Voluntary disclosure responses to mandated disclosure: evidence from Australian corporate tax transparency

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Voluntary Disclosure Responses to Mandated Disclosure: Evidence
from Australian Corporate Tax Transparency

By
Allison Kays

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

May 2018

i
Dedication
This dissertation is dedicated to my husband, Ryan Martin for his unwavering support
over the past four years.

ii
Acknowledgements
I would like to thank Shane Heitzman for his dedication to helping me with this project
and all of his invaluable comments and suggestions. I would also like to thank my dissertation
committee: Eric Allen, Randy Beatty, Oguz Ozbas, and Mark Soliman, as well as Tyler
DeGroot, Tina Lang, Michele Mullaney, Stacey Ritter, and Satish Sahoo for their comments and
suggestions.

iii
Table of Contents
DEDICATION i

ACKNOWLEDGEMENTS ii

LIST OF TABLES v

LIST OF FIGURES vi

ABSTRACT vii

SECTION 1: Introduction 1

SECTION 2: Background Information on Public Tax Return Disclosure in Australia 7

SECTION 3: Literature Review 9
3.1 Costs and Benefits of Public Tax Return Disclosure 9
3.1.1 Reputation Costs 9
3.1.2 The Informative Value of Tax Information 12
3.1.3 The Cost of Analyzing Tax Disclosures 13
3.1.4 The Cost of Revealing Proprietary Information 18
3.2 The Effect of Mandatory Disclosure on Voluntary Disclosure 19
3.2.1 Theoretical Models on the Relationship between Mandatory and Voluntary
Disclosure 19
3.2.2 Empirical Literature on the Relationship between Mandatory and Voluntary
Disclosure 21

SECTION 4: Hypothesis Development 23
4.1 The Probability and Timing of Voluntary Disclosure 23
4.2 Market Reactions to Tax Return Disclosures with and without Supplementary
Information 28

SECTION 5: Empirical Design 30
5.1 Predicting New Voluntary Disclosures 30
5.2 Investor Reactions to Tax Return Disclosures and the Effect of Voluntary Disclosures
34

SECTION 6: Sample Selection and Descriptive Statistics 37
6.1 Sample Selection and Data Collection 37
6.2 Summary Statistics 39
iv

6.3 Fundamental Drivers of Tax Return – Financial Statement Differences 41

SECTION 7: Main Results 46
7.1 Voluntary Disclosure Responses to Public Tax Return Disclosure 46
7.2 Investor Reactions to Public Tax Return Disclosure 52

SECTION 8: Supplementary Analyses 55
8.1 Investor Reactions to Public Tax Return Disclosure Conditional on the Sign of Tax
Return Differences 55
8.2 Investor Reactions to Public Tax Return Disclosure in the Second Year of Disclosures
58
8.3 Tax Transparency Code Disclosures 62
8.3.1 Determinants of TTC Disclosures 64
8.3.2 Investor Reactions to TTC Disclosures 66

SECTION 9: Conclusion 67

REFERENCES 70

APPENDIX A: Variable Definitions 77

APPENDIX B: Supplemental Disclosure Examples 81

v

List of Tables
TABLE 1: Sample Selection 87

TABLE 2: Summary Statistics 88

TABLE 3: Distribution of Variables by Quintile of Tax Return Differences 89

TABLE 4: Predicting Tax Return Differences 90

TABLE 5: Tax Return Differences and Voluntary Disclosure 91

TABLE 6: Market Reactions to the First Year of Tax Return Disclosures, 2015 95

TABLE 7: Market Reactions to the First Year of Tax Return Disclosures Dependent on the Sign
of TL_Diff and the Timing of Voluntary Disclosures 98

TABLE 8: Market Reactions to the Second Year of Tax Return Disclosures, 2016 101

TABLE 9: TTC Summary Statistics 105

TABLE 10: Tax Transparency Code Disclosure 106

TABLE 11: Market Reactions to Tax Return Disclosure Conditional on Firm Issued TTC
Disclosures 107

vi

List of Figures

FIGURE 1: Voluntary Disclosure Dates in Relation to the ATO’s Tax Return Disclosures 85

FIGURE 2: Tax Return Effective Tax Rates 86

vii

Abstract

In this paper, I study how mandated disclosures influence firms’ voluntary disclosure
decisions and subsequently, how voluntary disclosures shape market reactions to mandatory
disclosures. Specifically, I examine Australian firms’ voluntary disclosure responses to the
involuntary disclosure of their tax returns. Tax return information is difficult to understand and
most difficult when it largely differs from a firm’s previously reported financial statements. As
the cost to understand a disclosure increases, theory predicts that the average precision of
investors’ beliefs decreases, which may prompt managers to issue clarifying information. I find
support for this theory as firms with large differences between their tax return and financial
statement values are more likely to issue voluntary disclosures that supplement their tax return
information. Further, my evidence suggests that voluntary disclosures shift investors’ attention
away from the difference between a firm’s tax return values and its financial statement values.

1

1. Introduction
Is a firm’s voluntary disclosure policy affected by mandatory disclosure requirements?
Despite large literatures on the determinants and consequences of both mandatory and voluntary
firm disclosures, relatively few studies examine the effect of one on the other (Beyer, Cohen, Lys
& Walther, 2010). When the firm issues a mandatory disclosure, the firm can voluntarily
supplement the disclosure with non-mandated information. Thus, the observed disclosure may be
a combination of both mandatory and voluntary disclosure which makes it difficult for
researchers to distinguish between the two (Heitzman, Wasley, & Zimmerman, 2010). In this
paper, I am able to side step this key issue by exploiting an empirical setting in which the
government involuntarily discloses proprietary firm level information. I then examine whether
and how this mandated and involuntary disclosure affects firms’ incentives for voluntary
disclosure.
In December 2015, the Australian Tax Office (ATO) initiated public disclosure of three
summary numbers from large corporations’ tax returns: total income, taxable income and income
taxes payable. A firm’s total income is a measure of gross receipts and is calculated under the
same rules as a firm’s financial statements (IFRS in Australia). These financial statements are
already publicly available so this number is unlikely to be news
1
. A firm’s taxable income and
income taxes payable, however, are calculated under Australian corporate income tax laws and
are not directly inferable from the firm’s financial statements (McGill & Outslay, 2004). As
such, these two numbers will be news in the sense that they were previously unobserved, and

1
Total income is line 6S on Australia’s Corporate Tax return and is similar to gross revenue. It includes gross
receipts, distributions from partnerships and trusts, interest income, rent income, dividends, and a few other
miscellaneous items as calculated under IFRS (the same as a firm’s financial statements). The ATO discloses tax
return information at least one if not two years after a firm’s tax year has ended. As such, the related financial
statements were filed long before the tax return disclosures are publicly available.
2

may cause investors to change their expectations of the amount, timing, and uncertainty of a
firm’s future cash flows.
Whether investors respond to this information will depend on whether these values are
materially different from what investors infer from existing reports and whether such differences
are incrementally informative. Prior researchers and investors have not historically had access to
a firm’s tax returns and instead have relied on a firm’s financial statement tax expense to
estimate its tax liability and taxable income. This research has found that a firm’s estimated
taxable income relative to its financial statement income is predictive of the firm’s future
earnings growth (Lev & Nissim, 2004) and the persistence of the firm’s earnings (Hanlon, 2005).
These studies point out that a firm’s tax expense is directly related to its taxable income and
although the laws that govern a firm’s taxable income are designed to raise revenues and
incentivize or deter certain political priorities (rather than calculate a value relevant summary
number), they are still based on a firm’s underlying economic transactions. Thus, it is reasonable
to believe that a firm’s actual taxable income and/or liability would only improve these
inferences.
Tax return values, however, are specific to a certain taxing jurisdiction and only include
certain subsidiaries of a consolidated entity (Graham & Mills, 2008). As such, tax return values
may not always be better sources of information on the full operations of a consolidated firm. On
the other hand, a firm’s financial statement tax expense includes accruals for uncertain tax
positions, excludes taxes related to discontinued operations, and allows for managerial
discretion. Thus, whether tax return values are incrementally informative relative to the
information already reported in the firm’s financial statements, should depend on the degree of
the difference as well as the attributes of the firm that created those differences.
3

Because the tax return disclosures in my setting are only partial and lack additional
important details, investors may find it costly to determine what drives any difference between a
firm’s tax return values and its financial statement values. Although some insight may be
available in a firm’s financial statements, such an analysis would require an investor to have a
thorough understanding of financial accounting for income taxes, Australia’s corporate tax laws,
and the specific structures and transactions of a given firm. If analysts, who are generally
considered the most sophisticated users of firm disclosures, find the cost of such knowledge
exceeds the benefit
2
, less sophisticated stakeholders likely do too. In such a case, tax return
disclosures may cause investors to become aware of how much they do not know. In fact,
anticipated investor uncertainty could explain why managers do not voluntarily disclose their tax
return information.
If managers expect the ATO’s tax return disclosure to negatively affect their information
environments, they should be more likely to voluntarily disclose supplemental information that
addresses the uncertainty and mitigates costly market reactions (Jung & Kwon, 1988). Managers
know the content and the approximate timing of the ATO’s disclosure which allows the firm to
anticipate the assumptions investors will make and the conclusions they may come to without
additional information. As such, managers can get ahead of the ATO and issue supplemental
information before or concurrent with the ATO disclosures to prevent an adverse shock to the
firm’s information environment. I expect larger deviations between a firm’s tax return values and
that implied by its financial statements (“tax return differences”) will cause larger shocks to a
firm’s information environment as such deviations may cause both increased uncertainty and a
larger revision in investors’ beliefs.

2
See Amir and Sougiannis (1999), Bratten, Gleason, Larocque and Mills (2017), Chen and Schoderbek (2000),
Erickson, Heitzman, and Zhang (2013), Plumlee (2003), and Weber (2009).
4

It is not clear, ex ante whether the sign of tax return differences will be correlated with
the sign of the news to investors. For instance, if a firm’s taxable income is lower than implied
by its financial statements this could be bad news, suggesting that a firm is less profitable than
previously thought. On the other hand, it could be good news if it reveals that a firm optimally
transferred more of its income to lower tax jurisdictions. Without additional information,
investors with varying abilities to process the new information may come to these two very
different conclusions about the same firm, which will lead to increased uncertainty. Thus, I
hypothesize that firms with large absolute tax return differences will be more likely to
voluntarily disclose supplemental tax return information.
Managers will only issue a disclosure if they expect the benefit will exceed the cost
(Verrecchia, 1983). Thus, I hypothesize that voluntary disclosures will mitigate the impact of the
ATO’s disclosures on investors’ beliefs. In other words, I anticipate that on average, the market’s
reaction to tax return information that is supplemented with voluntary disclosure will depend less
on tax return differences.
I use the first two years of public tax return disclosure in Australia to test which firms are
most likely to voluntarily issue supplemental tax disclosures and how these disclosures affect the
market’s reaction to a firm’s tax return information. To identify voluntary disclosures, I search
publicly listed Australian firms’ websites, stock exchange disclosures and annual reports for
press releases or reports that specifically reference the ATO’s tax return disclosures.
Approximately 10% of my sample issues such disclosures. I then use a probit model to test
whether firms with greater absolute differences between their tax return values and their
financial statement values are more likely to issue these voluntary disclosures. As predicted, I
find that firms with large differences between their tax return values (taxable income or tax
5

liability) and that implied by their financial statement tax expense are 11% more likely to issue
supplemental tax disclosures. Interestingly, I find that this result is concentrated in firms that
issue voluntary disclosures ahead of or concurrent with the ATO’s disclosure. In contrast, the
decision to disclose after the fact seems to be driven by the market’s reaction to the ATO’s
disclosure. This suggests that ex-ante voluntary disclosure is driven by anticipated market
reactions whereas ex-post voluntary disclosure is driven by actual market reactions.
Next, I examine whether investor reactions to the ATO’s disclosure differs when the firm
issues supplemental disclosures. I focus on unsigned returns and trading volume to capture
changes in investors’ beliefs. On the day of and the day after the ATO’s first and second public
tax return disclosure, unsigned returns and trading volume are positively associated with the
difference between a firm’s tax return values and its financial statement values. Specifically, in
the first year of disclosure, I find that firms in the top decile of tax return differences (the
absolute difference between a firm’s tax return tax liability and its current tax expense)
experience absolute returns that are 47 basis points higher and trading volume that is 0.05
percentage points higher than firms that are in the bottom decile. These represent increases of
27% and 29% over the market’s normal behavior in my sample period, respectively. This
suggests that tax return disclosures that largely differ from disclosures in a firm’s financial
statements provide information that changes investors’ beliefs.
As predicted, I find that voluntary disclosures help offset some of these market reactions.
On the day of and the day after the ATO’s first tax return disclosure, the market’s reaction to
firms that issue preemptive or concurrent voluntary disclosures is not positively associated with
the difference between the firm’s tax return values and its financial statement values. This
suggests that voluntary disclosures increased investor certainty (as there was less volatility) and
6

mitigated investors’ reactions to the difference between the firm’s tax return values and its
financial statement values.
The ATO’s tax return disclosure is a unique setting to study how a mandatory disclosure
affects a firm’s voluntary disclosure decisions because, in this case, the mandatory disclosure is
not issued by the firm but by a third party. When the firm is responsible for providing a
mandatory disclosure, they can supplement the required information in order to maximize
investors’ understanding of the disclosure. This is the behavior that I document, however,
inferences would be clouded in a setting in which the firm issues a new mandatory disclosure as
variation between firms in the amount and quality of the disclosure may be caused by both
voluntary disclosures as well as differences in materiality thresholds (Heitzman et al., 2010). In
my setting, in which the mandatory disclosure is outside of the firm’s control, I can cleanly
separate mandatory and voluntary disclosures while still speaking to our overall understanding of
the relationship between these two types of disclosure.
The results of this paper contribute to three strands of literature. First, this paper adds to
our understanding of the interdependencies of firm disclosures. Specifically, I provide new
evidence on mandatory disclosures effect on firms’ incentives to produce new information.
Guay, Samuels and Taylor (2017) find that firms that are mandatorily required to increase the
complexity of their financial statements (adopt SFAS 133 and/or SFAS 157) voluntarily issue
more earnings forecasts to help investors synthesize the new information. However, both Guay et
al. (2017) and Dyer, Lang, and Stice-Lawrence (2017) point out that it is difficult to disentangle
a casual chain from alternative explanations. My paper looks to compliment and build on their
results by studying a voluntary disclosure that specifically references the mandatory disclosure
so as to provide a more direct causal link.
7

Second, this paper contributes directly to the new stream of empirical evidence on the
costs and benefits of tax return disclosure. Prior work on tax return disclosure focuses on the
reputational costs that such disclosures often intend (Hoopes, Robinson, & Slemrod, 2017). In
this paper, I show that when tax return disclosures are partial and disclosed out of context, they
can shock a firm’s information environment. This may affect capital markets as well as
incentivize managers to issue additional disclosures. My study adds insight to our understanding
of partial tax return disclosure by showing that such disclosures shift some firms’ voluntary
disclosure equilibriums. As a result, beyond the three new tax return numbers, the total
information available to society increases.
Lastly, this paper contributes to literature on disclosure regulation. Leuz and Wysocki
(2016) point out that disclosure regulation first affects firm disclosures and these disclosures then
have economic consequences. Most studies ignore variation in firm disclosures and study the
relationship between a regulatory change and economic consequences. However, my study
illustrates that a uniform regulatory change can lead to differential changes in firm disclosures,
which leads to cross sectional differences in economic consequences. This highlights that the
link between a regulatory change and disclosure outcomes is important to consider when
studying the economic consequences of disclosure regulation.
2. Background Information on Public Tax Return Disclosure in Australia
In June 2013, the Australian Government mandated the Australian Tax Office (ATO) to
publicly disclose three line items from large corporations’ tax returns: total income, taxable
income, and income taxes payable. The new regulation only affected Australian public
8

corporations (listed and unlisted)
3
and all foreign corporations (public and private) that report
total income of $100 million
4
or more and Australian private corporations that report total
income of $200 million or more on their tax returns. The law only requires the ATO to publish
positive taxable incomes and/or liabilities. For loss firms, the ATO only publishes the firm’s total
income. Similarly, if a firm has positive taxable income but is owed a refund, the firm’s tax
payable is not disclosed. The ATO provides some background information on their website to
help users interpret the disclosures but notes that what can be learned from only three numbers is
limited and that “some entities may provide further context and explanation on their own
websites or in financial or tax reports”.
The first year of data was published on December 17, 2015 and related to the tax returns
of 1,538 large public corporations with tax years ending December 1, 2013-November 30, 2014.
Of this total, 553 were Australian public corporations (both listed and unlisted) and the
remaining 985 were subsidiaries of foreign corporations. On March 22, 2016 the ATO disclosed
the same information for 321 private firms. In the first year, the disclosure of private firms’ tax
return information was delayed due to a legislative hold up, however, in all future years, private
firms’ information will be disclosed at the same time as all other firms. The second year of data,
relating to the 2014/2015 tax year was published on December 9, 2016. This report disclosed tax
return information for 1,378 of the first-year firms as well as information on 83 new Australian
public corporations and 118 new foreign corporations
5
. These disclosures will continue once a
year for the foreseeable future.

3
In Australia, proprietary (private) companies can only issue a limited number of shares. An unlisted public firm can
issue an unlimited number of shares but is not listed on a stock exchange. An example is a co-operative.
4
For ease of notation, all $ in this paper refer to Australian dollars or AUD.
5
This information is provided on the Australian Tax Office’s website: https://www.ato.gov.au/Business/Large-
business/In-detail/Tax-transparency/Corporate-tax-transparency-report-for-the-2014-15-income-
year/?page=4#Population_overview
9

3. Literature Review
3.1 Costs and Benefits of Public Tax Return Disclosure
In many countries, including the US, tax return information is considered confidential
and is protected by the government. When the government initiates public disclosure of such
information, three outcomes may occur: (1) reputational considerations may discourage firms
from engaging in tax planning, (2) the informative value of the disclosure may cause capital
markets to react, and (3) awareness of corporate tax laws may affect the general electorate’s
opinion (Lenter, Slemrod, & Shackelford, 2003). Whether such outcomes are considered benefits
or costs depends on the differing objective functions of affected stakeholders. The government,
who wishes to maximize tax revenues, may benefit if firms decrease their tax planning.
Investors, who wish to maximize after-tax cash flows, may suffer for the same reasons.
Similarly, although the release of new firm information may improve the functioning of capital
markets it may also hurt firm value if proprietary information is revealed. Lastly, public tax
return disclosure could increase general awareness of corporate tax laws but it may also create
confusion and mislead uninformed citizens, thus the net impact of informing the electorate is an
open question (Lenter et al., 2003). Several empirical studies, discussed below, attempt to
quantify such outcomes to better inform debates.
3.1.1 Reputation costs. Prior research argues that public disclosure of tax return
information may constrain aggressive tax planning by forcing taxpayers to consider its impact on
public perceptions. To date, the evidence is mixed. After individual income tax data became
more accessible to the public, sole proprietorships in Norway increased their reported earnings
(Bo, Slemrod & Thoresen, 2015). While such a response may be due to individuals desiring to
appear wealthier to others, the result still suggests that public disclosure leads to less tax
10

avoidance. Similarly, after the public disclosure of corporate tax returns ended in Japan, public
corporations’ taxable incomes decreased (Hasegawa, Hoopes, Ishida & Slemrod, 2013). This
suggests that firms engaged in less tax planning when they were subject to public disclosure,
however, the results are not robust to various specifications. In Australia, Hoopes et al. (2017)
find that after the initiation of public tax return disclosure, the percentage of public firms that
paid no taxes, slightly increased. This is contrary to the prior studies and suggests that public tax
return disclosure encouraged some public firms to increase tax avoidance. Interestingly, the
result flips for private firms, as they were less likely to not pay taxes after the disclosure. This
leads Hoopes et al. (2017) to conclude that public firms experience relatively more pressure from
shareholders to reduce taxes while private firms experience relatively more pressure from the
public to pay taxes. The results in Hoopes et al. (2017) only apply to whether or not the firm
pays taxes. Conditional on paying some taxes, there is no evidence that public or private firms
change the amount of taxes they pay.
To provide more evidence on why firms did or did not change their tax planning, Hoopes
et al. (2017) examine changes in consumer sentiment surrounding public tax return disclosure.
When the government publicly discloses a firm’s tax returns, the firm may decrease its tax
planning if it is sensitive to reputational costs arising from the disclosure. Such costs could arise
if consumers decide to purchase their goods and services from a firm’s less tax aggressive
competitor. Based on a sample of 230 public firms with well-known brands in Australia, Hoopes
et al. (2017) find no significant changes in consumer sentiment surrounding their tax return
disclosures, even among firms with significant media coverage of their tax return information
and those that paid no Australian tax. This suggests that in the first year of Australia’s tax return
disclosures, public corporations experienced little to no reputational costs. This likely explains
11

why these firms did not materially alter their tax planning in response to the public disclosure.
On the other hand, when private firms’ tax returns were publicly disclosed, consumers’
perceptions of these firms were negatively altered (Hoopes et al., 2017). This is likely what
drove private firms to decrease their tax planning and shows that public tax return disclosure had
differential effects on public and private firms.
Although public tax return disclosure did not change consumers’ opinions of public
firms, investors in these firms may still negatively react if they expect the new information to
change public opinion and subsequently, motivate the public to lobby for corporate tax reform.
Investors did respond negatively when public tax return legislation was proposed in Australia,
which suggests that investors anticipated some cost to disclosure (Hoopes et al., 2017). In
particular, public firms that reported zero income tax expense incurred the most negative market
reactions. Further, when actual tax returns were disclosed, firms that reported positive income
tax expense but whose tax returns revealed a zero tax liability incurred the least positive market
reactions. This suggests that the market expects non-taxpaying firms to be differentially hurt by
public tax return disclosure. Firms that engage in the most tax planning are the most sensitive to
tax policy changes (Heitzman & Ogneva, 2017). Thus, this result is consistent with investors
anticipating tax policy changes (Hoopes et al., 2017).
On the other hand, over four key dates related to Australia’s tax return disclosure
legislation, Chen (2017) finds that firms most likely to be affected by the new law, experience a
net positive return. This suggests that the market expected a net benefit, rather than a cost to
disclosure. Chen (2017) also finds that on average, there is no market reaction when the ATO
disclosed actual tax return information (signed returns on average, are insignificant). She
12

concludes that partial tax return disclosure was a nonevent that did not provide new information
to the market.
3.1.2 The informative value of tax information. Public tax return disclosure may
provide new information to the market if it is incrementally informative about a firm’s tax
planning activities and/or its pretax profitability. Hanlon, Laplante, and Shevlin (2005) explain
that a firm’s true profitability is unobservable. Using a firm’s financial statements, shareholders
observe a proxy for a firm’s income calculated under US GAAP or IFRS accounting principles.
In those same financial statements, shareholders also observe a second proxy for a firm’s
income, current tax expense, which can be used to estimate a firm’s taxable income. Taxable
income is an alternative measure of profitability calculated using rules designed by a country’s
lawmakers. While these rules are designed for different purposes, such as to raise revenues
and/or encourage certain behaviors, they nevertheless reflect variation in a firm’s true, but
unobservable, income. To the extent that a firm’s financial accounting income is an imperfect
proxy, an alternative signal of firm profitability, such as taxable income, can be incrementally
informative about firm value (Einhorn, 2005).
Lev and Nissim (2004) are one of the first to argue that taxable income may be
informative of the quality of a firm’s pretax income. They argue that managers can use
discretionary accruals to manipulate book earnings (pretax income) but not taxable income.
Since managers have incentives to maximize book income and minimize taxable income the
relationship between these two measures may be informative of a firm’s true income. Indeed,
Lev and Nissim (2004) finds that firms with high taxable income relative to pretax income have
higher future earnings growth. Similarly, firms with large temporary differences between their
13

