forecasts for firm i in year t (I/B/E/S
numest).
Instit equals firm i’s ratio of the number of shares held by institutional investors
(Thompson Reuters 13F database shares/1,000,000) to total shares outstanding
(Thompson Reuters 13F database shrout1) at the end of year t.
LnSalesit equals the natural log of firm i’s sales (sale) at the end of year t.
Foreign%it equals firm i’s ratio of foreign pre-tax income (pifo) to worldwide pre-tax income
(pi) at the end of year t.
HHIit equals the decile ranking of the Herfindahl index in year t for firm i’s industry. The
Herfindahl index is calculated as the sum of the squares of the market shares of all
preprint
firms listed on Compustat within a particular two-digit SIC in year t. A firm's
market share equals the ratio of its sales (sale) in year t to the sum of sales for all
firms in its two-digit SIC code in year t.
Segmentsit
accepted
equals the number of geographic segments reported by firm i in year t (geoseg).

manuscript
AbsDiscAccrit equals the absolute value of performance-matched discretionary accruals for firm i
in year t, computed following Kothari, Leone and, Wasley (2005, 174, equation 7),
where total accruals are calculated based on the statement of cash flows (ib-(oancf-
xidoc)) as suggested in Hribar and Collins (2002).
PRE=0: Reasonsit an indicator variable equal to one if firm i had zero PREEstimatedit in year t for the
following Reasons: FETR>=35% if the FETR is greater than 35 percent;
Repatriation if ReconcilingRepat results in estimated TAX, and thus estimated PRE,
equal to zero; No ETR Rec Item if no ETR reconciling item is disclosed from years
t-2 to t, which results in estimated TAX and, thus estimated PRE, equal to zero.
(PRE) Other an indicator variable equal to one if firm i repatriated PRE from year t-2 to t and
Repatriationit PREEstimatedit does not equal zero.
(TAX) Other an indicator variable equal to one if firm i repatriated PRE from year t-2 to t and
Repatriationit TAXEstimatedit does not equal zero.
FETR Volatilityit an indicator variable equal to one if the standard deviation of firm i’s foreign ETR
reconciling items was above the sample median in year t, and zero otherwise.

32
APPENDIX A (continued)
Variable Definitions
Variable Definition
Havenit an indicator variable set equal to one if firm i had a tax haven in year t, and zero
otherwise. Haven classifications are based on Dyreng and Lindsey (2009).
#Exhibit21 an indicator variable set equal to one if firm i’s listed number of material operations
Countriesit in countries in Exhibit 21 exceeds the sample median in year t, and zero otherwise.
Data on the location of firms’ foreign subsidiaries are from Scott Dyreng’s website.
TAX=0: Reasonsit an indicator variable equal to one if firm i had zero TAXEstimatedit in year t for the
following Reasons: Repatriation if ReconcilingRepat exceeds estimated TAX; No
ETR Rec Item if no ETR reconciling item is disclosed from years t-2 to t.

preprint
accepted
manuscript

33
APPENDIX B
Examples of Sample Footnote Disclosures

Part 1: Typical Disclosure

Altria 12/31/2001 Tax Footnote:


--------------------------------------------------------------------------------------
2001 2000 1999
--------------------------------------------------------------------------------------
U.S. federal statutory rate 35.0% 35.0% 35.0%
Increase (decrease) resulting from:
State and local income taxes,
net of federal tax benefit 2.3 2.6 2.5
Rate differences--foreign
operations (2.3) (1.2) (0.6)
Goodwill amortization 2.3 1.3 1.4
Other 0.6 1.0 0.9
--------------------------------------------------------------------------------------
Effective tax rate 37.9% 38.7% 39.2%
======================================================================================

At December 31, 2001, applicable United States federal income taxes and foreign withholding taxes have
not been provided on approximately $5.6 billion of accumulated earnings of foreign subsidiaries that are
expected to be permanently reinvested. The Company is unable to provide a meaningful estimate of
additional deferred taxes that would have been provided were these earnings not considered permanently
reinvested.

Treatment: In the most straightforward case, the firm separately discloses its foreign tax rate

preprint
differential and either an estimate of TAX or a statement that calculating TAX is not practicable.
From this footnote, we collect the PRE disclosure ($5.6 billion), the ETR reconciliation items
(−2.3 percent in 2001), and the TAX related disclosure (“not practicable”).

accepted
Reconciling items can be disclosed as a dollar amount or as a percentage. When reconciling
items are disclosed only as a percentage, as is the case in the above disclosure, we convert the
manuscript
percentage into a dollar amount by multiplying the percentage by pre-tax income.
Part 2: Comprehensive Disclosure

Kellogg 1/3/2009 Tax Footnote:


2008 2007 2006
U.S. statutory income tax rate 35.0% 35.0% 35.0%
Foreign rates varying from 35% −5.0 −4.0 −3.5
State income taxes, net of federal benefit 1.0 1.1 1.3
Foreign earnings repatriation 1.6 2.3 1.2
Tax audits −1.5 — −1.7

Net change in valuation allowances — −.5 .5


Statutory rate changes, deferred tax impact −.1 −.6 —
International restructuring — −2.6 —
Other −1.3 −2.0 −1.1
Effective income tax rate 29.7% 28.7% 31.7%