taxable income and their pretax income (which is more likely to be driven by accruals than
permanent differences) have less persistent earnings (Hanlon, 2005).
Although a firm’s taxable income may provide information on firm profitability, Thomas
and Zhang (2014) argue that investors’ primary source of information regarding firm profitability
is likely nontax information. As long as a firm provides sufficiently informative nontax
information, the market should negatively value taxable income as it is related to an expected
cash outflow, the firm’s tax liability. Prior studies estimate a firm’s taxable income with its
financial statement tax expense, which fundamentally, is a cost to the firm. As such, Thomas and
Zhang (2014) show that in properly specified tests (that control for a firm’s expected
profitability), tax expense only provides incremental information about future profitability if
other nontax measures do a poor job of measuring future profitability. For instance, a loss or a
large swing in pretax income is generally temporary and as such, less informative about a firm’s
future profitability. In these cases, where investors cannot rely on nontax values to predict a
firm’s future profitability, they instead rely on tax expense. Thus, although tax expense is always
informative of the firm’s cash outflow for taxes it only provides incremental information on a
firm’s pretax cash flows when other nontax measures of profitability are insufficient.
3.1.3 The cost of analyzing tax disclosures. While there is compelling evidence that tax
disclosures are informative of a firm’s future cash flows, several studies show that investors’
responses to tax disclosures appear incomplete and can take several months to impound into
stock price. For instance, although Lev and Nissim (2004) find that the ratio of a firm’s taxable
income to its book income predicts a firm’s earnings growth, they also find that investors do not
fully incorporate this information into stock prices. Investors appear to be aware that this ratio is
informative of earnings growth as the ratio is associated with current price. However, the ratio is
14

also predictive of future returns, which suggests that the available information is not fully
incorporated into stock price. Similarly, Hanlon (2005) finds evidence that investors correctly
anticipate that firms with large temporary book-tax differences have less persistent accruals.
However, they overestimate the effect of large temporary book-tax differences on the persistence
of a firm’s cash flows. Overall, investors seem to be aware that book-tax differences are
informative of the persistence of a firm’s pre-tax earnings; however, they have not fully
incorporated this information into firm price.
Thomas and Zhang (2011) show that a firm’s quarterly tax expense is also related to
future returns. They show that a firm’s quarterly tax expense is predictive of both next quarter’s
tax expense and next quarter’s pre-tax income. The market, however, does not fully appreciate
these predictable relationships and only incorporates the information into stock price when the
persistence is later revealed in the next quarter’s earnings. As such, the information is
incorporated with a delay. This suggests one of two things: (1) investors do not fully understand
a firm’s tax disclosures, or (2) tax expense (and/or taxable income) is related to a risk factor that
investors are compensated for with higher future returns.
To disentangle these two possibilities, I turn to research on analysts. Income tax
accounting rules are very complex. If the effect on firm value is relatively small, analysts (and
shareholders) may find that the marginal cost of incorporating information from income tax
accounting into their analyses exceeds the benefit of obtaining more precise estimates of a firm’s
future cash flows. Although researchers cannot directly measure investors’ or analysts’ cost to
understand tax information, the consistent evidence that analysts fail to fully incorporate tax
information into their analyses indirectly reveals that this cost must be high. In such a case, the
15

association between taxable income and future returns is likely due to incomplete pricing rather
than risk.
Chen and Schoderbek (2000) show that analysts do not incorporate predictable effects of
a known corporate tax rate change into their earnings forecasts. When tax rates change, the book
value of a firm’s deferred tax assets and liabilities change, which affects book income. This is a
onetime change in earnings that will not persist; however, Chen and Schoderbek (2000) find that
investors value the change in earnings as if it will. Amir and Sougiannis (1999) find that
although analysts correctly value the future benefit of tax loss carryforwards, they overestimate
these firms’ future earnings as they fail to anticipate that firms with tax losses are more likely to
remain unprofitable. Similarly, Weber (2009) shows although book tax differences predict future
earnings growth, analysts fail to incorporate such information into their earnings forecasts and
investors fail to incorporate it into their valuation of a firm.
Erickson et al. (2013) paint a slightly more nuanced picture as they find that analysts fail
to incorporate predictable tax information into their forecasts but also that analysts seem to learn
from their mistakes overtime. They study managers’ incentives to carryback tax losses. When a
firm incurs a tax loss, they can carryback that loss to offset taxable income from prior years and
receive a refund for taxes previously paid. Tax losses can only be carried back a certain number
of years and otherwise, are carried forward. Due to the time value of money, managers have
incentives to maximize tax losses in years in which they can carryback such losses. To the extent
that taxable income is tied to a firm’s book income, a firm’s book income will be predictably
lower in these years. Erickson et al. (2013) find that analysts fail to incorporate these predictable
effects; however, they also find that analysts seem to understand that such losses will not persist
and correctly, do not incorporate them into future forecasts.
16

Although the above studies consistently show that analysts ignore available tax
information, it is unclear if this reflects inefficient investment by analysts or the deliberate
outcome of a cost- benefit analysis. There is evidence consistent with the latter. Plumlee (2003)
finds that after the Tax Reform Act of 1986, analysts incorporated less complex tax law changes
into their earnings forecasts but did not incorporate the more complex tax law changes.
Similarly, Bratten et al. (2017) find that analysts’ effective tax rate forecasts are less accurate and
more disperse when a firm has more complex tax reporting. These last two results provide
stronger evidence that analysts intentionally ignore more complex tax information because of
higher perceived costs and indirectly suggests that investors likely do the same.
Tax disclosures may be especially difficult to understand when a firm engages in tax
planning. Tax planning often requires complex corporate structures that may make it harder for
investors or analysts to understand a firm’s operations (Desai & Dharmapala, 2006). Further,
managers who engage in tax planning may purposefully provide opaque tax disclosures to hide
information from tax authorities (Bozanic, Hoopes, Thornock, & Williams, 2017). In line with
these expectations, Balakrishnan, Blouin and Guay (2017) find that analysts of firms that engage
in more tax planning than their peers have larger forecast errors and more forecast dispersion.
They also find that the adverse selection component of these firms’ bid-ask spreads is larger.
On the other hand, Gallemore and Labro (2015) challenge the idea that tax planning
requires opacity. They argue that in order to successfully identify, coordinate, implement and
defend complex corporate structures, a firm must have a strong internal information
environment. Indeed, their evidence suggests that firms with stronger internal information
environments (as proxied by faster filing speeds, more accurate management forecasts, and the
absence of restatements or material internal control weaknesses) have consistently lower cash
17

effective tax rates. They also find that the tax planning of these firms is less risky and more
sustainable which they conjecture is because the firm’s tax positions are supported by the firm’s
strong information environment.
While on the surface these studies may appear to be in direct contrast with each other,
Balakrishnan et al. (2017) argue that even if firms that successfully engage in tax planning have
high quality internal information, it may still be difficult to communicate tax planning structures
to external stakeholders. Balakrishnan et al. (2017) argue that managers recognize this issue and
attempt to offset some of their transparency problems with more disclosures. They find that firms
with abnormally low GAAP and cash effective tax rates tend to have longer manager discussion
and analysis reports and longer conference calls, both of which have more tax related
discussions. They find, however, that this additional information has limited success in
mitigating the transparency issues caused by tax planning.
Interestingly, in a concurrent working paper, Demeré (2018) finds evidence that tax
return information can help reduce the cost of understanding financial statement tax disclosures.
In his setting, Demeré (2018) does not have access to firms’ tax return information. Instead, he
relies on the syndicated loan market to test whether institutional investors who participate in the
secondary loan market use private loan information (such as a firm’s tax return) when trading in
a firm’s equity. He finds that after a firm issues a syndicated loan that is traded on the secondary
loan market, several of the tax related market anomalies discussed above, are arbitraged away.
This suggests that information included within a firm’s full tax return decreases the cost of
understanding a firm’s financial statement tax information such that previously known
information is now properly impounded into price.

18

3.1.4 The cost of revealing proprietary information. Although there is evidence that
full tax return disclosure may improve firms’ information environments, it comes with the cost
of revealing affected firms’ private and sensitive data (Lenter et al., 2003). In Demeré (2018)’s
setting in which tax returns are only disclosed to members of a syndicated loan, disclosure costs
are minimized as no information is disseminated to tax authorities or the firm’s competitors.
Further, the information is only disclosed to highly sophisticated investors which reduces the
chance of confusion. When tax return information is publicly disclosed, however, none of this is
true which may impose costs on affected firms.
To avoid such costs, firms may manipulate measures that determine disclosure if
disclosure thresholds are known ex-ante (Gao, Wu, & Zimmerman, 2009). In Japan, both
individuals and corporations near disclosure thresholds manipulated their taxable incomes
downward to avoid public tax return disclosure (Hasegawa et al., 2013). This behavior suggests
that that both individuals and firm managers anticipated personal costs to public tax return
disclosure. Hoopes et al. (2017) provide similar evidence from Australia and document that the
effect is strongest for private and foreign owned firms. This is consistent with the idea that
private firms experience both higher proprietary costs and higher reputational costs to disclosure,
and as such are more likely to actively manage their reporting in order to avoid disclosure.
Firms that cannot avoid disclosure because they are too far from the disclosure threshold
may attempt to mitigate proprietary costs by omitting or obfuscating important information in
their tax returns. This would dilute the information content of their tax returns, which may inhibit
tax enforcement and reduce the informative value of tax return disclosure. Lenter et al. (2003)
argue that a policy to disclose only summary line items from corporate tax returns would be less
costly and potentially beneficial to society as such numbers would not reveal proprietary firm
19

information but could still improve the flow of information to society. If a corporation felt that
its summary number(s) was not reflective of its true economic position, the corporation could
always clarify by voluntarily disclosing more information (Lenter et al., 2003). Hoopes et al.
(2017) do a limited analysis of ten firms and note that while there is some evidence of new
voluntary disclosures, it is very rare for firms to react to mandatory tax return disclosure with
voluntary disclosure.
3.2 The Effect of Mandatory Disclosure on Voluntary Disclosure
My paper dives deeper into this relationship between involuntary tax return disclosures
and firm reporting. As such, I next review the theoretical and empirical literature that links
mandatory disclosure policies with voluntary disclosure decisions.
3.2.1 Theoretical models on the relationship between mandatory and voluntary
disclosure. If disclosure avenues are substitutes, intuition suggests that an increase in required
disclosures will lead to a decrease in a firm’s voluntary disclosures. However, contrary to this
intuition, theoretical work suggests that an increase in mandatory disclosures may actually lead
to an increase in a firm’s voluntary disclosures (Bagnoli & Watts, 2007; Dye, 1986; Einhorn,
2005). Although the theory models vary in their approach, they all conclude that mandatory and
voluntary disclosures are compliments when the information in the two disclosures are correlated
or in some way interdependent. For example, in a simplified model in which a firm has only two
signals regarding firm value, they have four disclosure options: disclose nothing, disclose only
one signal, disclose only the other signal, or disclose both signals. When no disclosures are
required, a manager will only voluntarily disclose both signals if it is preferable to all other
options. When disclosure of one signal is required, however, the firm is only left with two
disclosure options: disclose only the required signal or disclose both signals. In this setting, full
20

disclosure is more likely as it only has to be preferable to one other option, disclosure of just one
signal. Specifically, if the two signals are correlated, partial disclosure of only one signal will
change investors’ conditional expectation of the second signal. If the second signal is more
favorable than investors’ new conditional expectation, managers will disclose both signals. Thus,
the requirement to disclose one signal may shift a firm’s optimal disclosure strategy from no
disclosure to full disclosure (Dye, 1986).
Einhorn (2005) studies a similar two signal model and argues that when the signals are
correlated, managers are more likely to disclose the second signal when the mandatorily
disclosed first signal is noisy and therefore costly to interpret. In such a case, investors can use
the information in the second signal to filter out the noise in first signal. The second signal helps
investors discern the first signal’s valuation implications at a lower cost. Importantly, Einhorn
(2005) shows that when the two signals are correlated, both “good” and “bad” news mandatory
disclosures may be followed by new voluntary disclosures. This is because the disclosure of the
second signal has two effects: one, it updates investors’ interpretations of the previous signal and
two, it has its own independent implications for a firm’s future cash flows. When the first signal
is interpreted as good news, manager’s disclosure threshold is higher; the second signal has to be
even better news to be disclosed. However, since the two signals are correlated, the second signal
is already more likely to be favorable given that the first was favorable. When the second signal
is interpreted as bad news, manager’s disclosure threshold is lower and they are more likely to
voluntarily disclose their second signal. However, again, since the two signals are correlated, the
second signal is less likely to be favorable as well. Thus, depending on the strength of the two
opposing forces, both good and bad news mandatory disclosures may be followed by more good,
or less bad news voluntary disclosures, respectively.
21

Bagnoli and Watts (2007) also show that both “good” and “bad” news mandatory
disclosures may be followed by new voluntary disclosures. Their model differs from Einhorn
(2005) in that the manager’s private information is not a second signal of a firm’s future cash
flows but rather information on the precision of the mandatorily disclosed signal. With this extra
information, investors can determine the precision of their estimate of a firm’s future cash flows.
If a mandatorily disclosed signal is interpreted to be good news and managers privately know
that the signal has very little variance, managers will voluntarily disclose their private
information to heighten the impact of the good news. If a mandatorily disclosed signal is
interpreted to be bad news but managers privately know that the signal has a lot of variance,
managers will voluntarily disclose their private information to diminish the impact of the bad
news on firm value. Taken together, the theoretical literature shows that managers should
voluntarily supplement mandatory disclosures when additional information is needed for
investors to fully understand the cash flow implications of the mandatory disclosure and when
the benefits to the firm exceed their cost.
3.2.2 Empirical literature on the relationship between mandatory and voluntary
disclosure. Despite the conceptual and practical importance of understanding the interrelation
between mandatory and voluntary disclosure, opportunities for empirical inquiry are infrequent.
Guay et al. (2017) do not explicitly test any of the above theories but they do provide empirical
evidence on the effect of mandatory disclosures on voluntary disclosures. Guay et al. (2017)
argue that when mandatory disclosures become long, complex and too costly to analyze,
managers can provide more concise, easy to analyze voluntary disclosures as a substitute. Over
time, the increasing information required to be disclosed in a firm’s annual reports makes them
longer and more complex. If the increase in required information pushes the marginal cost of
22

processing that information above the marginal benefits of having the information, the quality of
a firm’s information environment will decrease. This creates an incentive for managers to
actively manage the firm’s information environment by issuing voluntary disclosures that are
less costly to understand but help recover some of the information lost in the complex mandatory
disclosure.
Consistent with their predictions, Guay et al. (2017) find that firms that are mandatorily
required to increase the complexity of their financial statements by adopting SFAS 133 and/or
SFAS 157, increase their voluntary disclosure of earnings forecasts. They also find that as firms’
10-Ks become longer and less readable over time, managers issue more earnings forecasts.
While earnings forecasts are forward looking and do not directly address uncertainty about a
firm’s past transactions as disclosed in their financial statements, Guay et al. (2017) explain that
earnings forecasts essentially synthesize the implications of the information in a firm’s 10-K.
Investors can then use this number as a substitute with which to predict a firm’s future cash
flows. Guay et al. (2017) also note that in line with Einhorn (2005) and Bagnoli and Watts
(2007), managers could instead issue voluntary disclosures that specifically address the obtuse
mandatory disclosures and in such a capacity, serve as a compliment.
Both Guay et al. (2017) and Dyer et al. (2017) point out that it is difficult to disentangle
their causal chain from alternative explanations. Empirically it is clear that there is a strong
correlation between the increase in the length and complexity of financial statements over time
and the issuance of management forecasts. What is unclear is if the mandatory disclosures
actually induce a change in voluntary disclosures. For instance, it is plausible that over time,
technology is improving which decreases the cost to compile and distribute lengthy and complex
23

disclosures as well as the cost to forecast earnings more frequently (Dyer et al., 2017). As such,
the evidence on the relationship between mandatory and voluntary disclosures is inconclusive.
Bischof and Daske (2013) also study firms’ voluntary disclosure responses to a
mandatory disclosure shock. For a limited time during the Eurozone’s sovereign debt crisis,
European banking regulators disclosed certain firms’ stress test rankings. Bischof and Daske
(2013) find that, after the onetime mandatory disclosure, firms experienced a decrease in market
liquidity. However, firms that voluntarily continued to disclose their stress test ratings did not
experience such a decrease. This suggests that firms’ voluntary disclosure strategies can offset
negative effects from a change in mandatory disclosure. In this specific context, the banking
regulator made the disclosure but firms had to invest in costly technology to perform stress tests
on their assets and liabilities. This was a onetime investment that firms could use in the future to
calculate their risk exposures. Thus, the onetime involuntary disclosure not only increased the
benefit of voluntarily continuing the disclosure (to maintain continuity and liquidity in the firm’s
information environment) but also lowered the direct costs of subsequent disclosures. As such, it
is unclear if the increased benefit of voluntary disclosure would have been enough to induce a
change in the firm’s voluntary disclosure strategies.
4. Hypothesis Development
In this study, I build off and add to the relatively new empirical evidence on the
relationship between mandatory and voluntary disclosures by studying managers’ direct
voluntary disclosure responses to the involuntary disclosure of their tax returns.
4.1 The Probability and Timing of Voluntary Disclosure
Involuntary disclosure of tax return information will shock a firm’s information
environment if: (1) investors believe that tax return information is informative of the level or
24

riskiness of a firm’s future cash flows and (2) the newly released information is materially
different from investors’ prior beliefs. I assume that investors’ prior beliefs are formed using a
firm’s financial reporting for income taxes (Hanlon & Slemrod, 2009). As discussed earlier, prior
research shows that a firm’s financial statement tax expense can be informative of the level or
riskiness of a firm’s future cash flows (Hanlon, 2005; Hanlon et al., 2005; Lev & Nissim, 2004;
Thomas & Zhang, 2014). This research relies on the firm’s tax expense to estimate its underlying
tax liability and/or its taxable income. Thus, it is reasonable to expect that disclosure of a firm’s
actual tax liability and/or taxable income from its tax return would improve investors’ cash flow
forecasts, especially if tax return information is less affected by discretion and capital market
incentives (Dhaliwal, Gleason & Mills, 2004; Frank & Rego, 2006; Krull, 2004; Schrand &
Wong, 2003).
Tax returns, however, typically represent information from a finite period and are often
subsidiary and jurisdiction specific. These features may make the tax return values of a
consolidated, multinational firm less informative than the firm’s financial statement tax expense
since financial statement tax expense consolidates the income tax obligations of all of a firm’s
controlled subsidiaries as well as both its domestic and foreign operations (Graham & Mills,
2008). As a result, whether a single tax return disclosure is incrementally informative about an
entire firm’s cash flows, relative to the information already reported in the firm’s financial
statements, should depend on the degree of the difference as well as the attributes of the firm that
created those differences.
Because the tax return disclosures in my setting are only partial (at most three line items)
and lack additional important details, investors may have difficulty determining a) whether there
is a difference between a firm’s tax return values and its financial statement values, and b)
25

whether it is incrementally informative of firm value. Although a firm’s financial statements
provide some insight into the factors that drive this divergence, even sophisticated analysts do
not fully analyze or understand income tax disclosures (Amir & Sougiannis, 1999; Bratten et al.,
2017; Chen & Schoderbek, 2000; Erickson et al., 2013; Plumlee, 2003; Weber, 2009). Thus, I
expect that the typical investor will also fail to fully understand a firm’s tax return disclosures.
When the marginal cost to understand a disclosure exceeds the marginal benefit, investors’
beliefs will become less precise and investors’ forecasts of a firm’s future cash flows will
become more uncertain (Grossman & Stiglitz, 1980). This causes investors to underreact to new
information as well as discount their previous beliefs. The latter will materialize as a discount on
firm value (Lambert, Leuz, & Verrecchia, 2011). Such a result, triggered by the partial tax return
disclosure, creates an incentive for managers to voluntarily provide additional information that
addresses investor uncertainty (Jung & Kwon, 1988).
Turning to my setting, I expect that larger deviations between a firm’s tax return values
and its financial statement values will create a larger shock to a firm’s information environment.
As tax return values increasingly deviate from investors’ prior beliefs (proxied by the firm’s
financial statements), investors will be more likely to revise their beliefs and more likely to be
uncertain about these revisions. As a result, the net benefit of voluntarily providing additional
information increases. This leads to my first hypothesis:

H1A: The probability of voluntary disclosure is increasing in the absolute difference
between a firm’s tax return and financial statement values.

In my main analyses, I focus on unsigned difference, as it is not clear ex ante whether the
sign of the deviation between a firm’s financial statements and its tax returns should predict the
26

direction of the revision in investors’ beliefs. Even if a firm’s tax returns only provide
information about a firm’s tax planning (and not on profitability), tax planning still requires a
tradeoff between costs and benefits. Thus, without knowing a firm’s optimum level of tax
planning, it is not possible to predict whether investors will perceive a given deviation as good or
bad news. Since managers did not voluntarily release the firm’s tax return information before, I
expect that deviations in either direction will negatively affect their information environments
and affect value through increased uncertainty.
It is important to note that managers know in advance what their released tax return
numbers will be and approximately when the tax authority will disclose them. This disclosure is
outside of the firm’s periodic reporting. Thus, managers may choose to issue their voluntary
disclosures before, concurrent with, or after the involuntary disclosure. Unfortunately, economic
theory does not provide strong guidance on when managers should issue supplemental voluntary
disclosures (Guay et al., 2017). In some firms, managers may anticipate strong market reactions
and find it optimal to preempt the ATO’s disclosure to set the tone for investors. In other firms,
the market’s reaction may be uncertain such that managers find it optimal to first observe how
the market reacts to the ATO’s disclosure before deciding whether to issue their own
supplemental disclosure. I hypothesize that a manager’s decision to disclose before or with the
ATO’s disclosure will be driven by expected market reactions. I expect that mangers will use the
difference between a firm’s tax return values and its financial statement values (observed by
them) as an ex ante measure of expected market reactions. This leads to my next hypothesis:

H1B: The sensitivity of disclosure to the absolute difference between a firm’s tax return
and financial statement values is strongest for preemptive and contemporaneous
disclosures.
27

Of course, the voluntary disclosure decision can also follow, rather than precede the
involuntary disclosure. Managers may withhold private information when they are unsure of how
the market will react to such information (Dutta & Trueman, 2002; Fishman & Hagerty, 2003;
Suijs, 2007). Indeed, this may partially explain why managers have not voluntarily disclosed
their tax return information. After public tax return disclosure, managers learn how the market
values the firm’s tax information and can respond accordingly (Luo, 2005; Zuo, 2016). If the
market reacts positively, managers may learn that investors respond favorably to their tax
information and may now release additional details to heighten this favorable response (Dutta &
Trueman, 2002). On the other hand, if the market reacts negatively, managers’ disclosure
threshold will decrease and they will be more likely to disclose information that they had
previously considered unfavorable but is now relatively favorable (Jung & Kwon, 1988). Thus,
managers may follow both favorable and unfavorable mandatory disclosures with supplementary
voluntary disclosures (Bagnoli & Watts, 2007; Einhorn, 2005). As such, I do not make a signed
prediction. I expect that managers of firms that experience larger market reactions to involuntary
tax return disclosure will be more likely to issue supplementary voluntary disclosures. This leads
to my next hypothesis:

H1C: The probability of ex-post voluntary disclosure is increasing in the market’s
reaction to a firm’s tax return information.