During the year, the Company repatriated approximately $710 million of earnings and capital which
carried a cash tax charge of $24 million. This amount was less than the charge recorded during the year
due to the impact of favorable movements in foreign currency. This cost was further offset by foreign tax
related items of $16 million, reducing the net cost of repatriation to $8 million. The Company has

34
provided $18 million of deferred taxes related to the remaining $290 million of unremitted foreign
earnings. These amounts are reflected in the foreign earnings repatriation line item. At January 3, 2009,
accumulated foreign subsidiary earnings of approximately $834 million were considered indefinitely
invested in those businesses. Accordingly, U.S. income deferred taxes have not been provided on these
earnings and it is not practical to estimate the deferred tax impact of those earnings.
Treatment: This example contains the maximum amount of disclosures a firm can make in one
year. While rare, this provides insight into our overall data collection process. We collect the (a)
cumulative amount of disclosed PRE ($834 million, PREReported), (b) cumulative amount of
disclosed TAX (or “not practicable” disclosure, as is the case here), (c) current year ETR benefit
from declaring current year earnings permanently reinvested (-5 percent, ReconcilingForeign), (d)
current year tax expense associated with PRE repatriation ($8 million, ReconcilingRepat), (e)
current year tax expense associated with removing the PRE designation from prior year foreign
earnings ($18 million, ReconcilingReversal), and (f) current year tax benefit associated with
converting prior year foreign earnings to PRE (not applicable - since Kellogg has item (e) this
year they will not have item (f)).

Part 3: Non-Discloser

Coach 6/30/2007 Tax Footnote:

The Company accounts for income taxes in accordance with SFAS 109, “Accounting for Income Taxes.”
Under SFAS 109, a deferred tax liability or asset is recognized for the estimated future tax consequences
preprint
of temporary differences between the carrying amounts of assets and liabilities in the financial statements
and their respective tax bases. Coach does not provide for U.S. income taxes on the unremitted earnings
of its foreign subsidiaries as the Company intends to permanently reinvest these earnings.

accepted Fiscal Year Ended

Amount
manuscript
June 30, 2007

Percentage
July 1, 2006

Amount Percentage
July 2, 2005

Amount Percentage

Tax expense at U.S. statutory rate $362,135 35.0% $261,565 35.0% $192,997 35.0%
State taxes, net of federal
benefit 38,910 3.8 27,164 3.6 29,287 5.3
Reversal of deferred U.S. taxes on foreign
earnings — 0.0 — 0.0 (16,247) (2.9)
Foreign income subject to reduced tax rates (13,892) (1.3) (11,548) (1.5) (4,458) (0.8)
Other, net 10,988 1.0 6,309 0.8 (447) (0.1)
Taxes at effective worldwide rates $398,141 38.5% $283,490 37.9% $201,132 36.5%

Treatment: This firm separately discloses its foreign tax rate differential, which we capture in
our data collection (-$13.892 million in 2007). Despite the presence of an ETR reconciling item
and an explicit statement that the firm does not record U.S. taxes on permanently reinvested
earnings, this firm makes no disclosure related to either PRE or TAX.

35
REFERENCES

Abahoonie, E., and Y. Barbut. 2012. Undistributed foreign earnings: Disclosure requirements
and considerations. PricewaterhouseCoopers Practical Tip No. 2012-03.

Accounting Principles Board (APB). 1972. Accounting for Income Taxes – Special Areas. APB
Opinion No. 23. New York, NY: APB.

Ajinkya, B., S. Bhojraj, and P. Sengupta. 2005. The association between outside directors,
institutional investors and the properties of management earnings forecasts. Journal of
Accounting Research 43 (3): 343-376.

Aubin, D. 2013. U.S. companies’ overseas earnings hit record $1.9 trillion: Study. Reuters (May
8).

Barth, M. E., M. F. McNichols, and G. P. Wilson. 1997. Factors influencing firms’ disclosures
about environmental liabilities. Review of Accounting Studies 2: 35-64.

Bauman, M. P., and K. W. Shaw. 2008. The usefulness of disclosures of untaxed foreign
earnings in firm valuation. The Journal of the American Taxation Association 30 (2): 53-
77.

Belsley, D. A., E. Kuh, and R. E. Welsh. 1980. Regression Diagnostics: Identifying Influential
Data and Sources of Collinearity. Hoboken, NJ: John Wiley and Sons.
preprint
Blouin, J. L., L. K. Krull, and L. A. Robinson. 2012. Is U.S. multinational dividend repatriation
policy influenced by reporting incentives? The Accounting Review 87 (5): 1463-1491.

accepted
Blouin, J., L. Krull, and L. Robinson. 2013. The Location, Composition, and Investment

manuscript
Implications of Permanently Reinvested Earnings. Working paper, University of
Pennsylvania, University of Oregon, and Dartmouth College.

Botosan, C. A. 1997. Disclosure level and the cost of equity capital. The Accounting Review 72
(3): 323-349.

Bryant-Kutcher, L., L. Eiler, and D. A. Guenther. 2008. Taxes and financial assets: Valuing
permanently reinvested foreign earnings. National Tax Journal 61 (4): 699-720.