It is important to note that although the benefit of additional disclosures has increased,
managers may still not issue voluntary disclosures if the cost is high. Managers should only issue
voluntary disclosures if the expected benefit exceeds the expected cost. The cost of disclosure
includes both the direct cost of assembling and distributing information as well as indirect
28

proprietary costs (Verrecchia, 1983). Firm issued tax disclosures could have high proprietary
costs if they are informative to a firm’s competitors or to tax authorities. Prior research finds that
although the tax authority is privy to a wealth of private disclosures, they still rely on public
disclosures in directing who and what to audit (Bozanic et al., 2017; Mills, 1998). As such, firms
may be reluctant to provide a detailed reconciliation of their financial statements to their
Australian tax returns if they think the information would aid tax authorities (foreign or
domestic) in their audits.
4.2 Market Reactions to Tax Return Disclosures With and Without Supplementary
Information
In this section, I turn the focus to the role of voluntary disclosures in shaping market
reactions to tax return information. In the absence of voluntary disclosures, I expect that the
market will react to tax return information if it differs from their expectations. As previously
discussed, I assume that investors’ prior expectations are based on a firm’s financial statement
tax expense. If a firm’s actual tax return values are similar to the firm’s financial statement tax
expense, tax return disclosure should simply confirm what investors already know. This may
reduce the uncertainty of investors’ cash flow forecasts but is unlikely to change the forecasts
themselves. However, as a firm’s actual tax return values diverge from their financial statement
values, investors are likely to become more uncertain of their beliefs. If investors interpret such a
divergence as an alternative signal of the nature or riskiness of a firm’s tax planning, the cost of
its actual tax liability, or the firm’s pretax profitability, they are likely to react to this divergence.
This leads to my next hypothesis:

H2: Investor reactions to a firm’s tax return information is increasing in the difference
between the firm’s tax return and financial statement values.
29

When a firm issues voluntary disclosures, however, I expect that the market will react
less to the difference between the firm’s tax return and financial statement values. H1B predicts
that firms issue preemptive or concurrent voluntary disclosures in anticipation of market
reactions to the difference between the firm’s tax return values and its financial statement values.
If this is true, I expect that managers issue these disclosures to prevent the anticipated market
reactions. Since managers are only likely to issue additional disclosures if they expect it to be
value enhancing, I expect that on average, voluntary disclosures will successfully mitigate
investors’ reactions to the difference between a firm’s tax return values and its financial
statement values. In other words, if two firms have the same difference between their tax return
and financial statement values, the firm that provides more supplementary information before or
with the ATO’s disclosure will experience a smaller shock. This leads to my last hypothesis:

H3: Preemptive and concurrent voluntary disclosures mitigate market reactions to the
difference between a firm’s tax return and financial statement values.

H3 focuses specifically on preemptive and concurrent disclosures as the market receives
the firm’s supplemental information ahead of or at the same time as the ATO’s tax return
information. This timing provides an opportunity to test market reactions to tax return
disclosures with supplemental information against market reactions to tax return disclosures
without supplemental information, at the same time. Although I expect ex-post voluntary
disclosure to also reduce market reactions to the difference between a firm’s financial statement
and tax return values, on the date of the ATO’s disclosure, ex-post disclosers’ additional
30

information is not yet available, as such I do not expect the market to react differently to their tax
return information.
6

5. Empirical Design
5.1 Predicting New Voluntary Disclosures
I examine which firms are most likely to issue voluntary disclosures by estimating the
following probit regression:

!"#$
%
= '
(
+'
*
|, -
.%//
%
0+'
1
|2, 3
456789
%
0+'
:
, 2
;<<=
%
+'
>
?",@ABC
%
+'
D
ln(2HHIJ)
%
+'
L
-32
%
+'
M
NIOIP@QI
%
+'
R
BBS
%
+'
(T
U,V
%
+W
%
(1)

where VolD is a measure of voluntary disclosure, |TR_Diff| is a measure of the difference
between a firm’s tax return values and its financial statement values, |ATO_Return| is a measure
of the market’s reaction to a firm’s tax return disclosure, and the rest are control variables.
VolD takes one of three forms. To test H1A, which predicts that voluntary disclosure, in
general, will be associated with larger values of |TR_Diff|, VolD_Any is an indicator variable
equal to one if a firm issues supplemental disclosures about their tax return information in the six
months before or after the ATO’s tax return disclosure and zero otherwise. To test H1B and
H1C, which predict differential drivers of preemptive and ex-post disclosures, I estimate two
additional regressions. In the first, VolD_Pre/Conc equals one if a firm issues supplemental
disclosures before or concurrent with the ATO’s tax return disclosure and zero if the firm never
issues supplemental disclosures related to that tax year. Given that a voluntary disclosure takes
time to draft and is likely subject to several layers of approval before it is issued, I expect that

6
Due to difficulty in obtaining precise event dates, I do not examine market reactions on the dates of voluntary
disclosure.
31

managers draft concurrent disclosures in anticipation of the ATO’s disclosure but time their
release to coincide with the ATO’s information release. This regression excludes firms that
disclose after the ATO’s disclosure to narrow in on which variables drive preemptive or
concurrent disclosure as opposed to no disclosure. In my third specification, VolD_Post equals
one if a firm issues supplemental disclosures after the ATO’s tax return disclosure and zero if the
firm never issues supplemental disclosures related to that tax year. This regression excludes
firms that disclose before or concurrent with the ATO’s disclosure to test whether ex-post
voluntary disclosure is driven by different factors than preemptive voluntary disclosure.
|TR_Diff| takes one of two forms and captures the difference between a firm’s previously
disclosed financial statements and its newly disclosed tax return values. The first specification,
|TI_Diff|, measures the absolute value of the difference between a firm’s actual taxable income
and a financial statement proxy for taxable income, scaled by assets. Similar to prior research,
my financial statement proxy for taxable income is a firm’s current tax expense grossed up by
the statutory rate (30% in Australia). The second version of |TR_Diff|, |TL_Diff| measures the
absolute value of the difference between a firm’s actual tax liability and a financial statement
proxy for a firm’s tax liability, scaled by assets. Again, I rely on a firm’s financial statement
current tax expense to proxy for its tax liability. My measures of |TR_Diff| are designed to
capture the shock caused by tax return disclosure, essentially “tax-tax” differences, rather than
previously known “book-tax differences”.
7

|ATO_Return| is the market’s reaction to the ATO’s tax return disclosures, measured as
the absolute value of a firm’s logarithmic cumulative return over the day of and the day after the

7
Book-tax difference is generally measured as the difference between a firm’s pre-tax financial statement income
(“book”) and its grossed up current tax expense (“tax”). My measures take this one step further by measuring how a
firm’s actual tax return values differ from the tax values suggested by its financial statements.
32

ATO’s tax return disclosure. I use the absolute value of ATO_Return and TR_Diff to capture the
magnitude of potential revisions to investors’ beliefs. I make no directional predictions and
instead focus on managers responses to any anticipated or actual changes in investors’ beliefs. In
additional tests, I examine whether voluntary disclosure is differentially driven by deviations in
one direction or the other.
In all specifications, I include a proxy for a firm’s tax planning to control for the
association between tax planning and information quality.
8
This may also control for proprietary
costs as the level and nature of a firm’s tax planning may be associated with the proprietary cost
of additional disclosures. I use Balakrishnan et al. (2017)’s measure of tax avoidance which
measures a firm’s effective tax rate relative to their industry and size peers. Firms with effective
tax rates that are significantly lower than their peers are considered to engage in more tax
avoidance. TA_GAAP is a firm’s three year average GAAP effective tax rate (the sum of a firm’s
total tax expense over the previous three years divided by the sum of a firm’s pretax income over
the previous three years) adjusted for the average three year GAAP effective tax rate of firms in
the same industry (as defined by GICS
9
) and of a similar size (same quintile of total assets).
10

The variable is specified such that positive values indicate a firm is more tax aggressive than
their peers and negative values indicate the opposite.
I also control for whether or not a firm owed any taxes in Australia. NoTaxPd is an
indicator variable that equals one if a firm had no tax liability or perhaps even a tax refund in

8
Tax return differences may be less of a surprise for firms with higher quality financial statement disclosures. Given
the opposing predictions in prior literature, it is not clear if firms that engage in more or less tax planning have
higher quality financial statements (Balakrishnan et al., 2017; Gallemore & Labro, 2015). Thus, the directional
effect of this control is unclear.
9
GICS stands for Global Industry Classification Standard. Standard & Poor’s and Morgan Stanley Capital
International created this classification system to aid in global investment analysis.
10
Due to the small size of my sample, independent sorts on GICS and size creates cells with fewer than 30
observations. For firms in these cells, I adjust their three-year GAAP effective tax rates for the average three-year
GAAP effective tax rate of firms in the same industry but of different sizes.
33

Australia and zero otherwise. These firms experience the most negative media attention and the
most negative market reactions (Hoopes et al., 2017). As such, I include this control to ensure
that |TR_Diff| is not driven by firms with no tax liability and to test if firms issue voluntary
disclosures in response to expected or actual negative media attention. I also control for size
(ln(asset)), profitability (ROA), leverage (Leverage), fixed assets (PPE), and growth
opportunities (MTB). All continuous variables in this model are decile ranked and scaled
between zero and one. This controls for outliers as well as any nonlinearities in the relationship.
My main specification relates to the first year of Australian tax return disclosure: the
2013/2014 tax year
11
. In additional tests, I examine whether firms behave similarly in the
following year of tax return disclosure. In the latter specification, I control for a firm’s previous
voluntary disclosure. If a firm issued supplemental disclosures related to the first tax return
disclosure, I expect they will be more likely to also issue supplemental disclosures related to
their second tax return disclosure in order to maintain continuity in their disclosure practices
(Bischof & Daske, 2013). VolD_PY equals one if a firm voluntarily issued a disclosure about the
first year’s tax return disclosure and zero otherwise.
In my second year specification, I specify |TR_Diff| four different ways. The first two are
the same as in the first year, as described above. The second two specifications are adjusted to
include the possibility that investors learned from the first year of tax return disclosure. When
investors observe a firm’s first year tax return disclosure, they learn the relationship between a
firm’s tax return values and its financial statement values. Given that tax planning is generally a
long-term investment and structural differences between a firm’s financial statement reporting
and tax return reporting are likely to be similar from year to year, a firm’s |TR_Diff| is likely

11
This includes firm years ending December 1, 2013-November 30, 2014.
34

proportionately similar from year to year. Investors may use this information to update their
expectations of a firm’s second year tax return values. Managers may also anticipate investors’
changed expectations and update their disclosure strategies. In my second year specification,
|TR_Diff| is one of two additional values defined below:
Variable Definition
|TI_Diff2| |((Tax Return Taxable Income- (Current Tax Exp/.3))*(PY Tax
Return Taxable Income/(PY Current Tax Exp/.3))/Assets|
|TL_Diff2| |(Tax Return Tax Liability- (Current Tax Expense*(PY Tax
Return Tax Liability/PY Current Tax Expense)))/Assets|

These definitions are the same as in the first year except that a firm’s financial statement proxy
for taxable income or tax liability is adjusted based on the prior year relationship between a
firm’s actual tax return value and its financial statement proxy for that value.
5.2 Investor Reactions to Tax Return Disclosures and the Effect of Voluntary Disclosures
Next, I ask whether voluntary disclosures affect investors’ reactions to the ATO’s tax
return disclosures. To provide insight on this relationship I estimate the following regression:

|-IJYPZ|
%,6
"P !"#Y]I
%,6
= ^
(
+^
*
2,3$_H`
6
+^
1
2,3$_H`
6
∗|,- _$_cc|
%
+
^
:
2,3$_H`
6
∗|,- _$_cc|
%
∗!"#$
%
+^
>
2,3 $_H`
6
∗!"#$
%
+^
D
|,- _$_cc|
%
+
^
L
!"#$
%
+^
M
|,- _$_cc|
%
∗!"#$
%
+W
%,6
(2)

where |Returni,t| is the absolute value of firm i’s logarithmic daily return on day t, Volumei,t is
firm i’s daily trading volume scaled by a firm’s total shares outstanding on day t, ATODisc is an
indicator variable that equals one on the day of and the day after the ATO’s tax return disclosure
and zero otherwise, |TR_Diff| is a measure of the difference between a firm’s tax return values
and its financial statement values, and VolD is a firm specific indicator variable that equals one if
35

a firm issued a voluntary disclosure and zero otherwise. Following Hoopes et al. (2017), I
estimate this regression over the period November 1, 2015 (2016)-January 31, 2016 (2017) for
the first and second year of tax return disclosures.
This regression is essentially an event study that tests whether the market reacts to tax
return disclosures (ATODisc) and whether this reaction depends on both the extent to which the
information differs from the market’s prior expectations (ATODisc*|TR_Diff|) and if a firm
issued voluntary disclosures (ATODisc*|TR_Diff|*VolD). Similar to Beaver (1968), I use
unsigned returns and trading volume to measure market reactions. Abnormal unsigned returns
capture the magnitude of the average change in investors’ beliefs about a firm’s future cash
flows. Volume, on the other hand, captures changes in individual investor’s beliefs. Although
trading volume and unsigned returns are generally positively related, an abnormal increase in
volume could suggest that investors either interpreted the new information differentially or
interpreted prior information differentially and are now converging to the same belief (Karpoff,
1986). In both cases, investors’ prior beliefs are changed which leads to trading. Unsigned
returns may not capture these changed beliefs if the average of investors’ beliefs does not
change. Thus, I use both to test for changes in consensus or individual investor beliefs. I also test
signed returns for descriptive purposes.
H2 predicts that investor reactions to tax return information will increase in |TR_Diff|, a
positive ^
1
. Ex-ante it is not clear if investors will react more to taxable income differences or
tax liability differences. Thus, I estimate this regression with both specifications. In both,
|TR_Diff| is decile ranked and scaled between zero and one to control for outliers and any
potential nonlinearities in the relationship.
36

H3 predicts that the market’s reaction to firms that issue preemptive or concurrent
voluntary disclosures will increase less in |TR_Diff|. As such, I include a triple interaction to test
if investors react differently to large values of |TR_Diff | when additional information is and is
not available through voluntary disclosures. I estimate this regression with two specifications of
VolD. In my main specification, I use VolD_Pre/Conc. H3 is specific to this group as these
firm’s voluntary disclosures can be analyzed in conjunction with the ATO’s disclosure. For these
firms, I expect a negative ^
:
to offset a positive ^
1
.
In my second specification, I split VolD into three indicator variables: VolD_Pre,
VolD_Conc, and VolD_Post. The market’s reaction to the first two groups on the date of the
ATO’s disclosure may differ due to the timing of their voluntary disclosures. I expect that the
market will react more to concurrent disclosures as the market receives both the ATO’s tax
return information and the individual firm’s supplemental information at the same time. For
preemptive disclosures, the market likely reacts when the voluntary disclosures are issued such
that the market has less need to react when the same information is later issued by the ATO.
However, the market may still update their understanding of preempting firms on the date of the
ATO’s disclosure as investors are still learning the informative value of tax returns by observing
the firm’s peers’ returns. This is similar to the market reacting to the earnings announcement of a
peer firm (Foster, 1981)
12
. For both groups of firms, however, I still expect that the market’s
reaction will not be dependent on |TR_Diff|. For the last group, I expect an insignificant ^
:
as on
the date of the ATO’s disclosure, ex-post disclosers’ supplementary information is not yet
available. However, given that something prompts these firms to issue ex-post disclosures, the
market may react differently to these firms as well.

12
Investors may use peer firms as a benchmark to determine if a given firm’s |TR_Diff| is abnormal and/or if the
firm’s tax planning is more or less aggressive than its peers.
37

^
>
tests if the market responds differently to firms that issue voluntary disclosures
regardless of |TR_Diff|. I expect voluntary disclosures will provide context and clarification to
the ATO’s tax return disclosure which should reduce the cost of interpreting the tax return
information. If this is true, investors may become more confident in their beliefs and more
willing to trade on such beliefs (Blume, Easley, & O’Hara, 1994; Miller, 2010). In such a case,
the market may react more to tax return disclosures that are supplemented with voluntary
disclosures, a positive ^
>
. The rest of the variables in this regression are essentially control
variables that capture average market reactions for different groups of firms over my period of
estimation.
6. Sample Selection and Descriptive Statistics
6.1 Sample Selection and Data Collection
My sample starts with the 2,121 Australian entities (parents and subsidiaries) whose tax
returns are partially disclosed in either the ATO’s first or second round of tax return disclosures.
I drop Australian private companies as my hypotheses are predicated on the availability of public
financial statements. I also drop foreign owned firms as I expect that the Australian operations of
most of these firms are immaterial. As such, I do not expect them to provide a strong test of my
hypotheses. Lastly, I drop unlisted public firms, as they are not publicly traded. In Australia,
private companies can only issue a limited number of shares. Thus, large private companies are
classified as public although they are not listed on a stock exchange (for example, a co-
operative).
I use Bureau Van Dyck’s Mint Global database to match each entity to its ultimate parent
company based on the entity’s Australian Business Number (ABN). I then match each ultimate
parent company to Compustat Global with its International Security Identification Number
38

(ISIN). I hand checked all entities that did not match to an ultimate parent company to determine
if the entity has a publicly traded parent company. I match 473 Australian tax returns to 395
publicly listed Australian parent companies but drop one entity and its related parent company
because its website is inaccessible. I need access to a firm’s website to search for voluntary
disclosures. Of the remaining 472 entities, the ATO discloses tax return information for 382 in
both years, 43 in only the first year (for a total for 425 first year tax returns), and 47 in only the
second year (for a total of 429 second year tax returns). My final sample includes firms with both
financial statement and stock return data available at the necessary dates. I lose a handful of
firms that are acquired, liquidated or go private in-between filing their tax return and the ATO’s
tax return disclosure. The sample is detailed in Table 1.
I hand collect data on voluntarily provided tax return disclosures. I searched each parent
company’s website (as listed in Compustat Global) as well as its stock exchange disclosures. I
used search functions on the websites as well as manually searched through each company’s
press releases, news releases, corporate responsibility reports, investor centers, and publicly
disclosed financial statements. Any firm that explicitly discusses the ATO’s tax return disclosure
in any firm issued report is considered to have voluntarily issued a supplemental tax return
disclosure. Approximately 10% of the parent companies in my sample issue additional voluntary
tax disclosures. Of these, about half are issued before the ATO’s disclosure, a quarter are issued
on the same day or the day after the ATO’s disclosure, and the remaining quarter are made after
the ATO’s disclosure. These statistics are available in Table 2 and the full distribution of
disclosure dates is shown in Figure 1. Most of these disclosures are made in separate reports,
however, a handful are within firms’ audited annual financial statements. Some reports are
39

purely qualitative while most are both qualitative and quantitative. Examples of these disclosures
are provided in Appendix B.
6.2 Summary Statistics
Table 2 presents summary statistics for all variables of interest as well as all control
variables. The average firm in my sample has a taxable income that differs from its grossed up
current tax expense by two percent of assets and a tax liability that differs from its current tax
expense by less than one percent of assets. On average, a firm’s taxable income tends to be
higher than suggested by its financial statements while its Australian tax liability tends to be
lower than its current tax expense. The former may be driven by the assumption of a 30% tax
rate while the average firm in my sample pays only a 16% tax rate as shown by TaxETR2.
Although Australia’s statutory rate is 30%, Figure 2 shows that only about 20% of my sample
pays the statutory rate while more than 35% of my sample has no Australian tax liability.
Table 3 provides additional descriptive statistics on my main variables of interest,
|TI_Diff| and |TL_Diff|. I sort firms based on quintiles of these two variables (separately) in 2015
and compare the average values of several variables across those sorts. I also test for differences
between means of the two extreme quintiles. My two measures of |TR_Diff|, |TI_Diff| and
|TL_Diff|, are highly correlated with each other as both monotonically increase across quintiles
of the other measure. Both measures are also highly correlated over time as the average value of
each variable in 2016 monotonically increases across a firm’s 2015 quintile. In addition, the
quintiles of both measures of |TR_Diff| are not disproportionately signed in one direction or the
other. Although there is some evidence that firms in the bottom quintile of both measures are less
likely to have tax return values that are lower than their financial statement values, the
distribution of positive and negative values in the other four quintiles are similar to each other.
40

Next, I examine how firm attributes that are most likely to create a difference between a
firm’s financial statements and its tax returns vary across quintiles of |TR_Diff|. As expected, I
find that firms with large differences between their actual taxable incomes and/or tax liabilities
and that suggested by their financial statements are more likely to have research and
development expenditures, foreign sales, estimation errors, and more subsidiaries. Specifically,
28% and 40% of firms in the top two quintiles of |TI_Diff| report research and development
expenses while only 14-19% of firms in the bottom three quintiles do. I also find that 36% and
52% of firms in the top two quintiles of |TI_Diff| have foreign sales versus only 22-30% of firms
in the bottom three quintiles. I do not find any evidence that loss firms are more likely to have
larger differences between their tax return and financial statement values and I only find some
evidence that discontinued operations are associated with tax return differences as they are
marginally associated with tax liability differences and are not associated with taxable income
differences. All of these attributes are discussed in more detail in the next section.
The rest of this table provides preliminary support for my hypotheses. Although
voluntary disclosure does not monotonically increase across quintiles of |TR_Diff|, firms in the
top quintile of both measures are more likely to issue voluntary disclosures than firms in the
bottom quintile of the respective measure. 18% (13%) of firms in the top quintile of |TI_Diff|
(|TL_Diff|) issue voluntary disclosures while only 6% (5%) of firms in the bottom quintile do the
same. This lends support to H1A, which predicts that voluntary disclosure will be positively
associated with |TR_Diff|.
I next examine how the market’s reaction to the ATO’s public tax return disclosure varies
across |TR_Diff| quintiles and between firms that issue voluntary disclosures versus those that do
not. In general, the market reacts more to larger values of |TR_Diff|. On the day of and the day
41

after the ATO’s disclosure, firms with larger values of |TR_Diff| experience larger absolute daily
returns as well as larger absolute cumulative returns. This relationship is stronger for |TL_Diff|
than |TI_Diff | which suggests that the market may be more interested in a firm’s tax liability than
its taxable income. In preliminary support of H3, which predicts that the market’s reaction to
firms that issue voluntary disclosures will be less dependent on |TR_Diff|, the relationship
between |TR_Diff| and market reactions is strongest for firms that do not issue voluntary
disclosures. Again, this lack of relationship is stronger for TL_Diff than TI_Diff, which suggests
that voluntary disclosures may also be more focused on a firm’s tax liability than its taxable
income.
6.3 Fundamental Drivers of Tax Return - Financial Statement Differences
Table 4 provides descriptive evidence on what drives differences between a firm’s
financial accounting income tax expense and its Australian tax return values. In Australia, the
main drivers of reporting differences are provision to return adjustments, tax credits, intra-period
tax allocation, and uncertain tax positions (Hanlon, 2003; Tran, 2015). I estimate the average
effect of each of these items using the following OLS regression:

|,- _$_cc|= '
(
+'
*
SHJSPP"P+'
1
-&$+'
:
,2 _o22B +'
>
p"PI_QZ+'
D
?Y]qYrH
+'
L
$_H`3sH +'
M
N"HH+W

where |TR_Diff| is as previously defined and the independent variables are proxies for the main
drivers mentioned above. I estimate this regression over the first two years of public tax return
disclosure and multiply the dependent variable by 100 for presentation purposes. All continuous
variables are decile ranked and scaled between zero and one.
42

EstError is the proportion of a firm’s current year total tax expense that is related to the
previous year’s estimation error. In Australia, as in the US, most firms file their financial
statements months before they file their tax returns. As such, a firm’s financial statement tax
expense is an estimate of the amount the firm expects to report when it files its tax return.
Immaterial estimation errors are trued up in a firm’s next financial statement as a provision to
return adjustment. In their financial statements, Australian firms separately report their provision
to return adjustments. I hand collected these adjustments and exclude them from my
measurement of current tax expense. However, to the extent that a firm’s future financial
statements have not yet been issued, a firm’s current year estimation error is not yet known.
Thus, I estimate a firm’s current year estimation error based on its prior year estimation error.
Table 4 shows that EstError is not a significant predictor of taxable income differences but there
is some evidence that it predicts tax liability differences. In general, it appears that firms
overestimate their current tax expense, as EstError is associated with tax liabilities that are
smaller than expected. This may be a conservative choice by the firm.
R&D equals one if a firm reports any research and development expense and zero
otherwise. Australia, like the US, allows a tax credit for research and development expenditures.
Tax credits reduce a firm’s income tax liability dollar for dollar (as opposed to a reduction equal
to the tax rate). This makes a firm’s tax liability less than 30% of its Australian taxable income
such that the firm’s grossed up current tax expense will underestimate its actual taxable income.
However, I only find weak evidence that R&D is associated with |TI_Diff| and although not
significant, R&D is negatively associated with signed TI_Diff. The use of a tax credit would
predict the opposite. Similarly, a tax credit should not affect the mapping between a firm’s
current tax expense and its actual tax liability; however, R&D is also associated with tax
43

liabilities that are lower than a firm’s current tax expense. Firms with research and development
expense, on average, have tax liabilities that are $3 million lower than their current tax expense
('
1
= −0.43,qJC SPP"P= 0.10)
13
. This suggests that firms with research and development
expenditures have some other characteristic that alters the mapping between their current tax
expense and their tax returns. One possibility is that research and development is associated with
an uncertain tax position. If the firm is uncertain about its tax credit calculation, it must accrue
additional current tax expense to account for any future payments that may be required upon
audit. Such an accrual causes a disconnect between a firm’s current tax expense and both its tax
liability and its taxable income.
TA_GAAP is as previously defined. I include it in this regression as a proxy for uncertain
tax positions as firms that engage in more tax planning are more likely to have uncertain tax
positions. Unlike GAAP, IFRS does not require firms to separately disclose their uncertain tax
position accrual. As such, I cannot perfectly control for these accruals and instead must rely on a
proxy. I find that firms that engage in more tax planning have larger taxable income and tax
liability differences. These differences, however, appear to go in both directions. If this was a
perfect measure of uncertain tax positions it would only be associated with negative differences,
however, to the extent that it is an imperfect proxy it may simply capture that firms that engage
in more tax planning have more complex structures that result in more tax return differences.
Foreign equals one if a firm reports any sales outside of Australia and zero otherwise.
Foreign earnings make it difficult to infer a firm’s tax return values from its financial statement
values for two reasons. First, if a firm has foreign income, it will pay taxes on this income in the
foreign jurisdiction in which it is earned. Australian firms do not separately report their domestic

13
The dependent variable, TL_Diff, is scaled by assets and multiplied by 100. Thus, a coefficient of -0.43 implies
.43% of a firm’s assets. At the median value of assets, this corresponds to approximately $3 million.
44

and foreign tax expense. Thus, a firm’s current tax expense includes its Australian tax liability as
well as any foreign tax liabilities. This will cause a difference between a firm’s current tax
expense and its Australian tax liability as well as make it harder to estimate the firm’s Australian
taxable income. Second, Australia has a worldwide tax base so a firm’s foreign income will
eventually be taxed again in Australia. When it is, the company will be able to take a credit for
any foreign taxes already paid on this income. The use of this credit makes it harder to estimate a
firm’s taxable income from its current tax expense. I find that Foreign is a significant predictor
of both |TI_Diff| and |TL_Diff|. Further, I find that Foreign is associated with taxable incomes
and tax liabilities that are lower than suggested by a firm’s current tax expense. Specifically,
firms with foreign sales have taxable incomes (tax liabilities) that on average, are $6 million ($2
million) lower than that suggested by their current tax expense ('
>
= −0.82,qJC SPP"P =
0.36 for ,| _$_cc and '
>
= −0.30,qJC SPP"P= 0.09 for ,N _$_cc). This suggests that the first effect
dominates. If a firm has any foreign current tax expense, its total current tax expense will
overestimate a firm’s Australian taxable income and its Australian tax liability.
NumSubs counts the number of subsidiaries in which a firm owns more than 50%. Both
wholly owned and less than wholly owned subsidiaries can make it difficult to infer a firm’s tax
return values from its financial statement values. Only wholly owned subsidiaries can elect to be
consolidated on a parent company’s tax return. Due to this, a consolidated entity for financial
statement purposes may file several different tax returns for both wholly owned subsidiaries that
are not consolidated for tax purposes and all less than wholly owned subsidiaries that the firm
controls. Thus, an individual tax return as disclosed by the ATO may not represent all of a firm’s
operations. When I calculated a firm’s actual taxable income and tax liability, I included all of its
controlled subsidiaries. However, I only have access to the tax returns of large subsidiaries (that
45

have total income above the ATO’s disclosure threshold). If a firm has any subsidiaries that are
below the disclosure threshold, their taxable income and tax liabilities are omitted from my
calculation. Since the firm’s current tax expense consolidates all of the subsidiaries in which it
owns more than 50% (the portion related to noncontrolling interests is backed out later, net of
tax) a firm may show a difference between its tax return and financial statements simply because
we do not have access to all of their subsidiaries’ tax returns.
Interestingly, Table 4 shows that NumSubs is associated with smaller absolute values of
TI_Diff. This suggests that firms with more subsidiaries have smaller differences between their
tax return and financial statement values. This may be the case if the subsidiaries of larger firms
with the most subsidiaries are more likely to be larger themselves and above the ATO’s
disclosure threshold while the subsidiaries of smaller firms with fewer subsidiaries are more
likely to be below the ATO’s disclosure threshold. In such a case, my measure of a firm’s tax
return values with more subsidiaries would represent more of the reporting entity and there
would be less of a difference between the two reporting sets
14
.
DiscOp equals 1 if a firm reports any profit or loss from discontinued operations in their
income statement and zero otherwise. IFRS, like GAAP, requires firms to separately report any
profit or loss from discontinued operations net of income tax. As such, income taxes related to
discontinued operations are not a part of a firm’s current tax expense but will be on a firm’s tax
return. I find some evidence that DiscOp is associated with larger values of |TR_Diff| but not in
any one direction. This is not surprising as the relationship can go in both directions. If a firm’s

14
Subsidiaries may also pay dividends to their parent company, which can further complicate a firm’s income tax
accounting. When a dividend is paid from a subsidiary to the parent company, this income will be included in
taxable income twice. Once when the income is earned on the subsidiary’s tax return and again as dividend income
to the parent. Australia allows a credit for any taxes already paid by the issuing corporation. This is similar in
concept to the dividend received deduction allowed in America. Therefore, although the dividend income is included
twice, it is not taxed twice. As mentioned before, credits may make it harder to estimate a firm’s taxable income
from its current tax expense.
46

discontinued operations are profitable, its taxable income will be higher than expected and vice
versa if the discontinued operations are not profitable.
Lastly, Loss equals one if a firm’s pre-tax book income is less than zero and zero
otherwise. Firms with book losses are more likely to also have tax losses. The ATO does not
report taxable losses or income tax refunds. I have coded these missed observations as zero.
Thus, a firm’s current tax expense may differ from its tax return values due to the truncation of
the tax return values at zero. Table 4 shows that Loss is not associated with unsigned values of
TR_Diff but it is associated with positive values of TR_Diff. This is to be expected as a negative
current tax expense would suggest a taxable loss and no tax liability. In my data, a taxable loss is
truncated at zero. Thus, taxable incomes and tax liabilities of zero are higher than suggested by a
negative financial statement tax expense.
7. Main Results
7.1 Voluntary Disclosure Responses to Public Tax Return Disclosure
In this section, I turn to Table 5, which presents results from estimating equation (1).
|TR_Diff| relates to a firm’s taxable income difference (|TI_Diff|) in columns (1)-(3) and to a
firm’s tax liability difference (|TL_Diff|) in columns (4)-(6). Columns (1) and (4) test the
relationship between |TR_Diff| and voluntary disclosures made at any point in time. I find strong
support for H1A, which predicts that |TR_Diff| is positively associated with voluntary disclosure.
An untabulated marginal effects analysis finds that firms in the top decile of |TI_Diff| are 11%
more likely to issue a voluntary disclosure than firms in the bottom decile ('
*
=
1.04,qJC SPP"P = 0.39)
15
. Similarly, firms in the top decile of |TL_Diff| are 8% more likely to
issue voluntary disclosures than firms in the bottom decile ('
*
= 0.77,qJC SPP"P = 0.36). In

15
All regressions in Table 5 are estimated with a probit model. I estimate the marginal effect of |TR_Diff| by holding
all other significant explanatory variables at their means. Results are similar when estimated at the median.
47

this specification, |ATO_Return| is positive but insignificant which suggests that voluntary
disclosure, in general, is not associated with larger market reactions to tax return disclosure.
Voluntary disclosure is, however, associated with size. Firms in the top decile of ln(asset) are
40% more likely to issue voluntary disclosures than firms in the bottom decile in both
specifications of |TR_Diff|. This is likely because larger firms have a lower cost of disclosure
16
.
Next, I focus on the timing of voluntary disclosure. H1B predicts that preemptive and
concurrent disclosure is more likely to be driven by managers’ expectations of market reactions,
or |TR_Diff| while H1C predicts that ex-post disclosure is more likely to be driven by actual
market reactions, or |ATO_Return|. The results in Table 5 support both hypotheses. Columns (2)
and (5), which predict preemptive or concurrent voluntary disclosure, show that these disclosures
are positively associated with |TR_Diff| ('
*
= 0.87,qJC SPP"P = 0.41 for |TI_Diff| and '
*
=
0.71,qJC SPP"P = 0.37 for |TL_Diff|) and not associated with |ATO_Return|. Although I cannot
observe the market’s reaction to these firm’s tax returns without the voluntary disclosure, one
interpretation is that voluntary disclosure mitigated the market’s reaction.
In contrast, columns (3) and (6), which predict ex-post voluntary disclosure, show that
ex-post disclosures are marginally and not at all associated with |TI_Diff| and |TL_Diff|,
respectively but are strongly associated with |ATO_Return| ('
1
= 1.20,qJC SPP"P =
0.56 @ZC '
1
= 1.37,qJC SPP"P = 0.60 in column (3)and (6) respectively). Firms that
experienced the largest market reactions to the ATO’s disclosure (in the top decile of
|ATO_Return|) are 2% (3%) more likely to issue ex-post voluntary disclosures than firms that
experienced the smallest market reactions (in the bottom decile of |ATO_Return|) when |TR_Diff|

16
Larger firms are more likely to have in house financial reporting and tax departments. This likely decreases the
direct cost of disclosure. Further, larger firms with in house tax departments may have stronger and more certain tax
positions such that the proprietary costs of disclosure are also lower (Mills, Robinson and Sansing 2010).
48

is estimated with |TI_Diff| (|TL_Diff|). This suggests that while tax return differences may still be
important in motivating ex-post disclosures, the market’s reaction to a firm’s tax return
disclosure is also important in motivating voluntary disclosure decisions after tax return
disclosure. Thus, firms are more likely to issue voluntary disclosures when tax return disclosures
cause a larger shock to their information environment.
In Panel B, I estimate the probability of voluntary disclosure in the second year of public
tax return disclosure. As expected, Panel B shows that first year voluntary disclosure is a strong
predictor of second year voluntary disclosure. The coefficient on VolD_PY is positive and
statistically significant in all six columns (' = 3.15,qJC SPP"P = 0.41 in column (1) and is
similar in the other five columns). This suggests that prior year disclosures are associated with
both ex-ante and ex-post second year disclosures. We know from Table 3 that |TR_Diff| is highly
correlated from year to year and that investors prefer continuity in their disclosures (Bischof &
Daske, 2013) so it is not surprising that firms that issued previously would continue to disclose.
Interestingly, |TR_Diff2| is not statistically significant. In untabulated results, I find that this is
true whether or not VolD_PY is included in the regression and whether I define |TR_Diff| with or
without regard to a firm’s prior year tax return information (i.e. as |TR_Diff| or |TR_Diff2|). The
only exception is that |TL_Diff|, calculated without regard to prior year information, predicts
voluntary disclosure (both preemptive and ex-post) when I do not control for prior year
disclosures.
In contrast, I continue to find that ex-post disclosure is strongly driven by actual market
reactions. Firms that experience the largest market reactions to the ATO’s disclosure are 4%
more likely to issue ex-post voluntary disclosures than firms that experience the smallest market
reactions to the ATO’s disclosure (this estimation is approximately the same in both column (3)
49

and column (6) specifications). This shows that after one year of tax return disclosure, some
firms do not fully anticipate the market’s reaction to their tax return information and that market
reactions are still important determinants of disclosure.
17

In Panels C and D, I provide more detail on the drivers of voluntary disclosure in the first
year of disclosure, 2015. In Panel C, I ask whether the sensitivity of disclosure to |TR_Diff| is
conditional on the underlying drivers documented in Table 4. I focus on R&D, Foreign, DiscOps
and TA_GAAP as Table 4 shows that these four variables are positively associated with
|TR_Diff|. I interact each of these variables with |TR_Diff| to test if firms with any one or all of
these four characteristics drive the relationship between |TR_Diff| and voluntary disclosure. The
results show that firms with research and development credits are the main drivers of the positive
relationship between |TR_Diff| and voluntary disclosure. Firms in the highest decile of |TI_Diff|
(|TL_Diff|) are 52% (33%) more likely to issue voluntary disclosures than firms in the lowest
decile if they have research and development expenses whereas |TI_Diff| and |TL_Diff| are
marginally insignificant when a firm does not have research and development expenses. The
disclosure examples provided in Appendix B are consistent with this result.
Although foreign sales predict |TR_Diff|, they are not associated with voluntary
disclosures. This may be because the proprietary cost of revealing a firm’s international tax
planning and/or transfer pricing is too high to justify voluntary disclosures. Similarly, although
tax planning predicts |TR_Diff| it is not associated with voluntary disclosures. This is again,
likely due to proprietary costs as firms that engage in more tax planning than their peers would
likely incur high costs if they revealed their strategies to both their peers and the tax authority.

17
Although eight firms issue ex-post disclosures in each year, only two of the fourteen total firms issue ex-post
disclosures in both years. Most firms who issue ex-post disclosures in the first year issue preemptive disclosures in
the second year which suggests that they learned and tried to get ahead of the market in the second year.
50

Lastly, although Discop is marginally associated with |TR_Diff| it is negatively associated with
voluntary disclosures. This may be because firms do not want to draw attention to their
discontinued operations.
Panel D presents evidence on disclosure responses to signed values of TR_Diff and
ATO_Return. I interact both |TR_Diff| and |ATO_Return| with indicator variables that equal one
if either measure is negative (TR_Lower and Neg, respectively). The ex-ante interpretation of
TR_Diff’s sign is not obvious. TR_Lower captures firms that have low taxable income or tax
liabilities relative to what was previously reported to shareholders. This may suggest lower
performance, more tax planning, or a structural difference that implies neither. Column (1),
which relates to voluntary disclosures issued at any point in time, shows that firms with negative
values of |TI_Diff| (TR_Lower=1) are not more or less likely to issue voluntary disclosure than
firms with positive values of |TI_Diff|. |TI_Diff| remains positive and statistically significant,
which shows that large deviations in either direction are associated with voluntary disclosure. In
column (2), which relates to preemptive and concurrent disclosures, there is some evidence that
firms whose taxable income is lower than expected are not more likely to issue voluntary
disclosures. While the coefficient on |TI_Diff| remains positive and statistically significant, the
coefficient on |TI_Diff|*TR_Lower is negative and marginally significant. This negative
coefficient almost entirely offsets the positive coefficient on |TI_Diff|. This suggests that the
relationship between |TI_Diff| and preemptive disclosure is predominantly driven by taxable
incomes that are higher than expected. This is in line with Panel C, which showed that the
relationship between |TI_Diff| and voluntary disclosure is predominantly driven by research and
development expenditures. Although Table 4 shows that when controlling for other
characteristics, the relationship between TI_Diff and R&D is negative, if firms with research and
51

development expense take a research and development credit, the firm’s tax liability will be less
than 30% of its taxable income such that its taxable income will be higher than expected. In
columns (4) and (5) where the same tests are estimated with |TL_Diff|, the additional interaction
causes |TL_Diff| to become statistically insignificant. This is likely due to a power issue, as the
coefficient on |TL_Diff| by itself remains positive and almost significant.
Interestingly, there is some evidence that firms that issue preemptive or concurrent
voluntary disclosure experience less negative market reactions to the ATO’s disclosure. While
preemptive disclosure is positively associated with negative market reactions in general (' =
1.19,qJC SPP"P = 0.54), it is also negatively associated with large negative market reactions
(' = −1.97,qJC SPP"P = 0.88). This is a backwards regression as the disclosure occurs before
the market reaction, but the associations provide preliminary evidence that voluntary disclosure
may have mitigated larger negative market reactions.
Columns (3) and (6) also show that ex-post voluntary disclosure was more strongly
driven by large negative market reactions than large positive market reactions. The negative and
marginally significant coefficient on Neg (' = −4.59,qJC SPP"P = 2.64 @ZC ' =
−3.95,qJC SPP"P = 2.25 for |TI_Diff| and |TL_Diff| respectively) suggests that overall, firms
with negative market reactions are less likely to issue subsequent voluntary disclosures.
However, the positive and marginally significant coefficient on |ATO_Return|*Neg (' =
5.54,qJC SPP"P = 3.01 @ZC ' = 4.91,qJC SPP"P = 2.58 for |TI_Diff |and |TL_Diff|
respectively) suggests that firms with large negative market reactions are more likely to issue
subsequent voluntary disclosures. This is consistent with the market reaction incentivizing
managers to follow really bad news with not as bad news (Jung and Kwon, 1988; Sletten, 2012).

52

7.2 Investor Reactions to Public Tax Return Disclosure
Next I focus on Table 6 which presents the results of estimating equation (2). In Panel A
the dependent variable is unsigned daily returns (multiplied by 100). |TR_Diff| represents taxable
income differences in columns (1)-(3) and tax liability differences in columns (4)-(6). The two
specifications provide very similar results. In all six columns, the coefficient on ATODisc by
itself is not significant. This shows that the market did not react unconditionally to the ATO’s tax
return disclosures. However, as predicted by H2, the coefficient on ATODisc*|TR_Diff| is
positive which suggests that the market’s reaction to tax return disclosures was positively
associated with the difference between a firm’s tax return values and its previously disclosed
values. This relationship, however, is only significant for tax liability differences (β=0.47, Std
Dev=0.13), not taxable income differences (β=0.28, Std Dev=0.33). On average, firms in the top
decile of tax liability differences experienced a daily price change of 47 basis points more than
firms in the bottom decile. This is a 27% increase over normal price changes in my sample
period which suggests that in the absence of additional information, the market experiences a lot
of volatility related to tax return differences.
In line with H3, however, I find that voluntary disclosure can mitigate this reaction.
Columns (2) and (4) show that the market reacted differently to firms that issued preemptive or
concurrent voluntary disclosures. The negative and significant coefficient on
ATODisc*|TR_Diff|*VolD_Pre/Conc (β=-1.64, Std Dev=0.56
18
) offsets the positive coefficient
on ATODisc*|TR_Diff|. Columns (3) and (6) reiterate this result and show that this result holds
for both preemptive and concurrently issued voluntary disclosures. This suggests that, in line
with H3, the market’s reaction to firms that issued preemptive or concurrent voluntary disclosure

18
Although the point estimates vary when |TR_Diff| is calculated with |TI_Diff| or |TL_Diff|, the overall story is the
same. Given that the market seems to react more to |TL_Diff|, I provide coefficients from columns (5) and (6).
53

was not positively associated with the difference between a firm’s tax return value and its
financial statement values. In other words, voluntary disclosure successfully mitigated volatility
related to tax return differences.
The positive and significant coefficient on ATODisc*VolD_Pre/Conc (β=0.68, Std
Dev=0.25) suggests that although the market’s reaction to firms that issued voluntary disclosures
was not affected by tax return differences, the market did react to firms that issued voluntary
disclosures. These reactions likely relate to the information within the voluntary disclosure rather
than to the tax return information itself. As one might expect, this reaction is mostly driven by
firms that issue concurrent disclosures (β=3.92, Std Dev=1.33) as the market receives these
firms’ voluntary disclosure information at the same time as the ATO disclosure. Interestingly,
preemptive disclosers also experience an increase in unsigned returns at tax return disclosure
(β=0.84, Std Dev=0.20). Although tax return disclosure does not provide new information about
these firms, I suspect that the market is reacting to information on their peer firms. This reaction
would of course not be conditional on the preempting firm’s tax return difference. Lastly, both
columns show that unsigned returns for ex-post disclosers were not different from firms that did
not issue any voluntary disclosures at all.
Panel B estimates changes in trading volume at tax return disclosure. I find that trading
volume, unlike unsigned returns, increased unconditionally at tax return disclosure. This is likely
due to investors’ holding different ex-ante expectations of a firm’s tax return values such that
when actual tax return values were disclosed, investors’ individual opinions changed without an
overall consensus price change. I also find that similar to unsigned returns, volume increased
more for firms with larger values of |TL_Diff|. There is some evidence that firms that issued
voluntary disclosures did not experience this same increase in trading volume, however, the
54

results are not as strong as they were with unsigned returns. There was also more trading in
general for concurrent disclosures, which is likely due to the market reacting to the voluntary
disclosure itself rather than the tax return information.
Panel C estimates signed market reactions to tax return disclosures. I find that on average,
the market reacted positively to tax return disclosures (^ = 0.61,qJC $IO = 0.26 in column (1)
and is similar in the other five columns). Given that Hoopes et al. (2017) finds that the market
reacted negatively when Australia’s tax return disclosure law was passed, the positive reaction
may suggest that the information was better than the market expected. The market’s reaction
does not appear to vary for different magnitudes of tax return differences, which suggests that tax
return differences led to increased volatility but no consensus price change.
I also find that the market reacted most positively to firms that issued preemptive
voluntary disclosures. Column (3) provides some evidence that this reaction is positively related
to |TI_Diff| | (^ = 1.78,qJC $IO = 1.05) and column (6) provides stronger evidence that the
positive reaction is increasing in |TL_Diff| (^ = 3.54,qJC $IO = 0.80). This is interesting
because unsigned returns of preemptive disclosures were not positively related to |TR_Diff|. This
suggests that firms that issue preemptive disclosures experienced a smaller but directionally
consistent change in firm value over the day of and the day after the ATO’s disclosure. The
opposite result for firms that do not issue voluntary disclosures suggests that these firms
experienced large daily returns that moved in different and offsetting directions across the two
days i.e. increased volatility. Together, the results suggest that voluntary disclosure decreased the
cost of understanding a firm’s tax return information such that the market was able to reach a
consensus on the meaning of these firm’s tax return differences quickly. In the short term, this
55

result, may be concentrated in preempting firms as the market has had the most time to digest
these firms’ information.
Panel C also shows that ex-post disclosers had larger signed returns over the two days. In
general, ex-post disclosers experienced negative market reactions to the ATO’s disclosure (^ =
−2.14,qJC $IO = 0.82). However, the reaction is positively associated with |TL_Diff| (^ =
2.59,qJC $IO = 0.81 ). This shows that the market’s reaction to ex-post disclosers was more
consistent and less volatile over the two days. These consistent reactions in both directions may
have spurred managers’ subsequent disclosures.
8. Supplementary Analyses
8.1 Investor Reactions to Public Tax Return Disclosure Conditional on the Sign of Tax
Return Differences
My first supplementary analysis examines market reactions to tax return disclosures
conditional on the sign of TR_Diff. Although I expect the market to react to tax return values that
differ from a firm’s financial statements in either direction, it is possible that investor responses
depend on the direction of deviation. This could be because higher or lower taxable incomes or
tax liabilities send different signals about the firm’s underlying performance or tax planning. In
Table 7, I estimate equation (2) separately for firms with positive and negative values of TL_Diff.
Since Table 6 showed that the results were similar when equation (2) was estimated with both
TI_Diff and TL_Diff, but stronger with TL_Diff, I estimate these regressions using a firm’s tax
liability difference.
In Panel A, the dependent variable is unsigned returns. Columns (1) – (3) relate to firms
with tax liabilities that are larger than their current tax expense while columns (4) - (6) relate to
firms with tax liabilities that are smaller than their current tax expense. The results are similar
56

between both groups of firms. There is some evidence that the market reacts more to tax
liabilities that are larger than their current tax expense (^ = 0.74,qJC $IO =
0.25 when ,- _N"áIP= 0 versus ^ = 0.26,qJC $IO = 0.05 when ,- _N"áIP = 1 ), however,
overall, the market reacts to large values of |TL_Diff| whether they are positive or negative.
The market reactions conditional on voluntary disclosure are also similar to that
discussed in Table 6. The market reacts less to the difference between a firm’s tax liability and
its current tax expense when the firm issues preemptive or concurrent voluntary disclosures.
Although the signs of the coefficients on ATODisc*|TL_Diff|*VolD_Pre/Conc are the same
when TL_Diff is positive or negative, there is some evidence that voluntary disclosures are more
effective in offsetting market reactions to tax liabilities that are larger than expected as the
coefficient in column (2) (as opposed to column (5)) is larger and statistically significant. This is
interesting because Table 5 also shows that firms are more likely to issue preemptive disclosures
when their tax liability is larger than expected. Although I cannot observe the counterfactual, this
pattern of results suggests that managers are more likely to issue voluntary disclosures when they
are most likely to be effective. In other words, managers anticipated that investors would react
the most to tax liabilities that were larger than expected and issued voluntary disclosures to offset
these reactions. This is also in line with Hoopes et al. (2017) who finds that public firms are
more likely to increase tax avoidance after the ATO initiated public tax return disclosure.
Without additional information, the market may infer that firms with tax liabilities that are larger
than expected are engaging in suboptimal tax planning. While some firms may attempt to
prevent such an inference by increasing their tax avoidance, others seem to respond by issuing
clarifying disclosures.
57