Bushee, B. J., D. A. Matsumoto, and G. S. Miller. 2003. Open versus closed conference calls:
The determinants and effects of broadening access to disclosure. Journal of Accounting
and Economics 34: 149-180.

Caramanis, C., and C. Lennox. 2008. Audit effort and earnings management. Journal of
Accounting and Economics 45: 116-138.

Chen, F., Y. Hou, G. Richardson, and M. Ye. 2013. Auditor Quality and Litigation Loss
Contingency Disclosures. Working paper, University of Toronto.

36
Collins, J. H., J. R. M. Hand, and D. A. Shackelford. 2001. Valuing deferral: The effect of
permanently reinvested foreign earnings on stock prices. In International Taxation and
Multinational Activity, edited by J. Hines, 143-171. Chicago, IL: University of Chicago
Press.

Dechow, P. M., W. Ge, C. R. Larson, and R. G. Sloan. 2011. Predicting material accounting
misstatements. Contemporary Accounting Research 28 (1): 17-82.

De Waegenaere, A., and R. C. Sansing. 2008. Taxation of international investment and


accounting valuation. Contemporary Accounting Research 25 (4): 1045-1066.

Dixon, K. 2012. Microsoft, HP skirted taxes via offshore units: U.S. Senate panel. Reuters
(September 20).

Donohoe, M. P., G. A. McGill, and E. Outslay. 2012. Through a glass darkly: What can we learn
about a U.S. multinational corporation’s international operations from its financial
statement disclosures? National Tax Journal 65 (4): 961-984.

Dyreng, S. D., and B. P. Lindsey. 2009. Using financial accounting data to examine the effect of
foreign operations located in tax havens and other countries on U.S. multinational firms’
tax rates. Journal of Accounting Research 47 (5): 1283-1316.

Edwards, A., T. Kravet, and R. Wilson. 2013. Trapped Cash and the Profitability of Foreign
Acquisitions. Working paper, University of Toronto, University of Texas at Dallas, and
University of Iowa. preprint
Eiler, L., and L. Kutcher. 2013. Disclosure Decisions Surrounding Permanently Reinvested

accepted
Foreign Earnings. Working paper, California State University, Fullerton and Colorado
State University.

manuscript
Financial Accounting Foundation (FAF). 2013. Post-Implementation Review Report on FASB
Statement No. 109, Accounting for Income Taxes. Norwalk, CT: FAF.

Financial Accounting Standards Board (FASB). 1992. Accounting for Income Taxes. Statement
of Financial Accounting Standards No. 109. Norwalk, CT: FASB.

Francis, J., D. Nanda, and P. Olsson. 2008. Voluntary disclosure, earnings quality, and cost of
capital. Journal of Accounting Research 46 (1): 53-99.

Frischmann, P. J., T. Shevlin, and R. Wilson. 2008. Economic consequences of increasing the
conformity in accounting for uncertain tax benefits. Journal of Accounting and
Economics 46: 261-278.

Gleason, C. A., and L. F. Mills. 2002. Materiality and contingent tax liability reporting. The
Accounting Review 77 (2): 317-342.

37
Graham, J. R., M. Hanlon, and T. Shevlin. 2011. Real effects of accounting rules: Evidence from
multinational firms’ investment location and profit repatriation decisions. Journal of
Accounting Research 49 (1): 137-185.

Hanlon, M., R. Lester, and R. Verdi. 2012. The Effect of Repatriation Tax Costs on U.S.
Multinational Investment. Working paper, Massachusetts Institute of Technology.

Healy, P. M., and K. G. Palepu. 2001. Information asymmetry, corporate disclosure, and the
capital markets: A review of the empirical disclosure literature. Journal of Accounting
and Economics 31: 405-440.

Hoffelder, K. 2012. Tax incentive helps multinationals repatriate 20% less, study finds.
CFO.com (November 12).

Hribar, P., and D. W. Collins. 2002. Errors in estimating accruals: Implications for empirical
research. Journal of Accounting Research 40 (1): 105-134.

Kocieniewski, D. 2011. Companies push for tax break on foreign cash. The New York Times
(June 19).

Kothari, S. P., A. J. Leone, and C. E. Wasley. 2005. Performance matched discretionary accrual
measures. Journal of Accounting and Economics 39: 163-197.

Krull, L. K. 2004. Permanently reinvested earnings, taxes, and earnings management. The
Accounting Review 79 (3): 745-767.
preprint
Levin, C. 2012. Statement before U.S. Senate Permanent Subcommittee on Investigations on
Offshore Profit Shifting and the U.S. Tax Code (September 20).
accepted
manuscript
Levin, C., and T. Coburn. 2012. Offshore Profit Shifting and the U.S. Tax Code. Senate
Permanent Subcommittee on Investigations (September 20).

Li, Y., G. D. Richardson, and D. B. Thornton. 1997. Corporate disclosure of environmental


liability information: Theory and evidence. Contemporary Accounting Research 14 (3):
435-474.

Linebaugh, K. 2012. Top U.S. firms are cash-rich abroad, cash-poor at home. Wall Street
Journal (December 4).