In Panel B, the dependent variable is volume. Again, Columns (1) - (3) relate to firms
with actual tax liabilities that are larger than their current tax expense while columns (4) - (6)
relate to firms with actual tax liabilities that are smaller than their current tax expense. The
results are similar for both positive and negative values of TL_Diff. However, there appears to be
more trading in relation to negative values of TL_Diff. This is opposite to Panel A, which showed
more price movement for positive values of TL_Diff.
Higher trading with little price movement suggests one of two things. Either investors
have more trouble forming a consensus valuation of the tax return information of firms with tax
liabilities that are lower than expected or, investors’ ex-ante expectations of these firm’s tax
liabilities were more diverse and tax return disclosure resolved these differences in opinions
(Karpoff, 1986). Given that it was not known ex-ante whether or not a firm’s tax liability was
lower than expected, the latter would only be the case if firms with lower tax liabilities had some
other characteristic that made their tax positions harder to infer. Thus, the former explanation
seems more plausible: when a firm’s tax liability is lower than expected, there is more
disagreement among investors.
Finally, Panel C of Table 7 provides results when the dependent variable is signed
returns. Similar to Table 6, I find that, overall, the market reacts positively to tax return
disclosures regardless of the sign or magnitude of the deviation between a firm’s tax liability and
its current tax expense. I do find, however, that the large positive reaction to preemptive
disclosures noted in Table 6 appears to be concentrated amongst firms with tax liabilities that are
lower than their current tax expense. The coefficient for preemptive disclosers with large and
negative values of TL_Diff is positive and significant (^ = 3.57,qJC $IO =
0.61 in column (6)), however, the coefficient in column (3) is also positive ((^ =
58

2.51,qJC $IO = 3.34). Given that the majority of firms have tax liabilities that are lower than
their current tax expense, the positive but insignificant coefficient in column (3) may be due to a
lack of power. This pattern of results is also similar for ex-post disclosures: the results
documented in Table 6 are again, stronger for firms with tax liabilities that are lower than
expected. This may suggest that the market reacted more consistently to firms with tax liabilities
that were lower than expected, however, the sign of the coefficients in columns (3) and (6) are
the same and the patterns of results are similar for firms with both directions of TL_Diff.
8.2 Investor Reactions to Public Tax Return Disclosure in the Second Year of Disclosures
I next estimate market reactions to the ATO’s second year of tax return disclosures. Table
8 is similar to Table 6 except that the regressions are estimated over the three months
surrounding the ATO’s second tax return disclosure on December 9, 2016. Overall, the results
show that the market reacted very similarly to the ATO’s first and second tax return disclosures.
This suggests that the documented market reactions are related to tax return disclosures and not a
different concurrent event.
In Panel A, the dependent variable is unsigned daily returns (multiplied by 100) and
|TR_Diff| represents taxable income differences in columns (1)-(3) and tax liability differences in
columns (4) –(6). Similar to the first year, the two specifications provide similar results. In all six
columns there is a negative and significant coefficient on ATODisc. This suggests that similar to
the first year, the market did not unconditionally react to tax return disclosures. In fact, the
negative coefficient suggests that the market was less volatile than normal on the day of and the
day after the ATO’s second tax return disclosure.
The market also continued to react to tax return differences. I find positive and significant
coefficients on the interaction of ATODisc with both |TI_Diff2| and |TL_Diff2|. This suggests that
59

in the second year of tax return disclosures, the market reacted to both taxable income and tax
liability surprises whereas in the first year, the market only reacted to tax liability surprises. The
market’s first year reaction may have been muted if it did not yet fully understand the
implications of a firm’s true taxable income. The fact that the market strongly reacts to taxable
income differences in the second year suggests that it learned over time. Further, |TR_Diff2| takes
into account the relationship between a firm’s tax return values and its financial statement values
revealed in its first tax return disclosure. In untabulated tests, I find that the market does not react
to tax return differences that do not take into account this prior year information (|TR_Diff|)
which shows that the market learned from the first year disclosure and now focuses on new tax
return information that is incremental to the first year’s disclosure.
As in the first year, there is evidence that the market reacted differently to firms that
issued voluntary disclosures. The negative and significant coefficient on
ATODisc*|TR_Diff2|*VolD_Pre in columns (3) and (6) offsets the positive and significant
coefficient on ATODisc*|TR_Diff| which suggests that the market did not react to tax return
differences for firms that issued preemptive disclosures. This is similar to the results for the first
year of tax return disclosures. However, unlike the first year, I do not find that concurrent
disclosures had the same effect. The coefficient is still negative which provides some evidence
that concurrent disclosures helped to mitigate market reactions to a firm’s tax return differences,
however, the coefficient is insignificant in both column (3) and (6).
In this table, I include an additional indicator variable for firms that issued voluntary
disclosures in the prior year. Given that most tax planning is long term, I expect supplemental
information contained in a first year voluntary disclosure will affect investors’ understanding of
a firm’s second year tax return information. I find some evidence that firms that issued prior year
60

disclosures experienced lower overall market reactions to their second year tax return
information. The negative and significant coefficient on ATODisc*VolD_PY in most
specifications indicates that these firms experienced less volatility than other firms at tax return
disclosure. These firms had the most ex-ante tax information available. As such, it is not
surprising that at tax return disclosure, the market had less new information with which to react.
Panel B estimates changes in trading volume at the second tax return disclosure. I find
little evidence of increased trading volume on the day of and the day after the ATO’s tax return
disclosure and find no evidence that firms with larger tax return differences experienced higher
trading volume than normal. However, I find some evidence that trading volume related to tax
return differences was lower for firms that issued preemptive voluntary disclosures. Overall,
there was not much trading activity in the second year of tax return disclosure. This may be
because the first year’s information helped market participants form similar expectations such
that there was less ex-ante disagreement, which subsequently lead to fewer reallocations when a
firm’s true tax return information was disclosed.
In Panel C, I study signed market reactions to the ATO’s second tax return disclosures.
Similar to the first year, I find that on average, the market reacted positively to tax return
disclosures. However, in the second year, the market’s reaction is affected by the difference
between a firm’s tax return values and its financial statement values. Specifically the market
reacted less positively and eventually negatively to firms with larger differences between their
tax return values and financial statement values. In the first year, although firms with larger tax
return differences experienced larger unsigned returns they did not incur a consistent change in
firm value. This reflects increased volatility and a lack of market consensus which was likely
caused by the high cost of understanding these new disclosures. I suspect that the more nuanced
61

reaction in the second year is because the market learned from the first tax return disclosures
such that it was able to reach a consensus faster in the second year.
There is some evidence that voluntary disclosure prevents a negative reaction to
|TR_Diff2|. Although the market’s reaction to concurrent disclosure is still negatively related to
the difference between a firm’s tax return values and its financial statement values, the markets’
overall reaction to concurrent disclosures is positive. At all levels of |TL_Diff2|, the market’s
reaction to firms that issued concurrent disclosures is more positive than to firms that do not
issue voluntary disclosures. This suggests that voluntary disclosures successfully mitigated
negative market reactions to a firm’s tax return information.
I also continue to find that the market reacted less on the day of and the day after tax
return disclosure for firms that issued prior year disclosures. The negative and significant
coefficient on ATODisc*VolD_PY in column (4) of Panel B offsets the positive and significant
coefficient on ATODisc. This indicates that the market did not unconditionally react to the tax
return disclosures of firms who issued prior year voluntary disclosures. However, I also find that
prior year disclosures did not mitigate the market’s reaction to a firm’s tax return differences.
Instead, the market’s signed and unsigned reaction to the tax return disclosures of firms that
issued prior year disclosures was positively associated with |TR_Diff2|. |TR_Diff2| essentially
captures a firm’s tax return “surprise”, if the firm did not also issue a current year voluntary
disclosure, it makes sense that the market would react to the new information. However, the
positive reaction (as opposed to the negative reaction for firms with no voluntary disclosures)
suggests that the prior year supplemental information painted the new information in a positive
light.
62

Similar to the first year, I also find that the signed market reaction to ex-post disclosures
was different from other firms. Ex-post disclosers with small differences between their tax return
and financial statement values experienced more negative market reactions than other firms
whereas firms with large differences experienced more positive market reactions. This again
reiterates that both positive and negative market reactions to involuntary disclosure can motivate
additional disclosures from the firm.
8.3 Tax Transparency Code Disclosures
Around the same time that the Australian government initiated public disclosure of
corporate tax returns, it also asked the Board of Taxation
19
to review and recommend updates to
firm issued income tax disclosures. The board worked on their recommendations in the summer
of 2015 and issued a first draft of their report, the Tax Transparency Code (TTC), on December
11, 2015. A final, updated report was issued to the public on May 3, 2016.
The TTC is entirely voluntary and suggests three new disclosures for firms with
Australian ordinary income
20
over $100 million and an additional three new disclosures for firms
with Australian ordinary income over $500 million (a total of six new disclosures). For firms
with incomes between $100 and $500 million, the TTC suggests that firm’s disclose: (1) a
reconciliation of pretax income times the statutory tax rate to the firm’s accrued income tax
expense with detail on material temporary and permanent differences (which is already required
under IFRS and disclosed by all public companies), (2) a second reconciliation from income tax
expense to the actual income taxes paid or payable and, (3) their Australian specific and global

19
The Board of Taxation is a group of nongovernmental officials appointed by the Treasurer who are tasked with
“improving the design of taxation laws and their operation” http://taxboard.gov.au/about/governance/
20
The law technically defines amounts relative to ‘turnover’, which is a firm’s ordinary income. The TTC defines
Australian turnover as the global turnover of a business headquartered in Australia and the Australian related
turnover of a foreign business.
63

effective tax rates as defined by the Australian Accounting Standards Board (AASB)
21
.
Corporations with income over $500 million are further encouraged to provide: (1) a summary of
corporate taxes paid, (2) a qualitative report that describes the firms' tax policy, tax strategy, tax
governance and approach to tax risk management and, (3) a description of any material
transactions with offshore related parties. The TTC notes that these new disclosures do not need
to be audited and can be included in other reports that the Company already makes (such as
financial reports, corporate social responsibility reports, or global taxes paid reports). The TTC
also asks that firms provide their reports to the ATO so that the ATO can create a central
depository of links to all participating companies’ reports.
22

For the time being, these disclosures are voluntary because the government believes that
this choice will lead to discussions amongst senior management and the board, which may lead
to increased awareness of the tax function. The government has noted, however, that if voluntary
implementation is slow, the requirement may become mandatory in the future
23
. Although the
TTC is not the primary focus of this paper, it is important to note that this code was created at the
same time that Australian firms became subject to involuntary tax return disclosure. Essentially,
the TTC provides a framework that firms can follow if involuntary tax return disclosure shifts
their tax disclosure equilibrium. This way, there is enhanced consistency and comparability
between firms’ disclosures.
Table 9 presents summary statistics on my TTC disclosure measures. CountTTC is a
count variable that ranges from 0-6 and counts the number of disclosures a firm makes out of the

21
The AASB issued a draft appendix to the TTC in May 2017. This draft defines ETR as income tax expense
without any adjustments for prior years, interest, and/or penalties divided by pre-tax accounting profit.
22
Which is housed here: http://taxboard.gov.au/current-activities/transparency-code-register/
23
To my knowledge, no prior research examines whether threats of mandatory disclosure affect firms’ voluntary
disclosure decisions. This may be an interesting avenue for future research.
64

TTC’s six recommendations. Most firms do not issue new TTC disclosures as a count of one
simply captures the first TTC recommendation, which is already required by IFRS. IncTTC
equals one if a firm increased its TTC disclosures in the year following the government’s
issuance of the Tax Transparency Code and zero otherwise and TTC equals one if a firm issues at
least three of the TTC’s recommended disclosures. Approximately 11% of my sample increases
their TTC disclosures in the year following the first issuance of the TTC and a similar percentage
discloses at least three TTC disclosures.
Panel B of Table 9 describes differences in the means of my TTC variables for firms that
also issue tax return specific voluntary disclosures and those that do not. I find that firms that
issue tax return specific voluntary disclosures are also more likely to issue TTC disclosures. This
provides preliminary evidence that TTC disclosures may be a compliment rather than a substitute
for tax return specific disclosures.
8.3.1 Determinants of TTC disclosures. To provide more evidence on why firms
voluntarily issue TTC disclosures, my next set of tests investigates whether firms issue TTC
disclosures instead of tax return specific voluntary disclosures. The ATO’s tax return disclosures
relate to tax years that are one to two years old. Firms may decide that explaining such
information is no longer value relevant and instead provide timely TTC disclosures related to
current year operations. For instance, a firm could provide a reconciliation of its current year tax
expense to current year income taxes paid and/or new information on its tax strategy or
Australian specific tax operations. Such information may still help a firm’s investors understand
their past tax return information. As such, I test whether the ATO’s tax return disclosures are
associated with changes in a firm’s TTC disclosures.
65

To do so I re-estimate equation (1) with a dependent variable that measures the change in
a firm’s TTC disclosures, IncTTC. The first draft of the TTC was issued on December 11, 2015,
just six days before the ATO’s first tax return disclosure. This did not provide enough time for
firms to draft a TTC disclosure before the ATO’s first tax return disclosure. However, firms
could use the TTC framework to structure ex-post disclosures in response to the ATO’s first tax
return disclosure or to structure preemptive disclosures in anticipation of the ATO’s second tax
return disclosure. Since it is not clear whether any TTC disclosures made in this period would be
driven by the ATO’s previous disclosure or the ATO’s next disclosure (they relate to a firm’s
most recent fiscal year rather than the tax return year disclosed by the ATO), I estimate the
equation twice, using tax return information from the first and second year of public disclosure,
separately. I exclude all firms that issue tax return specific disclosures, as I am interested in the
issuance of TTC disclosures as an alternative voluntary disclosure rather than simply additional
supplemental information.
Table 10 provides results. In the first and third column, |TR_Diff| and |ATO_Return| relate
to the first year of tax return disclosure whereas in the second and fourth columns these variables
are calculated with the second year tax return disclosures. I also vary whether |TR_Diff| relates to
taxable income or tax liability, in the first and second column |TR_Diff| is calculated with a
firm’s taxable income whereas in the third and fourth column |TR_Diff| is estimated with a firm’s
tax liability. In all four columns, I fail to find any evidence that anticipated or actual market
reactions to tax return disclosures are associated with a change in a firm’s TTC disclosures. This
suggests that TTC disclosures are not motivated by tax return disclosures and are driven by
concerns other than the ATO’s disclosures. I continue to find that size is a strong predictor of
66

voluntary disclosures, which reiterates that larger firms can issue additional disclosures at a
lower cost.
8.3.2 Investor reactions to TTC disclosures. Lastly, I ask if additional tax disclosures
that are not directly related to a firm’s tax return information, TTC disclosures, affect investors’
reactions to tax return information. In Table 11, I re-estimate equation (2) with an additional
measure of voluntary disclosure, TTC. I estimate these tests in the second year of tax return
disclosure, as the TTC was not formally created until after the first year of disclosures.
The dependent variable is unsigned returns in columns (1) and (2), trading volume in
columns (3) and (4) and signed returns in columns (5) and (6). As most of these results have
already been discussed in section 8.2, I focus specifically on the differences. Controlling for TTC
disclosures does not affect my prior inferences related to the effect of direct voluntary
disclosures on market reactions. I find, however, that TTC disclosures have a similar effect on
the market’s reaction to tax return disclosures as direct voluntary disclosures do. The negative
and significant coefficient on ATODisc*|TR_Diff2|*TTC in columns (1) and (2) offsets the
positive and significant coefficient on ATODisc*|TR_Diff2|. This suggests that although TTC
disclosures were not issued to address tax return differences, they still provide some information
that mitigates the market’s reaction to the difference between the firm’s tax return values and its
previously released financial statements. In addition, the unsigned price reaction to firms that
issue TTC disclosures in general, is higher which suggests that TTC disclosures also helped the
market react to the firm’s tax return information with more certainty.
Columns (5) and (6) show that this stronger market reaction to firms that issued TTC
disclosures is related to positive returns. However, although the market’s reaction is generally
positive for firms that issue TTC disclosure, the market’s reaction to these firms is still
67

negatively related to the difference between a firm’s tax liability its current tax expense. This
suggests that while TTC disclosures may help the market react with more certainty (less
volatility) to a firm’s tax return disclosures, they do not prevent the market from negatively
reacting to large differences between the firm’s tax return information and its current tax
expense.
9. Conclusion
Understanding the incentives for voluntary disclosure is a long-standing area of interest
in accounting. Whether and how that is affected by mandatory disclosure requirements are
important questions that have generated recent empirical attention. In most settings, the firm is
responsible for issuing mandatory disclosures which allows managers to infuse that information
with voluntary disclosure. This likely leads to a more informative total disclosure but makes it
difficult for researchers to tease out the voluntary components. In this paper, I avoid this
empirical difficulty by studying a setting in which a third party issues a standardized firm level
disclosure that is outside of the firm’s control. If the firm wishes to supplement the required
information, it must do so through a disclosure that is entirely voluntary and easily separable
from the mandatory disclosure.
I find that in Australia, where tax return information is involuntarily disclosed by the tax
authority, firms that both anticipate and experience larger shocks to their information
environments, are more likely to issue supplemental disclosures. Specifically, I find that firms
with larger differences between their tax return values and their previously issued financial
statements are more likely to issue supplemental disclosures prior to or concurrent with the tax
authority’s involuntary disclosure. I also find that some firms appear to wait to first observe the
market’s reaction to the involuntary disclosure before issuing their own disclosures. Of these
68

firms, those that experience larger market reactions to the tax authority’s disclosure are more
likely to follow up with a voluntary disclosure. This result holds in both the first and second year
of tax return disclosure and is mainly driven by large negative market reactions. This suggests
that voluntary disclosures may be issued in an attempt to mitigate both anticipated and actual
reductions in firm value.
I then focus my attention on market reactions to voluntary disclosures. A key limitation
to my study is that I cannot study how the market would have reacted to a firm’s involuntary tax
return disclosure in the absence of a voluntary disclosure. However, I do find that when two
firms have the same difference between their tax return and financial statement values, the firm
that voluntarily discloses additional information experiences a muted market reaction to this
difference. This suggests that voluntary disclosure can enrich a firm’s information environment
in a way that shifts investors’ attention away from the simple difference between the firm’s
financial statement and tax return values. In other words, voluntary disclosures appear to alter the
context of the information which in turn, affects how investors react to the information.
Lastly, I find that when a firm does not voluntarily disclose additional information, the
market reacts to the difference between the firm’s tax return values and its financial statement
values. In the first year, the market reacts strongly to tax liability differences, however, this
reaction is mostly related to increased volatility with no consensus price change. In the second
year, the market reacts strongly to both tax liability and taxable income differences and these
reactions are consistently negative. This suggests that the market may have learned how to value
tax return information over time such that in the second year, it was able to reach a consensus
valuation faster.
69

This paper contributes to our understanding of the relationship between mandatory and
voluntary disclosures as well as the costs and benefits of public tax return disclosure. While
concurrent research finds that public tax return disclosure imposed little to no reputational costs
on public firms, I find that public tax return disclosure did shock firms’ information
environments, which motivated some firms to provide new details on their tax positions.

70

References
Amir, E., & Sougiannis, T. (1999). Analysts’ interpretation and investors’ valuation of tax
carryforwards. Contemporary Accounting Research, 16(1): 1- 33.
Bagnoli, M., & Watts, S. (2007). Financial reporting and supplemental voluntary disclosures.
Journal of Accounting Research, 45(5): 885-913.
Balakrishnan, K., Blouin, J. & Guay, W. (2017). Tax aggressiveness and corporate
transparency. Working Paper.
Beaver, W. (1968). The information content of annual earnings announcements. Journal of
Accounting Research, 6: 67-92.
Beyer, A., Cohen, D., Lys, T., & Walther, B. (2010). The financial reporting environment:
Review of the recent literature. Journal of Accounting and Economics, 50: 296-343.
Bischof, J., & Daske, H. (2013). Mandatory disclosure, voluntary disclosure, and stock market
liquidity: Evidence from the EU bank stress rests. Journal of Accounting Research,
51(3): 997- 1029.
Blume, L., Easley, D., & O’Hara, M. (1994). Market statistics and technical analysis: The role
of volume. The Journal of Finance, 49(1): 153-181.
Bo, E., Slemrod, J., & Thoresen. T. (2015). Taxes on the internet: Deterrence effects of public
disclosure. American Economic Journal: Economic Policy, 7(1): 36-62.
Bozanic, Z., Hoopes, J., Thornock, J., & Williams, B. (2017). IRS attention. Journal of
Accounting Research, 55(1): 79-114.
Bratten, B., Gleason, C., Larocque, S., & Mills, L. (2017). Forecasting taxes: New evidence from
analysts. The Accounting Review, 92(3): 1-29.
71

Chen, S. (2017). Do investors value corporate tax return information? Evidence from Australia.
Working Paper.
Chen, K. & Schoderbek, M. (2000). The 1993 tax rate increase and deferred tax adjustments: A
test of functional fixation. Journal of Accounting Research, 38(1): 23-44.
Comprix, J., Graham, R., & Moore, J. (2011). Empirical evidence on the impact of book-tax
differences on divergence of opinion among investors. Journal of the American
Taxation Association, 33(1): 51-78.
Demeré, P. (2018). Is tax return information useful to equity investors? Working Paper.
Desai, M., & Dharmapala, D. (2006). Corporate tax avoidance and high-powered incentives.
Journal of Financial Economics, 79: 145-179.
Dhaliwal, D., Gleason, C., & Mills, L. (2004). Last-chance earnings management: Using the
tax expense to meet analysts’ forecasts. Contemporary Accounting Research, 21(2):
431-459.
Dhaliwal, D., Goodman, T., Hoffman, P.J., & Schwab, C. (2016). The incidence, valuation, and
management of tax-related reputational costs: Evidence from negative media attention
during periods characterized by high scrutiny of corporate tax avoidance. Working Paper.
Dutta, S., & Trueman, B. (2002). The interpretation of information and corporate disclosure
strategies. Review of Accounting Studies, 7: 75-96.
Dye, R. (1986). Proprietary and nonproprietary disclosures. The Journal of Business, 59(2.1):
331-366.
Dyer, T., Lang, M., Stice-Lawrence, L. (2017). Do managers really guide through the fog? On
the challenges in assessing the causes of voluntary disclosure. Journal of Accounting and
Economics, 62(2-3): 270-276.
72

Dyreng, S., Hoopes, J., & Wilde, J. (2016). Public pressure and corporate tax behavior. Journal
of Accounting Research, 54(1): 147-185.
Einhorn, E. (2005). The nature of the interaction between mandatory and voluntary disclosures.
Journal of Accounting Research, 43(4): 593-621.
Erickson, M., Heitzman, S., & Zhang, X. (2013). Tax-motivated loss shifting. The Accounting
Review, 88(5): 1657-1682.
Frank, M.M., & Rego, S. (2006). Do managers use the valuation allowance account to
manage earnings around certain earnings targets? The Journal of the American
Taxation Association, 28(1): 43-65.
Fishman, M. & Hagerty, K. (2003). Mandatory versus voluntary disclosure in markets with
informed and uninformed customers. Journal of Law, Economics and Organizations,
19(1): 45-63.
Foster, G. (1981). Intra-industry information transfers associated with earnings releases. Journal
of Accounting and Economics, 3: 201-232.
Frank, M. (2009). Discussion of: Inferring U.S. tax liability from financial statement
information. Journal of the American Taxation Association, 31(1): 65-74.
Gallemore, J. & Labro, E. (2015). The importance of the internal information environment for
tax avoidance. Journal of Accounting and Economics, 60: 149-167.
Gallemore, J., Maydew, E., & Thornock, J. (2014). The reputational costs of tax avoidance.
Contemporary Accounting Research, 31(4): 1103-1133.
Gao, F., Wu, J.S., & Zimmerman, J. (2009). Unintended consequences of granting small firms
exemptions from securities regulation: Evidence from the Sarbanes-Oxley act. Journal
of Accounting Research, 47(2): 459- 506.
73

Graham, J. & Mills, L. (2008). Using tax return data to simulate corporate marginal tax rates.
Journal of Accounting and Economics, 46: 366-388.
Graham, J., Raedy, J., & Shackelford, D. (2012). Research in accounting for income taxes.
Journal of Accounting and Economics, 53: 412-434.
Grossman, S. & Stiglitz, J. (1980). On the impossibility of informationally efficient markets.
The American Economic Review, 70(3): 393-408.
Grossman, S.J. (1981). The informational role of warranties and private disclosure about product
quality. The Journal of Law and Economics, 24(3): 461-483.
Guay, W., Samuels, D., & Taylor, D. (2017). Guiding through the fog: financial statement
complexity and voluntary disclosure. Journal of Accounting and Economics, 62 (2-3):
234-269.
Hanlon, M. (2003). What can we infer about a firm’s taxable income from its financial
statements? National Tax Journal, 56(4): 831-863.
Hanlon, M. (2005). The persistence and pricing of earnings, accruals, and cash flows when firms
have large book-tax differences. The Accounting Review, 80(1): 137-166.
Hanlon, M. Laplante, S., & Shevlin, T. (2005). Evidence for the possible information loss of
conforming book income and taxable income. The Journal of Law and Economics, 48(2):
407-442.
Hanlon, M. & Slemrod, J. (2009). What does tax aggressiveness signal? Evidence from stock
price reactions to news about tax shelter involvement. Journal of Public Economics, 93:
126-141.
74

Hasegawa, M., Hoopes, J., Ishida, R., & Slemrod, J. (2013). The effect of public disclosure on
reported taxable income: evidence from individuals and corporations in Japan. National
Tax Journal, 66(3): 571-608.
Heitzman, S., Wasley, C., & Zimmerman, J. (2010). The joint effects of materiality thresholds
and voluntary disclosure incentives on firms’ disclosure decisions. Journal of Accounting
and Economics, 49: 109-132.
Heitzman, S., & Ogneva, M. (2017). Corporate tax planning and stock returns. Working Paper.
Hoopes, J., Robinson, L., & Slemrod, J. (2017). Public tax-return disclosure. Working Paper.
Hume, N. (2013). Australia Eyes Corporate Tax Transparency. The Financial Times (April 3).
https://www.ft.com/content/748c2952-9c49-11e2-9a4b-00144feabdc0
Jung, W., & Kwon, Y. (1988). Disclosure when the market is unsure of information endowment
of managers. Journal of Accounting Research, 26(1):146-153.
Kandel, E., & Pearson, N. (1995). Differential interpretation of public signals and rrade in
speculative markets. Journal of Political Economy, 103(4): 831-872.
Karpoff, J. (1986). A theory of trading volume. The Journal of Finance, 41(5): 1069-1087.
Krull, L. (2004). Permanently reinvested foreign earnings, taxes, and earnings management.
The Accounting Review, 79 (3): 745-767.
Lambert, R., Leuz, C., & Verrecchia, R. (2011). Information asymmetry, information precision,
and the cost of capital. Review of Finance, 16: 1-29.
Lenter, D., Slemrod, J., & Shackelford, D. (2003).Public disclosure of corporate tax return
information: Accounting, economics, and legal perspectives. National Tax Journal,
56(4): 803-830.
75

Leuz, C., & Wysocki, P. (2016). The economics of disclosure and financial reporting
regulation: Evidence and suggestions for future research. Journal of Accounting
Research, 54(2): 525-622.
Lev, B., & Nissim, D. (2004). Taxable income, future earnings, and equity values. The
Accounting Review, 79(4): 1039-1074.
Lisowsky, P. (2009). Inferring U.S. tax liability from financial statement information. Journal of
the American Taxation Association, 31(1): 29-63.
Lundholm, R. (1988). Price-signal relations in the presence of correlated public and private
information. Journal of Accounting Research, 26(1): 107-118.
Luo, Y. (2005). Do insiders learn from outsiders? Evidence from mergers and acquisitions. The
Journal of Finance, LX (4): 1951-1982.
McGill, G., & Outslay, E. (2004). Lost in translation: Detecting tax shelter activity in financial
statements. The National Tax Journal, 57(3): 739- 756.
Miller, B. (2010). The effects of reporting complexity on small and large investor trading. The
Accounting Review, 85(6): 2107-2143.
Mills, L. (1998). Book-tax differences and Internal Revenue Service adjustments. Journal of
Accounting Research, 36(2): 343-356.
Mills, L., Robinson, L., & Sansing, R. (2010). FIN 48 and tax compliance. The Accounting
Review, 85(5): 1721- 1742.
Plumlee, M. (2003). The effect of information complexity on analysts’ use of that information.
The Accounting Review, 78(1): 275-296.
Robinson, L., & Schmidt, A. (2013). Firm and investor responses to uncertain tax benefit
disclosure requirements. Journal of the American Tax Association, 35(2): 85-120.
76

Schrand, C. & Wong, M.H. (2003). Earnings management using the valuation allowance for
deferred tax assets under SFAS No. 109. Contemporary Accounting Research, 20(3):
579-611.
Sletten, E. (2012). The effect of stock price on discretionary disclosure. Review of Accounting
Studies, 17: 96-133.
Suijs, J. (2007). Voluntary disclosure of information when firms are uncertain of investor
response. Journal of Accounting and Economics, 43: 391-410.
Thomas, J., & Zhang, F. (2011). Tax expense momentum. Journal of Accounting Research,
49(3): 791- 821.
Thomas, J., & Zhang, F. (2014). Valuation of tax expense. Review of Accounting Studies, 19:
1436-1467.
Tran, A. (2015). Can taxable income be estimated from financial reports of listed companies in
Australia? Australian Tax Forum (30): 569-594.
Verrecchia, R. (1983). Discretionary disclosure. Journal of Accounting and Economics, 5:179-
194.
Weber, D. (2009). Do analysts and investors fully appreciate the implications of book-tax
differences for future earnings? Contemporary Accounting Research, 26(4): 1175-1206.
Zuo, L. (2016). The informational feedback effect of stock prices in management forecasts.
Journal of Accounting and Economics, 61: 391 -413.
77

Appendix A – Variable Definitions

Variable Definition
TI_Diff (TrTaxInc-(txc/.3))/Assets

TL_Diff (TrTaxLiab-txc)/Assets

TI_Diff2 ((TrTaxInc- (txc/.3))*(PY TrTaxInc/(PY txc/.3))/Assets

TL_Diff2 (TrTaxLiab- (txc*(PY TrTaxLiab/PY txc)))/Assets

TR_Lower Indicator that equals one if TR_Diff (TI_Diff or TL_Diff) is negative and
zero otherwise

TrTaxInc Tax return taxable income disclosed on the ATO's website and added
across all available subsidiaries of a given financial reporting entity

TrTaxLiab Tax return tax liability disclosed on the ATO's website and added across all
available subsidiaries of a given financial reporting entity

Txc Current tax expense hand collected from firms’ annual reports

Pi Pre-tax income from Compustat Global

VolD_Any Hand collected indicator variable that equals 1 if the firm issued
supplemental disclosures referencing their tax return disclosures at any
point in time

VolD_Pre Hand collected indicator variable that equals 1 if the firm issued
supplemental disclosures referencing their tax return disclosures before the
ATO's Disclosure

VolD_Conc Hand collected indicator variable that equals 1 if the firm issued
supplemental disclosures referencing their tax return disclosures on the day
of or the day after the ATO's disclosure

78

VolD_Post Hand collected indicator variable that equals 1 if the firm issued
supplemental disclosures referencing their tax return disclosures more than
one day after the ATO's disclosure

VolD_Pre/Conc Hand collected indicator variable that equals 1 if the firm issued
supplemental disclosures referencing their tax return disclosures before, on
the day of or on the day after the ATO's disclosure

VolD_PY Indicator variable that equals 1 if the firm issued supplemental disclosures
referencing their prior year tax return disclosure

CountTTC Hand collected count of the number of TTC recommended disclosures a
firm makes; 0-6

IncTTC Indicator variable that equals one if a firm issues more TTC disclosures
than it did in the prior year and zero otherwise

TTC Indicator variable that equals one if a firm issues 3 or more of the TTC’s 6
recommended disclosures

EstError |Tax expense related to prior year provision|/(|current tax
expense|+|deferred tax expense|)

DiscOp Indicator that equals one if a firm has any income from discontinued
operations and zero otherwise

R&D Indicator that equals one if a firms has any research and development
expenses and zero otherwise

NumSubs Number of subsidiaries in which a company owns more than 50%

Foreign Indicator that equals one if a firm has any sales outside of Australia and
zero otherwise

Loss Indicator if a firm's pretax income is less than zero and zero otherwise

79

Volume (A firm's daily trading volume scaled by its total shares outstanding)*100

Return Logarithmic daily return*100

ATO_Return Cumulative two day logarithmic return on the day of and the day after the
ATO's disclosure

Neg Indicator that equals one if ATO_Return is less than zero and zero
otherwise

ATODisc Indicator that equals one on the day of and the day after the ATO's
disclosure

ln(Asset) Natural log of total assets collected from Compustat Global

ROA Pretax income divided by total assets, collected from Compustat Global

Leverage Total liabilities scaled by total assets, collected from Compustat Global

PPE Property, plant and equipment scaled by total assets, collected from
Compustat Global

MTB Year End Market Value/Common Book Equity, collected from Compustat
Global

CurrETR txc/pi if pi>0 & winsorized at 0 and 1

CashETR Cash taxes paid/pi

TaxETR TrTaxLiab/pi if pi>0 & winsorized at 0 and 1

TaxETR2 TrTaxLiab/TrTaxInc if TrTaxInc>0

TA_GAAP (Mean GAAP ETR of firms in the same industry and size quintile) -
(Firm’s own GAAP ETR); GAAP ETR is the sum of total tax expense over
80

years t, t-1 and t-2 divided by the sum of pretax income over years t, t-1 and
t-2, winsorized at 0 and 1

NoTaxPd Indicator if a firm had no tax return tax liability and zero otherwise

81

Appendix B – Voluntary Disclosure Examples
Arrium Limited 12/17/2015

82

AMP Limited 12/10/2015
This document serves as an addendum to the AMP 2014 tax report, originally published on 26 March 2015. This
information is to assist with understanding disclosures expected to be made by the Australian Taxation Office (ATO) in
December 2015 on AMP’s tax position in its tax transparency report for our tax year ended 31 December 2013. It is being
released at this time as the timing of the ATO’s disclosure does not coincide with AMP’s financial reporting period.
In 2013, Australia introduced legislation requiring the ATO to publish specific income tax return data for companies with income
of $100 million or more, including AMP Limited and AMP Capital Holdings Limited (AMPCH). These disclosures apply to
AMP’s Australian income tax position and do not cover any foreign taxes.
This income tax return data will contain both tax attributable to shareholder and policyholder related profits and will therefore
differ from the tax paid analysis in the AMP 2014 tax report, which only relates to tax paid on shareholder profits.
The ATO has published documents on its website providing background on its corporate tax transparency report and
acknowledges that for companies that include a life insurance business (eg AMP), the effective tax rate for accounting purposes
will be impacted by the aggregation of shareholder and policyholder tax. The ATO has also noted that differing applicable tax
rates, in particular the 15% rate for superannuation, may give an impression of an artificially low tax paid. These documents also
comment on the impact of certain items, including the research and development tax concession (R&D) and imputation credits.
Tax transparency disclosures for the year ended 31 December 2013 (2013 year)
The first set of data will be released by the ATO in the week commencing 14 December 2015 and will include the following
information from the AMP Limited and AMPCH 2013 Australian income tax returns:

Footnotes:
1 Gross accounting income for AMP and its wholly-owned Australian subsidiaries and their foreign branches as disclosed in the tax returns, prior
to offsetting expenses (such as claims, commissions, employee costs or interest). This is not readily comparable to AMP’s income statement,
which also includes income from foreign subsidiaries and Australian entities that fall outside of the tax consolidated groups.
2 Under Australian tax law, AMP Limited has two classes of taxable income (ordinary class and complying superannuation class). Ordinary class
taxable income (2013: $908m) is taxed at the corporate tax rate of 30%, and complying superannuation class taxable income (2013: $4,031m) is
taxed at a statutory tax rate of 15%. The taxable income reported by the ATO for AMP Limited is the total of both classes of taxable income.
Note – AMP Limited’s tax payable, as a percentage of taxable income, is 7.6% (and 8.1%, including AMP Capital Holdings Limited). This is
largely due to how tax payable is determined for life insurance businesses and is explained further below under the heading ‘Why tax does not
equal 30% of taxable income’. How tax payable is determined for life insurance businesses is also outlined in the ATO’s 2013 transparency
disclosure report.

Why tax does not equal 30% of taxable income

Separate tax treatment for life insurance business
AMP’s tax profile differs to other Australian corporates due to its life insurance and superannuation businesses.
Under the tax rules for life insurance businesses, a significant portion of AMP’s gross income, as reported in its tax returns, is not
subject to tax (not included in taxable income) or is subject to a concessional rate of tax where it relates to policyholder interests.
This includes:
– certain life insurance premiums invested in AMP out of ‘after-tax’ earnings by policyholders which are not subject to further
contributions tax
– income relating to AMP’s pension business is exempt from tax, consistent with the taxation of pension phase earnings in a
superannuation fund, and
– income relating to complying superannuation business which is taxed at 15%, consistent with the taxation of standalone
complying superannuation funds.

Breakdown between the statutory rate of 30% and AMP’s tax rate
The difference between AMP’s tax payable and a tax payable rate of 30% is a result of:
83

– Complying superannuation class income taxed at a statutory rate of 15%, consistent with the rate of taxation on complying
superannuation funds. This contributed a $604m reduction to tax from a 30% tax rate in 2013.
– Imputation credits received by AMP from franked dividends, which reduce its tax payable. These imputation credits arise on
dividends received by AMP that have been paid from taxed profits and prevents double taxation of these profits. This contributed
a $389m reduction to tax from a 30% tax rate in 2013.
– Foreign tax offsets reflecting payments of foreign tax by AMP. These are allowed as a credit against Australian tax to prevent a
double taxation. This contributed a $47m reduction to tax from a 30% tax rate in 2013.
– AMP receives R&D tax offsets from conducting R&D activities in Australia. Under the relevant Australian tax law, R&D
expenditure was not deductible and gave rise to a 40% tax offset, which is equivalent to a 10% tax concession. This contributed a
$66m reduction to tax from a 30% tax rate in 20131.
The nature of AMP’s business means that under Australian tax laws its Australian tax payable ($410m for 2013) is lower than
30% of taxable income (30% tax payable would equate to $1,516m for 2013).

84

Footnotes:
1 Shareholder income is subject to tax at a 30% tax rate.
2 Policyholder income is subject to tax at either 30%, 15% or exempt from tax (as discussed above).
3 Tax offsets include R&D offsets of $66m, imputation credits received by AMP on dividends and foreign tax offsets. The $66m of R&D offsets
is the total of the 40% benefit under the R&D concession (this includes the $16m disclosed in the AMP’s 2014 tax report (in respect of the 2013
year) which represents the excess over the notional 30% deduction benefit).
4 This differs to the amount of tax paid shown in the table on page 4 of AMP’s 2014 tax report, due to refunds received by AMP from an
amendment request in 2015.

Difference between tax payable and income tax expense

The tax payable disclosed in the tax returns of AMP Limited and AMPCH differs to the shareholder income tax expense
disclosed in AMP’s accounts and 2014 tax report ($218m for 2013) for reasons including:
– the total tax payable includes tax referable to policyholder income which is not included in shareholder tax expense
– timing differences in respect of the recognition of income and expenses for tax and accounting purposes (eg gains on
investments are not included in taxable income until the investments are realised)
– AMP’s shareholder income tax expense includes foreign taxes, as well as Australian tax referable to entities that are not
included in the AMP Limited and AMPCH consolidated tax groups (eg income of AMP’s foreign branches already taxed
overseas are not subject to further Australian tax).
Further, the tax payable attributable to policyholders disclosed above represents the cash tax paid in respect of the 2013 income
year. The policyholder income tax expense disclosed in AMP’s 2013 accounts ($564m) represents both current and deferred tax
expense (expected to be paid in future income years).
85

Figure 1
Voluntary Disclosure Dates in Relation to the ATO's Tax Return Disclosures
This figure plots the number of firms that make a voluntary disclosure by event day relative to the ATO
disclosure date (T = 0) for 2015 and 2016.
0
2
4
6
8
10
Unknown
Before
More than
2 weeks
Before
-3 : -10 -2 -1 0 1 2 3 : 10 More than
2 weeks
After
Unknown
After
Number of Voluntary Disclosures
Disclosure Date Relative to the ATO Disclosure Date
Voluntary Disclosure Dates
2015 2016
86

Figure 2
Tax Return Effective Tax Rates
This figure plots firms' effective tax rates (tax liability/taxable income) according to their Australian tax
returns for 2015 and 2016. The maximum effective tax rate, 30%, is in line with Australia's statutory tax
rate. Firms are grouped based on effective tax rates in 5% intervals. Firms in a given bucket have effective
tax rates that are at least as big as the value stated but smaller than the value stated for the next bucket.
Firms in the 30% bucket have effective tax rates exactly equal to 30%.

0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 0.05 0.1 0.15 0.2 0.25 0.3
Percentage of Firms
Tax Liability/Taxable Income
Tax Return Tax Rate
2015 2016
87

Table 1
Sample Selection
This table describes my sample selection. I start with entities whose tax returns are disclosed by the
ATO in either the first or the second year of tax return disclosure. I then drop private companies,
foreign companies, and unlisted public companies**. I drop one entity in both years whose website is
inaccessible and prevents me from obtaining necessary data. I then match the remaining tax returns to
publicly listed parent companies. Some companies that are consolidated for financial reporting
purposes file more than one tax return with total income that exceeds the ATO's disclosure threshold.
My analyses are done at the parent company level but include the tax return data of all relevant
subsidiaries. Parent companies are dropped from my sample if they do not have necessary financial
statement data or stock data.
2015 2016
Total
Observations
Total Unique
Firms
Separate Tax Returns Disclosed 1859 1904 3763 2121
Less:
Australian Private Companies (321) (325) (646) (382)
Foreign Owned (985) (998) (1983) (1103)
Unlisted Australian Public Companies (127) (151) (278) (163)
Website Inaccessible (1) (1) (2) (1)
Total Tax Returns in Sample 425 429 854 472
Less:
Subsidiaries whose Parent Company's
Tax Return is also Disclosed (66) (66) (132) (78)
Australian Parent Companies 359 363 722 394
Less:
Acquired by Non Australian Firm (10) (11) (21) (11)
Acquired by Australian Firm* (10) (10) (20) (10)
Went Private (2) (5) (7) (3)
Liquidated (0) (1) (1) (1)
No Stock Data until 2016 (1) (0) (1) (0)
Final Sample 336 336 672 369

* These firms and their subsidiaries remain in the sample but are now part of a different parent
company.
**All firms with a certain number of shareholders are classified in Australia as public whether or not
they are publicly traded.