Long, J. S., and J. Freese. 2005. Regression Models for Categorical Outcomes Using Stata 2nd
Edition. College Station, TX: Stata Press.

Mendenhall, W., and T. Sincich. 2003. A Second Course in Statistics: Regression Analysis 6th
Edition. Upper Saddle River, NJ: Pearson Education.

Miller, G. S. 2002. Earnings performance and discretionary disclosure. Journal of Accounting


Research 40 (1): 173-204.

38
Mills, L. F., S. E. Nutter, and C. M. Schwab. 2013. The effect of political sensitivity and
bargaining power on taxes: Evidence from federal contractors. The Accounting Review 88
(3): 977-1005.

Peters, G. F., and A. M. Romi. 2013. Discretionary compliance with mandatory environmental
disclosures: Evidence from SEC filings. Journal of Accounting and Public Policy 32 (4):
213-236.

Rauf, D. S. 2012. Guess who’s pushing for tax holiday? Politico (April 5).

Reilly, D. 2012. Foreign profits can create earnings illusion. Wall Street Journal (October 8).

Rice, S. C., and D. P. Weber. 2012. How effective is internal control reporting under SOX 404?
Determinants of the (non-)disclosure of existing material weaknesses. Journal of
Accounting Research 50 (3): 811-843.

Robinson, J. R., Y. Xue, and Y. Yu. 2011. Determinants of disclosure noncompliance and the
effect of the SEC review: Evidence from the 2006 mandated compensation disclosure
regulations. The Accounting Review 86 (4): 1415-1444.

Robinson, L. A., and A. P. Schmidt. 2013. Firm and investor responses to uncertain tax benefit
disclosure requirements. The Journal of the American Taxation Association 35 (2): 85-
120.

preprint
Smith Jr, C. W., and R. L. Watts. 1992. The investment opportunity set and corporate financing,
dividend, and compensation policies. Journal of Financial Economics 32: 263-292.

Tully, S. 2011. The investor’s stake in corporate tax reform. CNN Money (April 13).
accepted
manuscript
Verrecchia, R. E. 1983. Discretionary disclosure. Journal of Accounting and Economics 5: 179-
194.

Verrecchia, R. E., and J. Weber. 2006. Redacted disclosure. Journal of Accounting Research 44
(4): 791-814.

Wagenhofer, A. 1990. Voluntary disclosure with a strategic opponent. Journal of Accounting


and Economics 12: 341-363.

Watts, R. L., and J. L. Zimmerman. 1986. Positive Accounting Theory. Upper Saddle River, NJ:
Prentice Hall.

Whitehouse, T. 2011. SEC squinting at overseas earnings. Compliance Weekly (June 14).

Zechman, S. L. C. 2010. The relation between voluntary disclosure and financial reporting:
Evidence from synthetic leases. Journal of Accounting Research 48 (3): 725-765.

39
FIGURE 1

Panel A: Plot of Year Marginal Effects – PRE Disclosure Regression

Year Marginal Effects for PRE


0.03
Year
0.02 Marginal
Effects
0.01

-1E-16

-0.01

-0.02

-0.03

-0.04
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Panel B: Plot of Year Marginal Effects – TAX Disclosure Regression


preprint
Year Marginal Effects for TAX
0.09
accepted
manuscript
0.08 Year
0.07 Marginal
Effects
0.06
0.05
0.04
0.03
0.02
0.01
0
-0.01
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Figure 1, Panels A and B plot the annual marginal effects of the year fixed effects for PRE and
TAX disclosure based on the estimation of equation (1).

40
TABLE 1
Sample Selection
Data Restrictions N
Compustat firm-year observations for all firms appearing
in the S&P 500 from 1999-2010 8,476
Less:
REIT firm-years (entities with no corporate level taxes) (236)
Firm-years missing basic Compustat data (SIC, CIK, (359)
stock price)
Foreign firm-years (e.g., 20-F filers) (44)
Firm-years subject to ownership by another firm (51)

Total observations to be hand collected 7,786

Less:
Firm-years with errors in disclosure (6)
Firm-years missing Form 10-K (112)
Total valid hand collected observations 7,668

Less:
Firm-years missing required info (e.g., Foreign ETR) (3,852)
Firm-years disclosing PRE less than or equal to 0 (114)
Firm-years not required to disclose PRE (382)

preprint
Full Sample 3,320

Less:
Firms missing necessary I/B/E/S, Compustat, and Thomson Reuters (343)

accepted
data (e.g., Number of Analysts, Segments, Institutional Ownership)
2,977
manuscript
Regression Sample
Note we hand collect PRE and TAX data beginning in 1997. Our final sample period includes observations from
1999 to 2010 because we require two prior years of data to estimate our PRE and TAX measures.

41
TABLE 2
Analysis of PRE Estimates
Panel A: PRE Estimates, Estimation Components, and Estimation Errors
Disclosers (N= 2,895)
Variable Mean StdDev Q1 Median Q3
Estimated and Reported PRE
PREReported 2,069.0 4,048.2 200.0 614.1 1,770.0
PREEstimated 1,292.5 3,440.4 0.0 114.4 753.7
PREReported/PI 0.5349 0.5608 0.1517 0.3622 0.6990
PREEstimated/PI 0.2673 0.5011 0.0000 0.0870 0.3413
Estimated PRE components
TAXEstimated 198.1 479.6 0.4 27.4 137.7
FETR 0.2630 0.1900 0.1281 0.2393 0.3553
PRE estimation error
PRE/PIEstimationError 0.2715 0.5762 0.0288 0.1760 0.4443
Panel B: Determinants of PRE Estimation Error
Variable Pred. Coef. (p-value)
Intercept ? 0.3090 (0.0361)
PRE=0: FETR>=35% + 0.0989 (0.0070)
PRE=0: Repatriation + 0.0354 (0.2980)
PRE=0: No ETR Rec Item + 0.0773 (0.0436)
Other Repatriation + -0.0251 (0.3006)
FETR Volatility
LnSales
+
? preprint 0.2538
-0.0130
(<0.0001)
(0.3995)
Segments ? 0.0035 (0.1346)
Haven ? -0.1001
accepted
(0.0490)
#Exhibit21Countries ? 0.1943 (<0.0001)

manuscript
Adj. R2 7.61
Panel C: PRE Correlations
Variable (1) (2) (3) (4)
(1) PREReported 1.0000 0.6870 0.2945 0.2018
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(2) PREEstimated 0.5729 1.0000 0.1475 0.6174
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(3) PREReported/PI 0.6183 0.3442 1.0000 0.3374
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(4) PREEstimated/PI 0.4109 0.9157 0.4429 1.0000
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
The results reported in Panel B are based on an OLS regression with the dependent variable equal to
the PRE estimation error (PRE/PIEstimationError). We report one-tailed p-values in Panel B where
predictions are made. PRE/PIEstimationError equals the absolute value of the difference between
PREReported/PI and PREEstimated/PI. See Appendix A for variable definitions.

42
TABLE 3
Analysis of TAX Estimates
Panel A: TAX Estimates and Estimation Errors
Disclosers (N= 751)
Variable Mean StdDev Q1 Median Q3
Estimated and Reported TAX
TAXReported 326.5 791.6 6.9 64.0 224.1
TAXEstimated 173.3 446.1 0.0 17.9 108.3
TAXReported/PI 0.0882 0.1380 0.0061 0.0355 0.1160
TAXEstimated/PI 0.0357 0.0496 0.0000 0.0140 0.0561
TAX estimation error
TAX/PIEstimationError 0.0526 0.1225 0.0000 0.0153 0.0637
Panel B: Determinants of TAX Estimation Error
Variable Pred. Coef. (p-value)
Intercept ? 0.0174 (0.7888)
TAX=0: Repatriation + 0.0738 (0.0557)
TAX=0: No ETR Rec Item + 0.0116 (0.1760)
Other Repatriation + 0.0076 (0.3857)
FETR Volatility + 0.0757 (<0.0001)
LnSales ? -0.0035 (0.6442)
Segments ? -0.0004 (0.6477)
Haven ? 0.0395 (0.0106)

preprint
#Exhibit21Countries ? 0.0041 (0.6932)
Adj. R2 10.49
Panel C: TAX Correlations

accepted
Variable (1) (2) (3) (4)
(1) TAXReported 1.0000 0.8722 0.4382 0.4783

manuscript
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(2) TAXEstimated 0.5868 1.0000 0.2877 0.5028
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(3) TAXReported/PI 0.8455 0.4612 1.0000 0.4365
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
(4) TAXEstimated/PI 0.4870 0.8776 0.5412 1.0000
(<0.0001) (<0.0001) (<0.0001) (<0.0001)
The results reported in Panel B are based on an OLS regression with the dependent variable equal to
the TAX estimation error (TAX/PIEstimationError). We report one-tailed p-values in Panel B where
predictions are made. TAX/PIEstimationError equals the absolute value of the difference between
TAXReported/PI and TAXEstimated/PI. See Appendix A for variable definitions.

43
TABLE 4
PRE Descriptive Statistics by Year
Panel A: PRE Disclosure Statistics (Totals) by Year
Sample composition 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Total
Disclosure firm-years 182 191 189 200 216 261 265 270 282 276 276 287 2,895
Non-disclosure firm-years 27 34 38 40 39 31 34 36 31 37 40 38 425
Total firm-years 209 225 227 240 255 292 299 306 313 313 316 325 3,320

Non-disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean
% of sample 12.9% 15.1% 16.7% 16.7% 15.3% 10.6% 11.4% 11.8% 9.9% 11.8% 12.7% 11.7% 12.8%
PREEstimated ($ in billions) 5.8 5.2 29.3 28.7 28.4 29.9 49.3 49.6 30.9 51.9 89.1 99.0 41.4

Disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean
% of sample 87.1% 84.9% 83.3% 83.3% 84.7% 89.4% 88.6% 88.2% 90.1% 88.2% 87.3% 88.3% 87.2%
PREReported ($ in billions) 202.0 243.5 267.6 306.5 372.8 453.2 391.9 513.7 666.4 743.7 833.6 994.9 499.1
PREEstimated ($ in billions) 110.7 131.1 133.9 161.0 189.7 258.1 271.6 373.3 446.3 529.8 519.7 616.8 311.8
PREReported/PREEstimated (%) 182.5% 185.8% 199.8% 190.4% 196.6% 175.6% 144.3% 137.6% 149.3% 140.4% 160.4% 161.3% 160.1%

Panel B: PRE Disclosure Statistics (Averages) by Year


preprint
Non-disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean

accepted
PREEstimated ($ in billions) 0.21 0.15 0.77 0.72 0.73 0.97 1.45 1.38 1.00 1.40 2.23 2.61 1.17
PREEstimated/PI 0.16 0.12 0.28 0.24 0.19 0.36 0.46 0.34 0.20 0.25 0.43 0.46 0.30

Disclosers 1999 2000 2001 2002 manuscript


2003 2004 2005 2006 2007 2008 2009 2010 Mean
PREReported ($ in billions) 1.11 1.27 1.42 1.53 1.73 1.74 1.48 1.90 2.36 2.69 3.02 3.47 2.07
PREEstimated ($ in billions) 0.61 0.69 0.71 0.80 0.88 0.99 1.02 1.38 1.58 1.92 1.88 2.15 1.29
PREReported/PI 0.39 0.44 0.49 0.55 0.66 0.60 0.43 0.46 0.49 0.52 0.61 0.70 0.53
PREEstimated/PI 0.18 0.19 0.18 0.18 0.25 0.26 0.23 0.25 0.27 0.34 0.35 0.39 0.27
See Appendix A for variable definitions. Measures based on reported values of PRE, i.e., PREReported and PREReported/PI, are not reported for non-disclosers because,
by construction, these values are zero.

44
TABLE 5
TAX Descriptive Statistics by Year
Panel A: TAX Disclosure Statistics (Totals) by Year
Sample composition 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Total
Disclosure firm-years 56 57 55 58 59 83 75 72 64 59 58 55 751
Non-disclosure firm-years 153 168 172 182 196 209 224 234 249 254 258 270 2,569
Total firm-years 209 225 227 240 255 292 299 306 313 313 316 325 3,320

Non-disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean
% of sample 73.2% 74.7% 75.8% 75.8% 76.9% 71.6% 74.9% 76.5% 79.6% 81.2% 81.6% 83.1% 77.4%
TAXEstimated ($ in billions) 10.20 16.36 20.16 24.41 26.50 32.36 36.85 46.57 57.24 66.65 67.36 76.88 40.13

Disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean
% of sample 26.8% 25.3% 24.2% 24.2% 23.1% 28.4% 25.1% 23.5% 20.4% 18.8% 18.4% 16.9% 22.6%
TAXReported ($ in billions) 7.55 8.43 8.59 8.12 13.52 16.62 14.61 21.73 26.31 31.26 42.33 46.11 20.43
TAXEstimated ($ in billions)
TAXReported/TAXEstimated (%)
3.07
246.1%
2.99
281.6%
3.12
275.6%
3.90
208.2%
preprint
6.17
219.2%
10.28
161.7%
10.41
140.3%
12.83
169.3%
14.03
187.5%
16.92
184.8%
20.93
202.3%
25.52
180.7%
10.85
188.4%

accepted
Panel B: TAX Disclosure Statistics (Averages) by Year
Non-disclosers 1999 2000 2001 2002 2003
2004 2005 2006 2007 2008 2009 2010 Mean
TAXEstimated ($ in billions)
TAXEstimated/PI
0.07
0.02
0.10
0.03
0.12
0.03
0.13
0.03 manuscript
0.14
0.15
0.04
0.05
0.16
0.05
0.20
0.05
0.23
0.05
0.26
0.05
0.26
0.05
0.28
0.06
0.19
0.04

Disclosers 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean
TAXReported ($ in billions) 0.13 0.15 0.16 0.14 0.23 0.20 0.19 0.30 0.41 0.53 0.73 0.84 0.33
TAXEstimated ($ in billions) 0.05 0.05 0.06 0.07 0.10 0.12 0.14 0.18 0.22 0.29 0.36 0.46 0.17
TAXReported/PI 0.05 0.05 0.08 0.07 0.11 0.10 0.07 0.07 0.09 0.10 0.12 0.14 0.09
TAXEstimated/PI 0.02 0.02 0.02 0.02 0.03 0.03 0.03 0.04 0.04 0.05 0.06 0.07 0.04
See Appendix A for variable definitions. Measures based on reported values of TAX, i.e., TAXReported and TAXReported/PI, are not reported for non-disclosers because,
by construction, these values are zero.

45
TABLE 6
Descriptive Statistics partitioned by Disclosure Status
Panel A: Partitioned by PRE Disclosure Status
Disclosers (N= 2,598) Non-Disclosers (N= 379)
Variable Mean StdDev Q1 Median Q3 Mean StdDev Q1 Median Q3
PREEstimated/PI 0.2759 0.4975 0.0000 0.0994 0.3563 0.3140 0.7313 0.0000 0.0569 ** 0.2519
TAXEstimated/PI 0.0430 0.0561 0.0012 0.0201 0.0643 0.0312 *** 0.0406 0.0017 0.0175 ** 0.0421
E/P 0.0285 0.1097 0.0263 0.0463 0.0629 0.0285 0.1046 0.0219 0.0460 0.0670
Leverage 0.2120 0.1774 0.0774 0.2010 0.3260 0.2527 *** 0.1652 0.1288 0.2552 *** 0.3644
PROA 0.0947 0.0906 0.0412 0.0928 0.1490 0.0829 ** 0.0962 0.0390 0.0806 ** 0.1300
NAnalyst 15.3468 7.4994 10.0000 14.0000 20.0000 14.7968 7.2556 9.0000 14.0000 20.0000
Inst 0.7686 0.1448 0.6820 0.7876 0.8726 0.7784 0.1470 0.6949 0.7907 0.8891
LnSales 8.7236 1.2009 7.8282 8.6173 9.5037 8.4953 *** 1.2232 7.6192 8.4914 ** 9.4278

preprint
Foreign% 0.4281 0.7409 0.1655 0.3991 0.6700 0.5406 *** 0.7548 0.1270 0.4548 * 0.7558
HHI 15.1705 14.0402 6.0000 10.0000 19.0000 14.6095 12.3740 6.0000 10.0000 18.0000
Segments 12.3422 7.1939 8.0000 12.0000 15.0000 13.7177 *** 8.4467 9.0000 12.0000 *** 16.0000
AbsDiscAccr 0.0440 0.0442 0.0144 0.0311
accepted
0.0580 0.0495 ** 0.0500 0.0177 0.0342 ** 0.0641
***, **, * indicate significant differences in means or medians between disclosers and non-disclosers at the one percent, five percent, and ten percent levels,

manuscript
respectively. Continuous variables are winsorized at the first and 99th percentiles. See Appendix A for variable definitions.

46
TABLE 6 (continued)
Descriptive Statistics partitioned by Disclosure Status
Panel B: Partitioned by TAX Disclosure Status
Disclosers (N= 618) Non-Disclosers (N= 1,980)
Variable Mean StdDev Q1 Median Q3 Mean StdDev Q1 Median Q3
PREEstimated/PI 0.2007 0.3318 0.0000 0.0769 0.2890 0.2994 *** 0.5368 0.0000 0.1102 *** 0.3792
TAXEstimated/PI 0.0372 0.0505 0.0000 0.0152 0.0576 0.0448 *** 0.0577 0.0019 0.0219 *** 0.0658
E/P 0.0320 0.1028 0.0260 0.0453 0.0616 0.0274 0.1118 0.0263 0.0467 0.0632
Leverage 0.1844 0.1778 0.0300 0.1734 0.2928 0.2207 *** 0.1764 0.0925 0.2083 *** 0.3402
PROA 0.1069 0.0983 0.0405 0.1038 0.1688 0.0908 *** 0.0878 0.0414 0.0890 *** 0.1427
NAnalyst 15.9757 7.9466 10.0000 15.0000 21.0000 15.1505 ** 7.3453 10.0000 14.0000 * 20.0000
Inst 0.7559 0.1468 0.6843 0.7824 0.8590 0.7726 ** 0.1440 0.6812 0.7887 * 0.8769
LnSales 8.5699 1.1123 7.6834 8.4299 9.3370 8.7715 *** 1.2235 7.8763 8.6627 *** 9.5531
Foreign% 0.4145 0.7194 0.1540 0.3976 0.6830 0.4323 0.7476 0.1688 0.3995 0.6659
HHI
Segments
16.1553
13.0146
14.2914
7.4380
6.0000
8.0000
11.0000
12.0000
24.0000
17.0000
preprint
14.8631 **
12.1323 ***
13.9502
7.1048
5.0000
8.0000
10.0000
11.0000
**
***
19.0000
15.0000
AbsDiscAccr 0.0478 0.0481 0.0145 0.0327 0.0663 0.0427 ** 0.0429 0.0143 0.0306 * 0.0564
accepted
***, **, * indicate significant differences in means or medians between disclosers and non-disclosers at the one percent, five percent, and ten percent
levels, respectively. Continuous variables are winsorized at the first and 99th percentiles. See Appendix A for variable definitions.
manuscript

47
TABLE 7
Determinants of Disclosure
Disclosureit = β0 + β1PREEstimated/PIit + β2TAXEstimated/PIit + β3E/Pit + β4Leverageit + β5PROAit + β6NAnalystit +
β7Instit + β8LnSalesit + β9Foreign%it + β10HHIit + β11Segmentsit + β12AbsDiscAccrit + γkIndustryk + αtYeart + εit
(1) (2)
PRE Disclosure TAX Disclosure
Coefficient dy/dx Coefficient dy/dx
Variable Pred. (p-value) Pred. (p-value)
Intercept ? 1.8000 ? 1.0871
(0.2030) (0.3880)
PREEstimated/PI ? -0.5255 *** -0.0242 - -0.6935 ** -0.0587
(0.0030) (0.0200)
TAXEstimated/PI ? 6.9192 *** 0.0326 - 0.7295 0.0070
(0.0060) (0.3780)
E/P ? -1.1870 -0.0112 ? -0.2040 -0.0038
(0.2050) (0.7590)
Leverage - -1.6763 ** -0.0255 - -0.9331 -0.0282
(0.0130) (0.1105)
PROA + 1.1075 0.0087 + 2.2857 ** 0.0353
(0.2235) (0.0345)
NAnalyst + -0.0016 -0.0010 + 0.0154 0.0196
(0.4630) (0.1820)
Inst ? -0.9671 -0.0121 ? -1.7293 ** -0.0426
(0.2120) (0.0220)
LnSales ? 0.1315 0.0137 ? -0.2319 ** -0.0474
(0.3200) (0.0290)

preprint
Foreign% - -0.2380 *** -0.0153 - -0.0225 -0.0028
(0.0050) (0.4005)
HHI + 0.0091 0.0109 + 0.0069 0.0165
(0.2435) (0.2090)
Segments - -0.0219 * -0.0139 ? 0.0164 0.0201

AbsDiscAccr -
(0.0965)
accepted
-3.6693 *** -0.0143 -
(0.3330)
1.9613 0.0148
(0.0085)
manuscript (0.9215)

Industry Fixed Effects? Yes Yes


Year Fixed Effects? Yes Yes
N 2,977 2,598
Pseudo-R2 10.19 7.28
% Correctly Predicted 72.9% 68.7%
Area under the ROC curve 0.731 0.690
***, **, * indicate significance at the one percent, five percent, and ten percent levels, respectively. Huber-White
robust standard errors are clustered by firm and are used to control for heteroscedasticity and serial correlation.
When predictions are made, p-values are one-tailed. Continuous variables are winsorized at the first and 99th
percentiles. Marginal effects (dy/dx) reflect the effect of a one standard deviation change across the mean of the
variable, from 1/2 standard deviation below the mean to 1/2 standard deviation above the mean or from zero to one
for categorical variables, with all other variables held at their means. Variables are defined in Appendix A.

48
TABLE 8
The Impact of the AJCA on PRE and TAX Disclosure
Panel A: Full Sample
PRE Disclosure TAX Disclosure
Pre-AJCA AJCA Post-AJCA Pre-AJCA AJCA Post-AJCA
(1999-2003) (2004-2005) (2006-2010) (1999-2003) (2004-2005) (2006-2010)
Avg. Coefficient on Year Fixed Effects 1.6379 2.0054 1.9889 1.2277 1.4445 1.2248

Test for increase in disclosure from Pre-AJCA to: Pred. Difference (p-value) Pred. Difference (p-value)
AJCA; test: AJCA – Pre-AJCA = 0 + 0.3675 (0.0267) ? 0.2169 (0.2060)
Post-AJCA; test: Post-AJCA – Pre-AJCA = 0 + 0.3510 (0.0574) ? -0.0029 (0.9892)
Panel B: Repatriating Firms
PRE Disclosure TAX Disclosure
Pre-AJCA AJCA Post-AJCA Pre-AJCA AJCA Post-AJCA
(1999-2003) (2004-2005) (2006-2010) (1999-2003) (2004-2005) (2006-2010)
Avg. Coefficient on Year Fixed Effects 2.2044 2.9606 2.9534 4.2674 4.4978 4.5015

Test for increase in disclosure from Pre-AJCA to: Pred. Difference (p-value) Pred. Difference (p-value)
AJCA; test: AJCA – Pre-AJCA = 0
Post-AJCA; test: Post-AJCA – Pre-AJCA = 0
+
+
preprint
0.7563
0.7491
(0.0024)
(0.0061)
?
?
0.2304
0.2341
(0.2868)
(0.3930)
Panel C: Firms with Estimated PRE > $250 Million

accepted Pre-AJCA
PRE Disclosure TAX Disclosure
Pre-AJCA AJCA Post-AJCA AJCA Post-AJCA

Avg. Coefficient on Year Fixed Effects


(1999-2003)
8.3746 manuscript
(2004-2005)
9.1908
(2006-2010)
(1999-2003)
9.4623
3.4396
(2004-2005)
3.6595
(2006-2010)
3.5726

Test for increase in disclosure from Pre-AJCA to: Pred. Difference (p-value) Pred. Difference (p-value)
AJCA; test: AJCA – Pre-AJCA = 0 + 0.8162 (0.0230) ? 0.2199 (0.4241)
Post-AJCA; test: Post-AJCA – Pre-AJCA = 0 + 1.0878 (0.0085) ? 0.1330 (0.6945)
This table provides evidence on whether PRE and TAX disclosure increases from the pre-AJCA (1999 - 2003) period to the AJCA (2004 - 2005) and
post-AJCA (2006 - 2010) periods. Based on the results from estimating equation (1), we compute the average year fixed effect coefficients during the
pre-AJCA, AJCA, and post-AJCA periods. We then test for differences in the average coefficients from the pre-AJCA period to both the AJCA and post-
AJCA periods. We perform these tests using the full sample, presented in Panel A, a subsample of firms that repatriate during our sample, presented in
Panel B, and a subsample of firms with estimated PRE greater than $250 million, presented in Panel C. When predictions are made, p-values are one-
tailed.

49

You might also like