88

Table 2
Summary Statistics
This table displays descriptive statistics for all variables used in my analyses. All amounts are in millions
of AUD and all variables are as defined in Appendix A. All continuous variables are winsorized at 1%
except ETRs, which are winsorized at 0 and 1. (0, 1) indicates an indicator variable.
Variable Obs Mean Std Dev P1 P25 P50 P75 P99
Financial Statement - Tax Return Differences
|TI_Diff| 672 0.025 0.042 0.000 0.002 0.011 0.030 0.287
|TL_Diff| 672 0.007 0.011 0.000 0.001 0.003 0.008 0.072
TI_Diff 672 0.001 0.041 -0.147 -0.012 0.000 0.010 0.172
TL_Diff 672 -0.004 0.011 -0.053 -0.006 -0.001 0.000 0.030
TI_Lower (0,1) 672 0.485 0.500
TL_Lower (0,1) 672 0.677 0.468
Voluntary Disclosure Frequencies
VolD_Any_2015 (0,1) 336 0.101 0.302
VolD_Pre_2015 (0,1) 336 0.051 0.219
VolD_Conc_2015 (0,1) 336 0.027 0.162
VolD_Post_2015 (0,1) 336 0.024 0.153
VolD_Any_2016 (0,1) 336 0.098 0.298
VolD_Pre_2016 (0,1) 336 0.051 0.219
VolD_Conc_2016 (0,1) 336 0.024 0.153
VolD_Post_2016 (0,1) 336 0.024 0.153
Market Variables
Volume 43,831 0.237 0.287 0.000 0.030 0.157 0.326 1.581
|Return| 43,831 0.015 0.019 0.000 0.002 0.009 0.021 0.101
Return 43,831 0.000 0.023 -0.078 -0.010 0.000 0.009 0.077
Financial Statement Variables
ln(Asset) 672 6.937 1.797 3.694 5.663 6.652 7.936 13.42
ROA 672 0.030 0.157 -0.786 0.009 0.052 0.093 0.415
Leverage 672 0.488 0.216 0.055 0.347 0.467 0.617 1.064
PPE 672 0.253 0.254 0.000 0.024 0.165 0.440 0.862
MTB 672 2.173 2.318 0.179 0.890 1.439 2.358 12.87
Tax Planning Variables
CurrETR 530 0.264 0.202 0.000 0.144 0.267 0.332 1.000
CashETR 530 0.243 0.227 0.000 0.051 0.231 0.331 1.000
TaxETR 530 0.188 0.191 0.000 0.000 0.185 0.288 1.000
TaxETR2 672 0.156 0.129 0.000 0.000 0.202 0.286 0.300
TA_GAAP 672 -0.011 0.187 -0.692 -0.080 -0.023 0.102 0.308
NoTaxPd (0,1) 672 0.351 0.478
Financial Statement - Tax Return Reporting Difference Drivers
DiscOp (0,1) 672 0.113 0.317
Foreign (0,1) 672 0.327 0.470
R&D (0,1) 672 0.222 0.416
Loss (0,1) 672 0.211 0.409
EstError 672 0.116 0.325 0.000 0.000 0.012 0.066 1.881
NumSubs 672 37.56 99.23 0.000 7.000 16.00 39.00 280.0

89

Table 3
Distribution of Variables by Quintile of Tax Return Differences
In this table, I sort firms based on quintiles of |TR_Diff|. I then calculate the average value of several
variables within each quintile and test for any difference between means in the two extreme quintiles. I
further sort firms based on whether or not they issue voluntary disclosures and calculate market returns on
the day of and the day after tax return disclosures. |TR_Diff| is calculated with |TI_Diff| in Panel A and
|TL_Diff| in Panel B. p-values (one sided) appear in brackets and test for differences in the means of the
extreme quintiles of tax return differences. All variables are as defined in Appendix A.
Panel A. Quintiles of |TI_Diff|
|TI_Diff| Quintile
Variable Q1 Q2 Q3 Q4 Q5 Diff Q5 - Q1 Diff P-Value
|TI_Diff| 0.000 0.004 0.011 0.026 0.083 0.083 [0.00]***
|TI_Diff| 2016 0.007 0.011 0.015 0.022 0.058 0.051 [0.00]***
|TL_Diff| 0.001 0.002 0.004 0.007 0.019 0.018 [0.00]***
TI_Lower 0.328 0.531 0.683 0.575 0.417 0.088 [0.15]
R&D 0.141 0.141 0.190 0.397 0.278 0.137 [0.03]**
Foreign 0.219 0.297 0.238 0.521 0.361 0.142 [0.03]**
EstError 0.027 0.041 0.035 0.050 0.072 0.045 [0.01]***
NumSubs 22.78 30.08 44.71 36.16 34.42 11.64 [0.07]*
DiscOp 0.078 0.125 0.111 0.082 0.139 0.061 [0.13]
Loss 0.265 0.156 0.111 0.192 0.194 -0.071 [0.84]
VolD_Any 0.063 0.106 0.048 0.096 0.181 0.117 [0.02]**
|Return Day of And Day After Tax Return Disclosure|
All 0.021 0.021 0.027 0.026 0.028 0.007 [0.05]*
VolD_Any=0 0.021 0.019 0.028 0.024 0.029 0.008 [0.07]*
VolD_Any=1 0.015 0.030 0.018 0.041 0.025 0.010 [0.12]
|Return Day 0| + |Return Day +1|
All 0.032 0.036 0.040 0.035 0.040 0.008 [0.07]*
VolD_Any=0 0.032 0.033 0.041 0.034 0.042 0.010 [0.07]*
VolD_Any=1 0.016 0.061 0.033 0.040 0.029 0.013 [0.07]*
Panel B. Quintiles of |TL_Diff|
|TL_Diff| Quintile
Variable Q1 Q2 Q3 Q4 Q5 Diff Q5 - Q1 Diff P-Value
|TL_Diff| 0.000 0.001 0.003 0.007 0.023 0.023 [0.00]***
|TL_Diff| 2016 0.001 0.002 0.004 0.007 0.016 0.014 [0.00]***
|TI_Diff| 0.010 0.008 0.014 0.026 0.070 0.060 [0.00]***
TL_Lower 0.274 0.704 0.855 0.811 0.788 0.514 [0.00]***
R&D 0.081 0.155 0.254 0.319 0.352 0.271 [0.00]***
Foreign 0.194 0.282 0.333 0.449 0.394 0.201 [0.01]***
EstError 0.023 0.034 0.047 0.042 0.079 0.056 [0.00]***
NumSubs 17.00 27.46 44.89 39.48 38.90 21.90 [0.00]***
DiscOp 0.065 0.099 0.095 0.116 0.155 0.090 [0.05]*
Loss 0.258 0.155 0.111 0.232 0.169 -0.089 [0.89]
VolD_Any 0.048 0.127 0.095 0.101 0.127 0.078 [0.06]*
|Return Day of And Day After Tax Return Disclosure|
All 0.017 0.027 0.023 0.030 0.025 0.008 [0.04]**
VolD_Any=0 0.017 0.027 0.023 0.029 0.025 0.008 [0.05]*
VolD_Any=1 0.022 0.028 0.019 0.038 0.027 0.004 [0.35]
|Return Day 0| + |Return Day +1|
All 0.029 0.035 0.037 0.040 0.041 0.012 [0.03]**
VolD_Any=0 0.030 0.033 0.037 0.040 0.043 0.013 [0.03]**
VolD_Any=1 0.023 0.050 0.032 0.040 0.031 0.008 [0.22]

90

Table 4

Predicting Tax Return Differences
This table reports the results from OLS regressions that describe what causes differences between a
firm's tax return and financial statement values (|TR_Diff|). In Columns 1 and 2, |TR_Diff| measures the
absolute difference between a firm's actual taxable income and that implied by its financial statements
(grossed up current tax expense). While in columns 3 and 4, |TR_Diff| measures the absolute difference
between a firm's actual tax liability and its financial statement current tax expense. In all regressions,
the dependent variable is multiplied by 100 and relates to the first two years of public tax return
disclosure, 2015 and 2016. Robust standard errors are reported in brackets below each coefficient. All
variables are as defined in Appendix A. ***, **, and * denote statistical significance at the .01, .05. and
.1 levels respectively.
Dependent Variable |TI_Diff|

TI_Diff

|TL_Diff|

TL_Diff

Variable (1)

(2)

(3)

(4)
EstError 0.473

-0.702

0.235* -0.319***

[0.510]

[0.472]

[0.127] [0.113]
R&D 0.608* -0.579

0.306*** -0.433***

[0.349] [0.406]

[0.105] [0.102]
TA_GAAP 1.202** 0.188

0.266* -0.133

[0.515] [0.526]

[0.145] [0.136]
Foreign 0.868** -0.821**

0.250** -0.296***

[0.384] [0.361]

[0.101] [0.091]
NumSubs -1.840*** -1.176**

-0.264 -0.094

[0.603] [0.569]

[0.171] [0.150]
DiscOps 1.470* 0.210

0.357* 0.052

[0.781] [0.643]

[0.206] [0.164]
Loss 0.240 0.598

0.088 0.449***

[0.459] [0.404]

[0.121] [0.096]

Constant 2.016*** 1.293***

0.349** -0.001

[0.547]

[0.483]

[0.145]

[0.127]

Observations 672

672

672

672

R
2
5% 3%

5%

9%

91

Table 5
Tax Return Differences and Voluntary Disclosure
This table presents the results of probit regressions that predict the probability a firm voluntarily issues supplementary tax disclosures in anticipation
of or in response to the ATO's public tax return disclosures. The main variables of interest are the absolute difference between a firm's tax return values
and its financial statement tax values (|TR_Diff|) and a firm's unsigned cumulative return over the day of and day after the ATO's tax return disclosure
(|ATO_Return|). In Columns 1, 2, and 3, |TR_Diff| measures the absolute difference between a firm's actual taxable income and that implied by its
financial statements (grossed up current tax expense). While in columns 4, 5, and 6, |TR_Diff| measures the absolute difference between a firm's actual
tax liability and its financial statement current tax expense. Panel A presents results from the first year of voluntary disclosure, 2015, while Panel B
presents results from the second year of voluntary disclosure, 2016. Panel C shows which underlying drivers of |TR_Diff| are associated with voluntary
disclosure. Lastly, Panel D presents results relating to voluntary disclosure and the sign of TR_Diff or ATO_Return. Robust standard errors are reported
in brackets below each coefficient. All variables are decile ranked, scaled from 0-1 and as defined in Appendix A. ***, **, and * denote statistical
significance at the .01, .05. and .1 levels respectively.
Panel A: Voluntary Disclosure Responses to Unsigned Differences and Unsigned Returns in the Initial Year of Tax Return Disclosure, 2015
|TR_Diff| = |TI_Diff| |TR_Diff| = |TL_Diff|
Dependent Variable VolD_Any VolD_Pre/Conc VolD_Post VolD_Any VolD_Pre/Conc VolD_Post
Variable (1) (2) (3) (4) (5) (6)
|TR_Diff| 1.04*** 0.87** 1.35* 0.77** 0.71* 0.73
[0.39] [0.41] [0.82] [0.36] [0.37] [0.56]
|ATO_Return| 0.48 0.30 1.20** 0.52 0.30 1.37**
[0.37] [0.39] [0.56] [0.36] [0.38] [0.60]
NoTaxPd 0.24 0.13 0.63 0.27 0.17 0.62
[0.24] [0.26] [0.45] [0.23] [0.25] [0.43]
TA_GAAP 0.00 -0.22 1.30* 0.09 -0.17 1.45*
[0.40] [0.42] [0.77] [0.40] [0.42] [0.82]
Ln(Asset) 2.96*** 2.83*** 2.49*** 2.90*** 2.80*** 2.37***
[0.58] [0.68] [0.67] [0.57] [0.68] [0.66]
ROA -0.01 -0.23 1.59 0.03 -0.19 1.51
[0.50] [0.55] [1.09] [0.49] [0.54] [0.99]
Leverage -0.56 -0.62 -0.14 -0.68* -0.72 -0.28
[0.40] [0.44] [0.68] [0.40] [0.45] [0.65]
PPE -0.29 -0.13 -1.13 -0.32 -0.14 -1.20
[0.38] [0.40] [0.87] [0.37] [0.39] [0.83]
MTB 0.18 0.02 0.33 0.22 0.03 0.50
[0.44] [0.49] [0.68] [0.45] [0.50] [0.69]
Constant -3.91*** -3.43*** -6.82*** -3.73*** -3.32*** -6.42***
[0.64] [0.67] [1.29] [0.62] [0.66] [1.11]
Observations 336 328 310 336 328 310
Pseudo R
2
23% 22% 32% 22% 21% 29%
92

Table 5 Continued
Panel B: Voluntary Disclosure Responses to Unsigned Differences and Unsigned Returns in the Second Year of Tax Return Disclosure, 2016
|TR_Diff2| = |TI_Diff2| |TR_Diff| = |TL_Diff2|
Dependent Variable VolD_Any VolD_Pre/Conc VolD_Post VolD_Any VolD_Pre/Conc VolD_Post
Variable (1) (2) (3) (4) (5) (6)
|TR_Diff2| 0.51 0.05 1.51 -0.04 -0.63 0.22
[0.65] [0.76] [1.24] [0.81] [1.00] [1.32]
|ATO_Return| 0.77 0.35 1.87** 0.90 0.44 2.37***
[0.65] [0.60] [0.74] [0.65] [0.59] [0.87]
NoTaxPd -0.70 -0.70 -0.88 -0.83 -1.04 -1.07
[0.51] [0.49] [0.85] [0.64] [0.72] [0.80]
TA_GAAP 0.08 0.16 -0.84 0.11 0.23 -0.74
[0.60] [0.55] [1.02] [0.60] [0.55] [1.05]
ln(Asset) 1.53** 1.59** 1.48 1.56** 1.60** 1.30
[0.75] [0.72] [0.98] [0.74] [0.72] [0.80]
ROA -0.06 -0.17 0.42 -0.01 -0.06 -0.05
[0.74] [0.71] [1.18] [0.74] [0.69] [1.21]
Leverage -0.31 -0.27 -0.26 -0.44 -0.31 -0.87
[0.72] [0.53] [0.59] [0.70] [0.53] [0.57]
PPE -0.13 0.28 -1.90** -0.08 0.34 -1.81**
[0.54] [0.62] [0.82] [0.54] [0.61] [0.88]
MTB 0.36 0.65 0.20 0.46 0.63 0.58
[0.82] [0.83] [1.22] [0.82] [0.81] [1.17]
VolD_PY 3.16*** 3.07*** 3.38*** 3.15*** 3.13*** 3.41***
[0.41] [0.40] [0.67] [0.41] [0.44] [0.68]
Constant -3.70*** -3.79*** -4.74*** -3.50*** -3.48*** -3.80***
[1.05] [0.93] [1.46] [1.10] [1.04] [1.10]
Observations 336 328 311 336 328 311
Pseudo R
2
66% 67% 62% 66% 67% 60%

93

Table 5 Continued
Panel C: Voluntary Disclosure Responses to the Underlying Drivers of Unsigned Differences in
the Initial Year of Tax Return Disclosure, 2015
Dependent Variable: VolD_Any
|TR_Diff| = |TI_Diff| |TR_Diff| = |TL_Diff|
Variable (1) (2) (3) (4)
|TR_Diff| 0.16 -0.68 -0.17 -0.50
[0.57] [0.92] [0.52] [0.79]
|ATO_Return| 0.29 0.39 0.46 0.62
[0.40] [0.43] [0.38] [0.40]
|TR_Diff|* R&D 4.79** 5.47** 3.36*** 3.39***
[1.97] [2.50] [1.24] [1.23]
|TR_Diff|* Foreign 0.77 0.88 0.62 0.76
[0.83] [0.92] [0.81] [0.86]
|TR_Diff|* Discop -2.31 -2.44*
[1.45] [1.29]
|TR_Diff|* TA_GAAP 1.96 0.78
[1.50] [1.32]
R&D -3.57* -4.14* -2.31** -2.20**
[1.64] [2.14] [0.97] [0.97]
Foreign -0.34 -0.32 -0.25 -0.29
[0.56] [0.62] [0.53] [0.56]
Discop 0.35 0.44
[0.92] [0.75]
TA_GAAP 0.07 -1.09 0.18 -0.27
[0.42] [0.94] [0.41] [0.83]
NoTaxPd 0.00 0.20 0.14 0.30
[0.27] [0.26] [0.25] [0.25]
Ln(Asset) 3.11*** 3.39*** 2.98*** 3.28***
[0.60] [0.63] [0.58] [0.61]
ROA -0.27 -0.23 -0.05 -0.01
[0.57] [0.57] [0.54] [0.54]
Leverage -0.63 -0.54 -0.66 -0.70*
[0.42] [0.42] [0.41] [0.42]
PPE -0.25 -0.16 -0.30 -0.33
[0.39] [0.41] [0.38] [0.39]
MTB -0.13 0.00 -0.05 0.08
[0.48] [0.51] [0.46] [0.50]
Constant -3.18*** -3.21*** -3.20*** -3.43***
[0.77] [0.81] [0.72] [0.78]
Observations 336 336 336 336
Pseudo R
2
29% 33% 26% 30%

94

Table 5 Continued
Panel D: Voluntary Disclosure Responses to Signed Differences and Signed Returns in the Initial Year of Tax Return Disclosure, 2015
|TR_Diff| = |TI_Diff| |TR_Diff| = |TL_Diff|
Dependent Variable VolD_Any VolD_Pre/Conc VolD_Post VolD_Any VolD_Pre/Conc VolD_Post
Variable (1) (2) (3) (4) (5) (6)
|TR_Diff| 1.62*** 1.71*** 1.21 0.74 1.01 0.13
[0.54] [0.56] [1.13] [0.60] [0.63] [1.09]
|TR_Diff|*TR_Lower -1.28 -1.68* 0.92 0.02 -0.51 2.60*
[0.86] [0.97] [1.44] [0.80] [0.85] [1.55]
|ATO_Return| 0.75* 0.76 0.54 0.65 0.56 0.72
[0.44] [0.46] [0.55] [0.40] [0.41] [0.53]
|ATO_Return|*Neg -0.96 -1.97** 5.54* -0.54 -1.43 4.91*
[0.84] [0.88] [3.01] [0.81] [0.92] [2.58]
TR_Lower 0.59 0.83 -1.01 0.04 0.53 -2.29**
[0.56] [0.60] [1.04] [0.47] [0.52] [1.02]
Neg 0.67 1.19** -4.59* 0.35 0.76 -3.95*
[0.54] [0.54] [2.64] [0.51] [0.53] [2.25]
NoTaxPd 0.34 0.21 0.51 0.28 0.18 0.37
[0.24] [0.25] [0.48] [0.23] [0.25] [0.49]
TA_GAAP -0.03 -0.25 1.53* 0.09 -0.15 1.58*
[0.41] [0.44] [0.85] [0.40] [0.43] [0.83]
Ln(Asset) 3.10*** 2.96*** 2.45*** 2.90*** 2.77*** 2.24***
[0.60] [0.70] [0.70] [0.57] [0.68] [0.68]
ROA 0.12 -0.11 1.52* 0.02 -0.19 1.38*
[0.52] [0.55] [1.23] [0.52] [0.57] [1.17]
Leverage -0.45 -0.44 0.07 -0.69* -0.76* 0.07
[0.41] [0.47] [0.73] [0.40] [0.46] [0.74]
PPE -0.23 -0.01 -1.12 -0.33 -0.09 -1.26
[0.37] [0.40] [0.80] [0.36] [0.38] [0.89]
MTB 0.24 0.04 0.46 0.19 -0.09 0.42
[0.49] [0.55] [0.76] [0.46] [0.50] [0.79]
Constant -4.68*** -4.52*** -6.39*** -3.82*** -3.75*** -5.39***
[0.85] [0.96] [1.51] [0.75] [0.86] [1.10]
Observations 336 328 310 336 328 310
Pseudo R
2
25% 25% 36% 22% 23% 36%
95

Table 6
Market Reactions to the First Year of Tax Return Disclosures, 2015
This table provides the results from estimating equation (2). The dependent variable is the absolute value of logarithmic daily stock returns in Panel A, daily trading
volume scaled by total shares outstanding in Panel B, and logarithmic daily stock returns in Panel C. The independent variables are indicator variables that represent the day
of and the day after the ATO's tax return disclosure (ATODisc), if a firm voluntarily disclosed tax return supplements before (VolD_Pre), concurrent with (VolD_Conc), or
after the ATO's disclosure (VolD_Post), and either before or concurrent with the ATO’s disclosure (VolD_Pre/Conc). The two continuous independent variables, the
absolute value of the difference between a firm's actual tax liability and its current tax expense(|TL_Diff|) and the absolute value of the difference between a firm's taxable
income and its grossed up current tax expense (|TI_Diff|), are decile ranked and scaled between 0 and 1. All variables are defined in Appendix A. All regressions relate to the
first year of tax return disclosure and are estimated from November 1, 2015-January 31, 2016 surrounding tax return disclosures in December, 2015. The table reports
coefficients for the main variables of interest and omits indicator variables that represent constant differences. Standard errors are clustered by day and reported in brackets
below each coefficient. ***, **, and * represent statistical significance at 1%, 5% and 10%, respectively.
Panel A: Unsigned Returns
Dependent Variable= |Return|
|TR_Diff|= |TI_Diff| |TR_Diff|= |TL_Diff|
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.03 -0.05 -0.04 -0.08 -0.13 -0.13
[0.26] [0.27] [0.27] [0.14] [0.15] [0.19]
ATODisc*|TR_Diff| 0.28 0.43 0.42 0.47*** 0.57*** 0.59***
[0.33] [0.37] [0.37] [0.13] [0.16] [0.21]
ATODisc*|TR_Diff|*VolD_PreConc -1.66*** -1.64***
[0.45] [0.56]
ATODisc*|TR_Diff|*VolD_Pre -1.09*** -1.58**
[0.33] [0.62]
ATODisc*|TR_Diff|*VolD_Conc -4.43*** -6.23***
[1.16] [1.60]
ATODisc*|TR_Diff|*VolD_Post 0.57 -0.29
[0.36] [1.79]
ATODisc*VolD_Pre/Conc 0.72*** 0.68***
[0.19] [0.25]
ATODisc*VolD_Pre 0.57** 0.84***
[0.23] [0.20]
ATODisc*VolD_Conc 2.82*** 3.92***
[1.03] [1.33]
ATODisc*VolD_Post -0.51 0.13
[0.37] [1.33]
Constant 1.54*** 1.54*** 1.54*** 1.54*** 1.53*** 1.53***
[0.06] [0.06] [0.06] [0.06] [0.06] [0.06]
Fixed Effects |TR_Diff|

|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|

|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD
N 21,742 21,742 21,742 21,742 21,742 21,742
R
2
0.10% 0.24% 0.48% 0.12% 0.26% 0.64%
96

Table 6 Continued
Panel B: Volume
Dependent Variable= Volume
|TR_Diff|= |TI_Diff| |TR_Diff|= |TL_Diff|
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.13*** 0.11*** 0.11*** 0.12*** 0.10*** 0.10***
[0.01] [0.02] [0.02] [0.01] [0.01] [0.01]
ATODisc*|TR_Diff| 0.03 0.05 0.05 0.05*** 0.07*** 0.07***
[0.03] [0.04] [0.04] [0.01] [0.02] [0.02]
ATODisc*|TR_Diff|*VolD_PreConc -0.21** -0.14
[0.10] [0.18]
ATODisc*|TR_Diff|*VolD_Pre -0.26 -0.05
[0.18] [0.22]
ATODisc*|TR_Diff|*VolD_Conc -0.21** -0.62***
[0.08] [0.11]
ATODisc*|TR_Diff|*VolD_Post -0.21* -0.25
[0.11] [0.20]
ATODisc*VolD_Pre/Conc 0.22** 0.16
[0.09] [0.13]
ATODisc*VolD_Pre 0.31** 0.18
[0.13] [0.15]
ATODisc*VolD_Conc 0.33*** 0.59***
[0.04] [0.10]
ATODisc*VolD_Post 0.26** 0.27*
[0.10] [0.16]
Constant 0.19*** 0.18*** 0.18*** 0.17*** 0.17*** 0.16***
[0.01] [0.01] [0.01] [0.01] [0.01] [0.01]
Fixed Effects |TR_Diff| |TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff| |TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD
N 21,742 21,742 21,742 21,742 21,742 21,742
R
2
1.44% 2.30% 3.79% 1.90% 2.54% 4.16%

97

Table 6 Continued
Panel C: Signed Returns
Dependent Variable= Return
|TR_Diff|=|TI_Diff| |TR_Diff|=|TL_Diff|
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.61** 0.60*** 0.59*** 0.68*** 0.71*** 0.76***
[0.26] [0.20] [0.21] [0.17] [0.12] [0.11]
ATODisc*|TR_Diff| -0.13 -0.19 -0.15 -0.27 -0.38 -0.45
[0.28] [0.35] [0.32] [0.45] [0.52] [0.54]
ATODisc*|TR_Diff|*VolD_PreConc 0.47 1.75
[1.12] [1.41]
ATODisc*|TR_Diff|*VolD_Pre 1.78* 3.54***
[1.05] [0.80]
ATODisc*|TR_Diff|*VolD_Conc -1.75 -0.52
[1.10] [3.66]
ATODisc*|TR_Diff|*VolD_Post -0.39 2.59***
[0.95] [0.81]
ATODisc*VolD_Pre/Conc 0.20 -0.56
[0.91] [1.06]
ATODisc*VolD_Pre -0.32 -1.29*
[0.97] [0.72]
ATODisc*VolD_Conc 1.17 0.39
[0.73] [2.32]
ATODisc*VolD_Post -0.18 -2.14**
[0.44] [0.82]
Constant -0.09 -0.09 -0.10 -0.11 -0.12 -0.12
[0.09] [0.09] [0.09] [0.09] [0.08] [0.08]
Fixed Effects |TR_Diff|
|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD |TR_Diff|
|TR_Diff|, VolD,
|TR_Diff|*VolD
|TR_Diff|, VolD,
|TR_Diff|*VolD
N 21,742 21,742 21,742 21,742 21,742 21,742
R
2
0.15% 0.16% 0.18% 0.14% 0.16% 0.21%

98

Table 7
Market Reactions to the First Year of Tax Return Disclosure Dependent on the Sign of TL_Diff and the Timing of Voluntary Disclosures
This table presents the results of estimating equation (2) separately for firms whose tax liability is higher than its current tax expense (TR_Lower=0), and firms whose tax
liability is lower than its current tax expense (TL_Lower=1). The dependent variable is the absolute value of logarithmic daily stock returns in Panel A, daily trading volume
scaled by total shares outstanding in Panel B, and logarithmic daily stock returns in Panel C. The independent variables are indicator variables that represent the day of and
the day after the ATO's tax return disclosure (ATODisc), if a firm voluntarily disclosed tax return supplements before (VolD_Pre), concurrent with (VolD_Conc), or after
the ATO's disclosure (VolD_Post), and either before or concurrent with the ATO’s disclosure (VolD_Pre/Conc), as well as the decile ranked absolute value of the difference
between a firm's actual tax liability and its current tax expense, scaled between 0 and 1 (|TL_Diff|). All variables are defined in Appendix A. All regressions relate to the first
year of tax return disclosure and are estimated from November 1, 2015-January 31, 2016 surrounding tax return disclosure in December 2015. The table reports coefficients
for the main variables of interest and omits indicator variables that represent constant differences. Standard errors are clustered by day and reported in brackets below each
coefficient. ***, **, and * represent statistical significance at 1%, 5% and 10%, respectively.
Panel A: Unsigned Returns
Dependent Variable: |Return|
TR_Lower=0 TR_Lower=1
Variable (1) (2) (3) (4) (5) (6)
ATODisc -0.21 -0.31 -0.33 0.06 0.08 0.08
[0.21] [0.21] [0.27] [0.08] [0.09] [0.10]
ATODisc*|TL_Diff| 0.74*** 0.94*** 1.01*** 0.26*** 0.27*** 0.26***
[0.25] [0.25] [0.34] [0.05] [0.07] [0.09]
ATODisc*|TL_Diff|*VolD_Pre/Conc -5.76*** -0.18
[0.39] [0.66]
ATODisc*|TL_Diff|*VolD_Pre -3.72*** -0.94
[1.33] [0.71]
ATODisc*|TL_Diff|*VolD_Conc -10.09*** -2.06
[1.17] [1.89]
ATODisc*|TL_Diff|*VolD_Post -1.95 1.14
[2.26] [3.00]
ATODisc*VD_Pre/Conc 3.70*** -0.38
[0.24] [0.29]
ATODisc*VD_Pre 1.59* 0.61***
[0.87] [0.22]
ATODisc*VD_Conc 8.03*** 0.65
[1.26] [1.46]
ATODisc*VD_Post 0.51 -0.88
[1.41] [2.45]
Constant 1.60*** 1.59*** 1.60*** 1.43*** 1.42*** 1.42***
[0.08] [0.08] [0.08] [0.06] [0.06] [0.06]
Fixed Effects |TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
N 6,695 6,695 6,695 15,047 15,047 15,047
R
2
0.24% 0.45% 0.83% 0.17% 0.53% 1.01%
99

Table 7 Continued
Panel B: Volume
Dependent Variable= Volume
TR_Lower=0 TR_Lower=1
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.11*** 0.11*** 0.09*** 0.14*** 0.10*** 0.10***
[0.01] [0.01] [0.01] [0.01] [0.01] [0.01]
ATODisc*|TL_Diff| 0.00 0.01 0.03 0.04*** 0.08*** 0.07***
[0.02] [0.02] [0.03] [0.01] [0.02] [0.02]
ATODisc*|TL_Diff|*VolD_Pre/Conc -0.36*** -0.15
[0.09] [0.23]
ATODisc*|TL_Diff|*VolD_Pre 0.21 -0.21
[0.19] [0.28]
ATODisc*|TL_Diff|*VolD_Conc -1.04*** -0.27***
[0.21] [0.08]
ATODisc*|TL_Diff|*VolD_Post -0.28 -0.69
[0.31] [0.75]
ATODisc*VD_Pre/Conc 0.12** 0.23
[0.05] [0.18]
ATODisc*VD_Pre -0.25* 0.35*
[0.13] [0.20]
ATODisc*VD_Conc 0.80*** 0.37***
[0.13] [0.08]
ATODisc*VD_Post 0.26* 0.65
[0.15] [0.64]
Constant 0.18*** 0.18*** 0.17*** 0.16*** 0.16*** 0.16***
[0.01] [0.01] [0.01] [0.01] [0.01] [0.01]
Fixed Effects |TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
N 6,695 6,695 6,695 15,047 15,047 15,047
R
2
1.39% 3.13% 4.76% 2.04% 2.45% 4.08%

100

Table 7 Continued
Panel C: Signed Returns
Dependent Variable= Return
TR_Lower=0 TR_Lower=1
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.83*** 0.85*** 0.94*** 0.50*** 0.51*** 0.52***
[0.12] [0.10] [0.12] [0.18] [0.16] [0.16]
ATODisc*|TL_Diff| -0.40 -0.47 -0.61 -0.04 -0.14 -0.15
[1.11] [1.30] [1.38] [0.30] [0.31] [0.28]
ATODisc*|TL_Diff|*VolD_Pre/Conc 1.78 1.55***
[6.13] [0.47]
ATODisc*|TL_Diff|*VolD_Pre 2.51 3.57***
[3.34] [0.61]
ATODisc*|TL_Diff|*VolD_Conc 3.47 -4.28
[10.42] [2.67]
ATODisc*|TL_Diff|*VolD_Post 3.20 6.22***
[2.09] [0.81]
ATODisc*VD_Pre/Conc -0.90 -0.32
[3.76] [0.42]
ATODisc*VD_Pre -0.69 -1.23
[1.25] [0.76]
ATODisc*VD_Conc -2.63 3.05
[7.92] [2.32]
ATODisc*VD_Post -2.24** -5.34***
[1.03] [0.69]
Constant -0.19* -0.19** -0.20** -0.02 -0.03 -0.03
[0.10] [0.10] [0.10] [0.10] [0.09] [0.09]
Fixed Effects |TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff| |TL_Diff|, VolD,
|TL_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
N 6,695 6,695 6,695 15,047 15,047 15,047
R
2
0.23% 0.23% 0.29% 0.15% 0.17% 0.24%

101

Table 8
Market Reactions to the Second Year of Tax Return Disclosures, 2016
This table provides the results of estimating equation (2) in the second year of public tax return disclosure. The dependent variable is the absolute value of logarithmic
daily stock returns in Panel A, daily trading volume scaled by total shares outstanding in Panel B, logarithmic daily stock returns in Panel C. The independent variables
are indicator variables that represent the day of and the day after the ATO's tax return disclosure (ATODisc), if a firm made a voluntary disclosure related to the first year
of tax return disclosures (VolD_PY), if a firm made a voluntary disclosure related to the second year of tax return information before (VolD_Pre), concurrent with
(VolD_Conc), or after the ATO's disclosure (VolD_Post), and either before or concurrent with the ATO’s disclosure (VolD_Pre/Conc). The two continuous independent
variables, the absolute value of difference between a firm's actual tax liability and its current tax expense adjusted for the prior year ratio between the same variables
(|TL_Diff2|) and the absolute value of the difference between a firm's taxable income and its grossed up current tax expense adjusted for the prior year ratio between the
same variables (|TI_Diff2|), are decile ranked and scaled between 0 and 1. All variables are defined in Appendix A. All regressions are estimated from November 1,
2016-January 31, 2017 surrounding tax return disclosures in December, 2016. The table reports coefficients for the main variables of interest and omits indicator
variables that represent constant differences. Standard errors are clustered by day and reported in brackets below each coefficient. ***, **, and * represent statistical
significance at 1%, 5% and 10%, respectively.
Panel A: Unsigned Returns
Dependent Variable= |Return|
|TR_Diff2|= |TI_Diff2| |TR_Diff2|= |TL_Diff2|
Variable (1) (2) (3) (4) (5) (6)
ATODisc -0.31** -0.29* -0.28* -0.30** -0.29* -0.28*
[0.14] [0.17] [0.16] [0.15] [0.16] [0.15]
ATODisc*|TR_Diff2| 0.52*** 0.49** 0.48** 0.52*** 0.50** 0.49***
[0.14] [0.20] [0.18] [0.17] [0.19] [0.17]
ATODisc*|TR_Diff2|*VolD_Pre/Conc -0.59** -0.32*
[0.28] [0.19]
ATODisc*|TR_Diff2|*VolD_Pre -1.07*** -1.03**
[0.24] [0.51]
ATODisc*|TR_Diff2|*VolD_Conc -0.55 -0.13
[0.41] [0.50]
ATODisc*|TR_Diff2|*VolD_Post 0.08 0.05
[0.24] [0.21]
ATODisc*|TR_Diff2|*VolD_PY 0.69 0.85 0.54* 0.77***
[0.79] [0.57] [0.30] [0.19]
ATODisc*VolD_Pre/Conc 0.32** 0.04
[0.14] [0.34]
ATODisc*VolD_Pre 0.41 0.26
[0.37] [0.66]
ATODisc*VolD_Conc 0.98 0.56
[0.84] [0.90]
ATODisc*VolD_Post 0.20 0.26
[0.70] [0.58]

102

Table 8 Continued
ATODisc*VolD_PY -0.44* -0.65** -0.21* -0.45
[0.24] [0.26] [0.12] [0.50]
Constant 1.59*** 1.61*** 1.61*** 1.67*** 1.71*** 1.71***
[0.07] [0.07] [0.07] [0.07] [0.07] [0.07]
Fixed Effects |TR_Diff2| |TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2| |TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
N 22,089 22,089 22,089 22,089 22,089 22,089
R
2
0.25% 0.47% 0.69% 0.61% 0.86% 1.05%

103

Table 8 Continued
Panel B: Volume
Dependent Variable= Volume
|TR_Diff2|= |TI_Diff2| |TR_Diff2|= |TL_Diff2|
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.02** 0.02 0.02 0.02 0.02 0.02
[0.01] [0.01] [0.01] [0.01] [0.02] [0.02]
ATODisc*|TR_Diff2| 0.01 0.01 0.01 0.02 0.02 0.02
[0.01] [0.01] [0.01] [0.01] [0.01] [0.01]
ATODisc*|TR_Diff2|*VolD_Pre/Conc -0.06*** -0.18**
[0.02] [0.07]
ATODisc*|TR_Diff2|*VolD_Pre -0.15*** -0.33***
[0.02] [0.06]
ATODisc*|TR_Diff2|*VolD_Conc 0.13* 0.01
[0.07] [0.04]
ATODisc*|TR_Diff2|*VolD_Post 0.06 -0.06
[0.05] [0.05]
ATODisc*|TR_Diff2|*VolD_PY 0.01 -0.02 0.12 0.15
[0.08] [0.04] [0.13] [0.10]
ATODisc*VolD_Pre/Conc 0.03 0.08***
[0.04] [0.02]
ATODisc*VolD_Pre 0.04 0.14***
[0.07] [0.04]
ATODisc*VolD_Conc -0.07* 0.00
[0.11] [0.10]
ATODisc*VolD_Post -0.08 0.02
[0.09] [0.08]
ATODisc*VolD_PY 0.01 0.05 -0.04 -0.05***
[0.02] [0.04] [0.03] [0.02]
Constant 0.22*** 0.21*** 0.21*** 0.23*** 0.22*** 0.22***
[0.01] [0.01] [0.01] [0.01] [0.01] [0.01]
Fixed Effects |TR_Diff2|
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD |TR_Diff2|
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
N 22,089 22,089 22,089 22,089 22,089 22,089
R
2
0.21% 0.64% 0.99% 0.10% 0.54% 0.80%

104

Table 8 Continued
Panel C: Signed Returns
Dependent Variable = Return
|TR_Diff2|= |TI_Diff2| |TR_Diff2|= |TL_Diff2|
Variable (1) (2) (3) (4) (5) (6)
ATODisc 0.46*** 0.48*** 0.50*** 0.37*** 0.39** 0.41**
[0.16] [0.14] [0.14] [0.18] [0.16] [0.16]
ATODisc*|TR_Diff2| -0.56*** -0.63*** -0.70*** -0.40*** -0.48*** -0.52***
[0.07] [0.10] [0.10] [0.08] [0.09] [0.08]
ATODisc*|TR_Diff2|*VolD_Pre/Conc -0.42 -0.72
[0.80] [1.35]
ATODisc*|TR_Diff2|*VolD_Pre 0.70 -0.76
[0.95] [1.44]
ATODisc*|TR_Diff2|*VolD_Conc -0.25 -1.28***
[0.90] [0.97]
ATODisc*|TR_Diff2|*VolD_Post 2.67*** 1.51
[0.34] [0.93]
ATODisc*|TR_Diff2|*VolD_PY 0.98*** -0.03 1.51* 1.52**
[0.24] [0.39] [0.81] [0.72]
ATODisc*VolD_Pre/Conc 0.32 0.40
[0.51] [0.77]
ATODisc*VolD_Pre -0.49 0.32
[0.63] [0.79]
ATODisc*VolD_Conc 0.94 1.55**
[0.69] [0.64]
ATODisc*VolD_Post -1.14*** -0.09
[0.32] [0.36]
ATODisc*VolD_PY -0.39** 0.06 -0.55 -0.78**
[0.17] [0.33] [0.39] [0.37]
Constant 0.08 0.07 0.07 0.08 0.08 0.08
[0.07] [0.07] [0.07] [0.07] [0.07] [0.07]
Fixed Effects |TR_Diff2| |TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2| |TR_Diff2|, VolD,
|TR_Diff2|*VolD
|TR_Diff2|, VolD,
|TR_Diff2|*VolD
N 22,089 22,089 22,089 22,089 22,089 22,089
R
2
0.04% 0.06% 0.08% 0.04% 0.06% 0.08%

105

Table 9
TTC Summary Statistics
Panel A displays descriptive statistics for TTC variables used in my supplemental analyses. Panel B presents
differences in the means of these variables for firms that issue tax return specific voluntary disclosures and
those that do not. All variables are as defined in Appendix A. (0, 1) indicates an indicator variable. ***
represents statistically significant differences in the means (two tailed) at the 1% level.
Panel A: Distribution of Variables Across the Full Sample
Variable Obs Mean Std Dev P1 P25 P50 P75 P99
CountTTC 336 1.527 1.304 1.000 1.000 1.000 1.000 6.000
IncTTC (0,1) 336 0.113 0.317
TTC (0,1) 336 0.113 0.317

Panel B: Difference in the Means of Firms that Do and Do Not Issue Tax Return Specific Voluntary
Disclosures
Mean
Variable Disclose
Do Not
Disclose Diff
CountTTC 3.242 1.340 1.903***
IncTTC (0,1) 0.424 0.079 0.345***
TTC (0,1) 0.515 0.069 0.446***

106

Table 10
Tax Transparency Code Disclosure
This table reports the results of probit regressions that predict the probability a firm voluntarily issues new
disclosures in line with the Tax Transparency Code. The dependent variable equals 1 if a firm made more TTC
disclosures in the year following the first issuance of the code than in the prior year and zero otherwise. The
main variables of interest are the absolute difference between a firm's tax return values and its financial
statement tax values (|TR_Diff|) and a firm's unsigned cumulative return over the day of and day after the
ATO's tax return disclosure (|ATO_Return|). In Columns (1) and (2), |TR_Diff| measures the absolute difference
between a firm's actual taxable income and that implied by its financial statements (grossed up current tax
expense). While in columns (3) and (4), |TR_Diff| measures the absolute difference between a firm's actual tax
liability and its financial statement current tax expense. Columns (1) and (3) use |TR_Diff| and |ATO_Return|
from the first year of tax return disclosure while columns (2) and (4) calculate those same variables with the
second year of tax return disclosure. Robust standard errors are reported in brackets below each coefficient. All
variables are decile ranked, scaled from 0-1 and as defined in Appendix A. ***, **, and * denote statistical
significance at the .01, .05. and .1 levels respectively.
Dependent Variable= IncTTC
|TR_Diff| = |TI_Diff| |TR_Diff| = |TL_Diff|
Tax Return Year First Second First Second
Variable (1) (2) (3) (4)
|TR_Diff| -0.33 0.02 -0.03 -0.07
[0.42] [0.42] [0.40] [0.47]
|ATO_Return| -0.29 0.18 -0.32 0.19
[0.37] [0.41] [0.37] [0.41]
NoTaxPd 0.00 -0.06 0.01 -0.10
[0.28] [0.26] [0.28] [0.32]
TA_GAAP 0.54 0.22 0.49 0.22
[0.45] [0.40] [0.45] [0.40]
Ln(Asset) 2.44*** 2.08*** 2.43*** 2.08***
[0.50] [0.52] [0.50] [0.53]
ROA 0.57 0.09 0.49 0.10
[0.59] [0.46] [0.58] [0.46]
Leverage -0.31 -0.56 -0.30 -0.57
[0.45] [0.49] [0.45] [0.48]
PPE 0.49 0.61 0.50 0.62
[0.41] [0.38] [0.41] [0.38]
MTB 0.35 0.16 0.36 0.16
[0.55] [0.54] [0.54] [0.54]
Constant -3.43*** -3.08*** -3.52*** -3.03***
[0.66] [0.74] [0.68] [0.78]
Observations 302 303 302 303
Pseudo R
2
19% 16% 19% 16%

107

Table 11
Market Reactions to Tax Return Disclosures Conditional on Firm Issued TTC Disclosures
This table presents the results of estimating equation (2) with an extra interaction for tax transparency code disclosures. The dependent variable is the absolute value of
logarithmic daily stock returns in columns (1) and (2), daily trading volume scaled by total shares outstanding in columns (3) and (4) and logarithmic daily stock returns in
columns (5) and (6). The independent variables are indicator variables that represent the day of and the day after the ATO's tax return disclosure (ATODisc), if a firm made
a voluntary disclosure related to the first year of tax return disclosures (VolD_PY), if a firm made a voluntary disclosure related to the second year of tax return information
before (VolD_Pre), concurrent with (VolD_Conc), or after the ATO's disclosure (VolD_Post), and if the firm issued 3 or more of the 6 TTC disclosure recommendations
before the ATO's tax return disclosure (TTC). The two continuous independent variables, the absolute value of difference between a firm's actual tax liability and its
current tax expense adjusted for the prior year ratio between the same variables (|TL_Diff2|) and the absolute value of the difference between a firm's taxable income and its
grossed up current tax expense adjusted for the prior year ratio between the same variables (|TI_Diff2|), are decile ranked and scaled between 0 and 1. All variables are
defined in Appendix A. The table reports coefficients for the main variables of interest and omits indicator variables that represent constant differences. All regressions
relate to the second year of ATO tax return disclosure and are estimated from November 1, 2016-January 31, 2017 surrounding tax return disclosures in December 2016.
Standard errors are clustered by day and reported in brackets below each coefficient. ***, **, and * represent statistical significance at 1%, 5% and 10%, respectively.
Dependent Variable= |Return| Volume Return
|TR_Diff|= |TI_Diff2| |TL_Diff2| |TI_Diff2| |TL_Diff2| |TI_Diff2| |TL_Diff2|
Variable (1) (2) (3) (4) (5) (6)
ATODisc -0.30* -0.36** 0.02* 0.01 0.48*** 0.34**
[0.16] [0.15] [0.01] [0.01] [0.14] [0.17]
ATODisc*|TR_Diff2| 0.52*** 0.65*** 0.01 0.03*** -0.70*** -0.42***
[0.18] [0.17] [0.01] [0.01] [0.10] [0.09]
ATODisc*|TR_Diff2|*VolD_Pre -0.63** 0.22 -0.19*** -0.29*** 0.48 -0.12
[0.25] [0.51] [0.06] [0.02] [0.82] [1.31]
ATODisc*|TR_Diff2|*VolD_Conc -0.13 0.88* 0.09 0.04 -0.56 -0.91
[0.42] [0.50] [0.12] [0.08] [0.76] [0.83]
ATODisc*|TR_Diff2|*VolD_Post 0.12 0.02 0.07 -0.06 2.93*** 1.56*
[0.24] [0.21] [0.05] [0.04] [0.37] [0.90]
ATODisc*|TR_Diff2|*VolD_PY 0.86 0.63*** -0.02 0.14 0.01 1.39*
[0.57] [0.19] [0.04] [0.10] [0.38] [0.73]
ATODisc*|TR_Diff2|*TTC -0.73*** -1.90*** 0.03 -0.08 -0.29 -1.29***
[0.14] [0.11] [0.09] [0.07] [0.20] [0.17]
ATODisc*VolD_Pre 0.26 -0.22 0.04 0.11** -0.53 -0.03
[0.37] [0.66] [0.09] [0.06] [0.62] [0.79]
ATODisc*VolD_Conc 0.89 0.30 -0.06 -0.01 0.99 1.42**
[0.84] [0.90] [0.13] [0.11] [0.66] [0.61]
ATODisc*VolD_Post 0.14 0.21 -0.08 0.01 -1.36*** -0.19
[0.71] [0.58] [0.10] [0.07] [0.36] [0.34]

108

Table 11 Continued
ATODisc*VolD_PY -0.65** -0.37 0.04 -0.05*** 0.00 -0.74**
[0.26] [0.50] [0.04] [0.02] [0.31] [0.37]
ATODisc*TTC 0.35*** 0.92*** 0.00 0.06** 0.54*** 0.95***
[0.09] [0.07] [0.04] [0.03] [0.17] [0.12]
Constant 1.63*** 1.73*** 0.21*** 0.22*** 0.08 0.09
[0.08] [0.08] [0.01] [0.01] [0.07] [0.07]
Fixed Effects |TI_Diff|, VolD,
|TI_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
|TI_Diff|, VolD,
|TI_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
|TI_Diff|, VolD,
|TI_Diff|*VolD
|TL_Diff|, VolD,
|TL_Diff|*VolD
N 22,089 22,089 22,089 22,089 22,089 22,089
R
2
0.75% 1.15% 1.23% 1.00% 0.01% 0.09%

Abstract (if available)

Abstract In this paper, I study how mandated disclosures influence firms’ voluntary disclosure decisions and subsequently, how voluntary disclosures shape market reactions to mandatory disclosures. Specifically, I examine Australian firms’ voluntary disclosure responses to the involuntary disclosure of their tax returns. Tax return information is difficult to understand and most difficult when it largely differs from a firm’s previously reported financial statements. As the cost to understand a disclosure increases, theory predicts that the average precision of investors’ beliefs decreases, which may prompt managers to issue clarifying information. I find support for this theory as firms with large differences between their tax return and financial statement values are more likely to issue voluntary disclosures that supplement their tax return information. Further, my evidence suggests that voluntary disclosures shift investors’ attention away from the difference between a firm’s tax return values and its financial statement values.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Leveling the playing field: unbiased tests of the relative information content of book income and taxable income

PDF

Essays in corporate finance

Asset Metadata

Creator Kays, Allison Megan (author)

Core Title Voluntary disclosure responses to mandated disclosure: evidence from Australian corporate tax transparency

School Leventhal School of Accounting

Degree Doctor of Philosophy

Degree Program Business Administration

Publication Date 04/10/2019

Defense Date 03/20/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag information environment,OAI-PMH Harvest,tax disclosure,tax return information,voluntary disclosure

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Heitzman, Shane (committee chair), Allen, Eric (committee member), Beatty, Randolph (committee member), Ozbas, Oguzhan (committee member), Soliman, Mark (committee member)

Creator Email allison.kays.2019@marshall.usc.edu,allison49@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-8790

Unique identifier UC11669153

Identifier etd-KaysAlliso-6203.pdf (filename),usctheses-c89-8790 (legacy record id)

Legacy Identifier etd-KaysAlliso-6203.pdf

Dmrecord 8790

Document Type Dissertation

Rights Kays, Allison Megan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA