SSRN Id2802426

CEO Severance Pay and Corporate Tax Planning
John L. Campbell
University of Georgia
johnc@uga.edu
Jenny Xinjiao Guan

Monash University
jenny.guan@monash.edu
Oliver Zhen Li
Shanghai Lixin University of Accounting and Finance
lizhen@lixin.edu.cn
Zhen Zheng*
Xiamen University
zhenzheng@xmu.edu.cn
June 2019
Running Head: CEO Severance Pay and Corporate Tax Planning
* Corresponding author. We appreciate helpful comments and suggestions from two anonymous reviewers, Kris Allee (discussant),
Steve Baginski, Ted Christensen, Matt DeAngelis, Dan Dhaliwal, Jeremy Douthit, Anne Ehinger, Paul Fischer, Fabio Gaertner
(discussant), Cristi Gleason, Michelle Hanlon, Ken Klassen, Robbie Moon, Santhosh Ramalingegowda, Susan Shu (discussant),
Anup Srivastava (discussant), Bridget Stomberg, Jake Thornock, Erin Towery, Dan Wangerin, James Warren, Connie Weaver,
Ben Whipple, Jaron Wilde (discussant), workshop participants at the University of Georgia and the National University of
Singapore, conference participants at the 2016 MIT Asia Conference in Accounting, 2017 Financial Accounting and Reporting
Section (FARS) Midyear Meeting, the 2017 Journal of the American Taxation Association (JATA) Conference, the 2017 Center
for the Economic Analysis of Risk (CEAR) accounting conference at Georgia State University, and the 2017 American Accounting
Association (AAA) Annual Meeting. Zhen Zheng acknowledges the grants from NSFC (No. 71702160, 71790602, 71772156) and
the Fundamental Research Funds for the Central Universities (No. 20720171077).
Electronic copy available at: https://ssrn.com/abstract=2802426

Abstract
We examine the association between CEO severance pay (i.e., payment a CEO would receive if
s/he is involuntarily terminated) and corporate tax planning activities. We find that CEO severance
pay is positively associated with corporate tax planning, consistent with CEO severance pay
providing contractual protection against managers’ career concerns and thereby inducing
otherwise risk-averse managers to engage in incremental levels of tax planning. This result holds
under an instrumental variable approach and a propensity score matching, and survives alternative
measures of CEO severance pay and corporate tax planning. Finally, we find that severance pay
provides stronger tax planning incentives in situations where managers are expected to face greater
career concerns – when they are less experienced, when they face stronger shareholder monitoring,
and when they manage firms with higher idiosyncratic volatility. Overall, our results suggest that
CEO severance pay represents a form of efficient contracting with otherwise risk-averse managers.
Keywords: corporate tax planning; CEO severance pay; risk-taking incentives

INTRODUCTION
Corporate tax planning activities have the potential to increase shareholders’ after-tax wealth,
but can also subject managers to risk.1 Tax planning increases a firm’s expected cash flows (from
the expected tax savings), but also increases the dispersion of these cash flows, widening the
probability distribution of future cash flows (Goh, Lee, Lim, and Shevlin 2016). Tax planning,
thus, is analogous to value-increasing investment projects favored by risk-neutral shareholders
(who can diversify risk), but not necessarily by risk-averse managers (who are under-diversified
with their wealth and human capital, and thus tend to avoid risk). To alleviate this tax-related
agency problem, shareholders can design compensation contracts that induce otherwise risk-averse
managers to take more risk (Crocker and Slemrod 2005).
Recent research finds that stock-based compensation contracts encourage CEOs to take more
risks and, thus, engage in more tax planning (Rego and Wilson 2012; Armstrong, Blouin,
Jagolinzer, and Larcker 2015). However, stock options only provide managers with upside gain
potential, and fail to protect managers from downside risk (Cadman, Campbell, and Klasa 2016).
Downside risk protection is important in the tax setting because tax positions have been shown to
draw scrutiny from outside parties (i.e., the IRS, the SEC, analysts, and the media), and thus have
the potential to attract unwanted attention on CEOs and their firms (Bozanic, Hoopes, Thornock,
and Williams 2017; Kubick, Lynch, Mayberry, and Omer 2016; Ehinger, Lee, Stomberg, and
Towery 2017; Chen, Powers, and Stomberg 2016). This attention can even potentially lead to CEO
job termination (Chyz and Gaertner 2018).
1
We define tax planning broadly as any action that reduces firms’ explicit tax liability. A high level of tax planning
decreases firms’ explicit tax liability to a greater extent, and thus represents particularly risky tax positions. We use
the term “a high level of tax planning” or “risky tax positions” throughout to describe firms’ aggressiveness in reducing
the explicit tax liability.

Prior research provides evidence that severance pay reduces managers’ career concerns and
encourages otherwise risk-averse managers to take higher levels of risk (Ju, Leland, and Senbet
2014; Chen, Cheng, Lo, and Wang 2015; Cadman et al. 2016). In this study, we examine whether
the increased risk-taking behavior induced by severance pay extends to the tax setting. As in these
other risk-taking settings, it is not necessary to assume that a CEO believes that s/he will be fired
solely for engaging in incremental tax planning. Rather, as long as tax planning increases the
attention paid to a CEO in a way that would increase her/his career concerns, and this expected
increase in career concerns prevents the CEO from taking these tax positions, severance pay will
mitigate this tendency and the CEO will take more tax risks than s/he would absent the severance
pay.
Severance pay represents a significant portion of CEOs’ compensation package, as over 75
percent of S&P 1500 firms have sizeable severance agreements with their CEOs (Cadman et al.
2016).2 Using a hand-collected sample of severance pay amounts, we find a positive association
between CEO severance pay and corporate tax planning, after controlling for risk-taking incentives
from stock options (i.e., delta and vega) as well as other known determinants of tax planning
activities (i.e., firm size, firm age, operating volatility, R&D, capital expenditures, leverage, and
foreign operations). We find this association when using firms’ GAAP and cash effective tax rates
(ETRs) to proxy for the level of tax planning as well as two additional measures that proxy for
firms’ tax planning: tax settlements with the IRS, and unrecognized tax benefits. Overall, our
findings imply that severance pay encourages otherwise risk-averse managers to engage in
incremental tax planning activities, and that it represents a form of efficient contracting between
shareholders and managers (Ju et al. 2014; Cadman et al. 2016).
2
Over the last decade, the ten largest CEO severance payouts cost shareholders over $2.4 billion (GMI 2013).

As with all empirical compensation analyses, our tests present associations for which we cannot
definitively ascribe causality. We mitigate the likelihood that our results are driven by endogeneity
through two additional analyses. First, we use propensity score matching to alleviate functional
form misspecification by matching high and low severance pay firms on non-severance observable
characteristics. Our result continues to hold. Second, we use an instrumental variable approach to
alleviate any correlated omitted variable bias. As instruments, we use a firm’s geographic
proximity to its local largest severance payer as well as the local median severance pay. Tests of
instrument strength and over-identification indicate that our instruments are not weak or over-
identified. Our result continues to hold.
To better understand the underlying mechanism that links severance pay to tax planning
activities, we perform a set of cross-sectional tests. If CEO severance pay increases corporate tax
planning because it reduces managers’ career concerns and protects otherwise risk-averse
managers against downside risk, such an effect should be stronger when CEOs face greater career
concerns. We find that CEO severance pay exhibits a stronger association with corporate tax
planning when CEOs are less experienced with their firms, face stronger shareholder monitoring,
and operate in firms with higher idiosyncratic volatility – situations where managers are expected
to face greater career concerns. Thus, managers with greater career concerns appear to be more
sensitive to the incentive effect of CEO severance pay.
Our study contributes to the literature in several ways. First, we further explain the variation in
corporate tax planning activity. The fact that some firms engage in more tax planning activities
than others is puzzling and not fully understood (Shevlin 2007; Hanlon and Heitzman 2010). The
majority of prior literature seeks to explain the determinants of tax planning by focusing on firm
characteristics (e.g., size, profitability, capital structure, foreign income, accounting metrics to

evaluate performance) and manager characteristics (e.g., education, ability, political affiliation,
personal tax aggression) (Gupta and Newberry 1997; Wilson 2009; Dyreng, Hanlon, and Maydew
2010; Chyz 2013; Christensen, Dhaliwal, Boivie, and Graffin 2015). More recent studies suggest
that a possible explanation for the substantial variation in corporate tax planning activities is the
heterogeneity in managerial incentives. We complement these studies, providing evidence that one
particular form of managerial incentives (i.e., CEO severance pay) can at least partially explain
the large variance in corporate tax planning activities.
Second, we add to the literature on compensation contracts and corporate tax planning. Prior
studies focus primarily on the risk incentives provided by stock options. Findings related to these
contracts largely suggest that they provide incentives for managers to engage in incremental tax
planning (Phillips 2003; Desai and Dharmapala 2006; Rego and Wilson 2012; Gaertner 2014;
Armstrong et al. 2015; Powers, Robinson, and Stomberg 2016). Importantly, these studies examine
compensation contracts that only provide incentives when firm performance is positive. In contrast,
we examine a compensation agreement, CEO severance pay, that completes the spectrum of
compensation incentives provided to risk-averse CEOs by mitigating their career concerns (i.e.,
providing protection for downside risk). Such evidence is important in light of recent research
showing that tax risk can result in an increase in the probability of an IRS audit, an SEC comment
letter, high analyst scrutiny and media attention (Kubick et al. 2016; Chen et al. 2016; Bozanic et
al. 2017; Ehinger et al. 2017). Furthermore, extreme tax planning can damage a firm’s reputation
and value (Graham, Hanlon, Shevlin, and Shroff 2014) and even result in CEO turnover (Chyz
and Gaertner 2018).
Third, our study extends the literature on CEO severance pay. Theoretical work suggests that
CEO severance pay provides a risk taking incentive due to its unique role in contractual protection

against managers’ downside risk. Motivated by this argument, prior research finds that severance
pay encourages managers to take risks that increase firm value (Almazan and Suarez 2003; Chen
et al. 2016; Cadman et al. 2016; Baginski, Campbell, Hinson, and Koo 2018). We extend this
empirical work to the tax setting, where tax scholars are gradually adopting the view that
investment in corporate tax planning can be risky. Our findings support the notion that severance
pay represents a form of efficient contracting that encourages otherwise risk averse managers to
take more risk. Similarly, our results are inconsistent with the idea that severance contracts
represent agency costs and are only obtained by powerful, entrenched CEOs.3
Finally, our study has practical implications. As a prevalent compensation practice, severance
pay has received a lot of attention from practitioners and regulators. As recounted in an article on
www.thestreet.com, “It’s one thing when a CEO gets paid millions of dollars for a job well done,
but these executives made off like bandits despite failing in their roles” (Fiegerman 2010). Our
findings, taken together with other severance research, suggest that arguments such as this can
‘miss the point’ because they focus strictly on the ex post termination payout amount. However,
our results suggest that ex ante severance pay, on average, provides incentives for otherwise risk-
averse managers to take value-increasing risks.
LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
Corporate Tax Planning within an Agency Framework
3
We are aware of the existence of a contemporaneous paper, Brown, Dong, and Ke (2016), that also examines the
association between severance agreements and tax avoidance. Brown et al.’s sample begins in 1992, long before the
SEC mandated disclosure of severance. Thus, their severance data rely heavily on firms’ voluntary disclosure of
severance. Furthermore, Brown et al. (2016) build on Desai and Dharmapala (2006)’s entrenchment view of corporate
tax planning, which has been challenged by recent studies (Gallemore and Labro 2015; Lennox, Lisowsky, and
Pittman 2013; Blaylock 2016). Our cross-sectional tests provide evidence that our results are stronger when managers
are less likely to be entrenched (e.g., lower tenure). This result is consistent with our managerial risk-taking
explanation and inconsistent with Brown et al. (2016)’s management entrenchment explanation.

Seminal work by Allingham and Sandmo (1972) lays the theoretical foundation for individual
tax planning bevahior. The optimal amount of individual tax planning increases in the income tax
rate, but decreases in the probability of being detected, the amount of the tax penalty due upon
detection, and the extent to which an individual is risk averse.
Slemrod (2004) argues that this simple framework does not necessarily apply to corporate tax
planning as modern corporations feature the separation of ownership and control. Shareholders,
holding well-diversified portfolios, are risk-neutral. They favor all activities that are expected to
add to shareholder value (Merton 1973; Jensen and Meckling 1976). Managers, however, invest
their human capital in specific firms. They are not able to diversify their portfolios in a manner
similar to shareholders. By definition, then, risk-averse managers do not take the level of risk
desired by shareholders (Chen and Chu 2005). Therefore, incentive plans need to be in place to
induce otherwise risk-averse managers to engage in corporate tax planning.
Empirical work on the association between managerial incentives and corporate tax planning
largely focuses on risk-based incentives such as bonus contracts, stock options and stock awards.
However, to date, the empirical findings are mixed. Phillips (2003) finds no evidence that bonus
contracts (based on after-tax earnings) encourage CEOs to engage in tax planning activities.
However, this incentive appears to work for business unit managers and tax department managers
(Phillips 2003; Robinson, Sikes, and Weaver 2010). Using an expanded sample, Gaertner (2014)
finds that linking bonuses to after-tax earnings does encourage CEOs to engage in more tax
planning. Furthermore, Powers et al. (2016) find that linking CEOs’ bonus to after-tax earnings
encourages them to report lower GAAP ETRs but similar Cash ETRs. In addition, using cash flow
metrics to evaluate CEOs’ performance for bonuses is associated with lower GAAP and cash ETRs
than using earnings metrics.

Related to stock options, Desai and Dharmapala (2006) use firm fixed effects regressions and
find a negative association between stock options and the level of corporate tax planning. However,
Rego and Wilson (2012) find that stock options increase corporate tax planning, as stock options
provide CEOs with convex payoffs by linking option values to stock return volatility. This feature
of stock options (i.e., vega) induces risk-averse managers to engage in risky tax planning to pursue
upside gain potential. Finally, Armstrong et al. (2015) find that, on average, stock options provide
only modest tax planning incentives. Instead, results from their quantile regressions show that
options matter the most when corporate tax planning is either extremely high or extremely low.
CEO severance pay plays a unique role in providing managers with downside risk protection.
Figures 1 through 3 illustrate how a severance contract functions differently than stock options and
bonus contracts. As shown in Figure 1, economic agents (including managers) have concave utility
functions. Winning $1,000 increases her/his utility by an amount less than losing $1,000 decreases
his/her utility. Thus, the manager is risk-averse, and this risk-aversion deters her/him from taking
the level of risk that risk-neutral shareholders prefer. To mitigate this agency problem and to
encourage the manager to be more risk-neutral, a firm can choose to grant her/him stock options.
Figure 2 presents the payoff function of a stock option.4 The option is “in-the-money” only when
the market price exceeds the strike price, otherwise the option has zero intrinsic value. Due to its
convex payoff function, the stock option helps bring the manager’s incentives more in line with
risk-neutral shareholders by effectively straightening out (i.e., removing the concavity of) the
upside of the manager’s utility function. On the other hand, severance pay plays a distinct role.
Figure 3 shows the payoff function of severance pay. It becomes “in-the-money” when the
4
Bonus contracts based on after-tax earnings offer similar incentives as options (i.e., only provide payment upon good
future outcomes). Thus, for brevity, we only discuss options in this example.

manager is close to being involuntarily terminated, and is worthless when the manager can safely
retain her/his job. Therefore, severance pay has a convex payoff function which straightens out
(i.e., removes the concavity of) the downside of the manager’s utility function. Putting these
together, a combination of stock options and severance pay can motivate the manager to be more
risk-neutral than either contract can do alone (Almazan and Suarez 2003; Ross 2004; Ju et al.
2014).
Prior studies find evidence consistent with the above theoretical analysis that CEO severance
pay reduces managers’ career concerns and induces them to take a higher level of risk than they
would otherwise take. In particular, Cadman et al. (2016) find that CEO severance pay increases
stock return volatility, the level and the change in firm leverage, and same-industry (rather than
diversifying) acquisitions. They also find that CEO severance pay is positively associated with
acquisition announcement returns and the value of cash holdings, consistent with CEO severance
pay motivating managers to undertake not only risky projects but also projects with a positive net
present value. Brown, Jha, and Pacharn (2015), focusing on the financial sector, find that CEO
severance pay is positively associated with market-based risk after controlling for the incentive
effect of equity-based compensation. This suggests that severance pay induces risk taking.
Consistent with this strand of literature, we expect that CEO severance pay encourages tax-
related risk taking. First, tax positions have been shown to draw scrutiny from outside parties, and
therefore have the potential to attract unwanted attention on CEOs and their firm. Specifically,
prior research finds that tax planning activities can generate attention from the IRS (Bozanic et al.
2017), the SEC (Kubick et al. 2016), analysts (Ehinger et al. 2017), and the media (Chen et al.
2016). In addition, public revelations of tax planning activities can induce outside parties to
question the existence of other possible unethical activities undertaken by CEOs (Hanlon and

Slemrod 2009). These negative public images can harm CEOs’ reputation and limit their
bargaining power in the labor market (i.e., their career concerns) (Hirshleifer 1993; Graham et al.
2014; Austin and Wilson 2017). In extreme cases, tax planning activities can result in CEO
turnover. Specifically, Chyz and Gaertner (2018) find that an extremely high level of corporate tax
planning is positively associated with the likelihood of a forced CEO turnover.5 These results
suggest that career concerns can discourage managers from undertaking particularly high levels of
corporate tax planning.
Based on the analyses above, we expect that severance contracts alleviate managers’ career
concerns which in turn encourage managers to engage in higher levels of tax planning. Anecdotal
evidence supports this argument. Xerox Corporation offered severance contracts to its executives.
Under such contractual protection, the CEO of Xerox, Richard Thoman, shifted corporate
operations to low-tax Ireland to lower its effective tax rate. However, tax benefits were not
achieved due to losses incurred in Ireland, and this in turn inflated the firm’s effective tax rate.
This income shifting for tax purposes was considered a “big mistake,” and in the end resulted in a
forced CEO turnover.6 Upon termination, Richard Thoman received severance payments in the
form of cash payment ($375,000) and immediate vesting of unvested pension ($800,000 per year),
and thus was compensated for losses due to his dismissal.
That said, it is not necessary to assume that a CEO must believe that s/he would be terminated
solely for taking incremental tax positions. Instead, all a CEO needs to believe is that incremental
tax positions have the potential to come under scrutiny from the IRS/SEC/analysts/media/board of
directors, will attract unwanted attention, put them on thin ice with the board of directors and
5
Chuck Collins. CEOs Rewarded for Tax Dodging Gymnastics. The Huffington Post. Oct 31, 2011.
6
See the following media coverage for more details. James Bandler and Mark Maremont Staff Reporters of The
Wall Street Journal. How Xerox's Plan to Reduce Taxes While Boosting Earnings Backfired. The Wall Street
Journal. April 17, 2001.

shareholders, and cause them to be closer to termination (i.e., increase their reputational cost and/or
career concerns). As previously argued, prior research establishes that severance pay can mitigate
managers’ career concerns and encourage them to take incremental levels of risk (Cadman et al.
2016; Baginski et al. 2018). If tax planning increases the attention paid to a CEO in a way that
would increase her/his career concerns, and this expected increase in career concerns prevents the
CEO from taking these tax positions, severance pay will mitigate this problem and the CEO will
take more tax risk than s/he would absent the severance pay. This leads to our main hypothesis:
H1: There is a positive association between CEO severance pay and corporate tax planning.
On the other hand, it is possible that the general risk-taking activities encouraged by severance
pay do not extend to corporate tax planning decisions. Several prior studies point to this possibility.
Specifically, in contrast to Chyz and Gaertner (2018), Gallemore, Maydew, and Thornock (2014)
find no evidence that CEOs are fired after public revelations of corporate tax sheltering.
Furthermore, Chen et al. (2016) find no adjustment of tax positions after negative media coverage
on corporate tax planning. Finally, most tax planning strategies survive IRS audits to yield positive
cash tax savings (Nessa, Schwab, Stomberg, and Towery 2016). Thus, it is possible that the
attention created by risky tax positions (from the IRS, the SEC, analysts, the media, the board of
directors) does not affect a CEO’s career concerns. If so, CEO severance pay will not be associated
with corporate tax planning.
DATA and SAMPLE
Hand-collected Data on CEO Severance Pay
On August 11, 2006, the SEC released new disclosure requirements regarding executive
compensation. As a result, firms are now required to quantify and disclose contracted severance
pay that CEOs would receive upon forced termination, either without cause or for good reasons
10

(as defined in the contracts). Prior to this change in regulation, disclosures of severance pay only
included whether such a contract existed and a narrative description of any material payments.
Further, these disclosures did not distinguish between whether the payments were already vested
(i.e., the CEO would receive them regardless of termination) or unvested (i.e., amounts that will
only be received by a CEO upon dismissal). Detailed information on CEO severance pay contracts
was only publicly disclosed once the CEO departed from the firm. 7 In contrast, firms now must
quantify and disclose all components of severance pay, including the unvested portion of the
payment (i.e., the amount that will be received by a CEO only upon dismissal). This requirement
enables researchers to examine the ex ante impact of CEO severance pay on corporate tax planning
in a comprehensive manner (Cadman et al. 2016).
We examine firms’ proxy statements (DEF-14, DEF-14A, or item 11 of Form 10-K) and hand-
collect the exact amount of each component of severance pay. A CEO will receive these payments
if s/he is dismissed “without cause” or resigns for “good reason” (see Cadman et al. (2016) for a
detailed description of these scenarios). We collect the total amount of severance payments, the
cash payment portion (e.g., salary and bonus continuation up to a certain number of years after
termination), the portion comprised of continuation of healthcare and other benefits, the unvested
stock options and stock awards that vest immediately upon dismissal, the incremental pension
benefits that vest upon dismissal, and other uncategorized portions.
We use hand-collected severance data rather than those provided by ExecuComp as Cadman et
al. (2016) identify several problems with the ExecuComp data. For example, when firms present
7
The restricted sample size problem and potential self-selection bias make ex post CEO severance pay a less attractive
setting compared with ex ante CEO severance pay. Furthermore, Cadman et al. (2016) show that when a CEO is
involuntarily terminated, the ex post payment is virtually identical to the ex ante contracted amount (ratio of ex post
payout to ex ante contracted amount of 1.04, correlation between the two of 0.99). This provides reasonable assurance
that the severance pay a CEO should expect to receive upon involuntary termination is the contracted amount provided
ex ante.
11

tables that do not report a “total” amount for severance payments, ExecuComp incorrectly reports
the first or last number in a column as the total. Also, ExecuComp includes vested payments to a
CEO, which do not reflect the incremental severance pay to the CEO if s/he is dismissed. Finally,
ExecuComp reports a zero value for severance pay in some cases when firms do in fact have
contracted payments disclosed in the proxy statements.8 Due to these issues with ExecuComp data,
our analyses rely on more accurate hand-collected data from firms’ proxy statements. Empirical
analysis presented in Appendix B presents the results of our tests if we use ExecuComp data rather
than our hand-collected sample. Our results do not hold on this sample, further highlighting the
need for our hand collected (i.e., less noisy) severance pay data.
Sample Selection
We start with firms covered in the ExecuComp for the years (i.e., 2006 and 2007) for which we
hand-collected severance pay data.9 We exclude financial firms, firms with foreign headquarters,
and firms with missing data to construct our regression variables. Our final sample consists of
1,413 firm-year observations. Among these observations, 661 fall in the fiscal year of 2006 and
752 in the fiscal year of 2007. We winsorize all continuous variables at the 1 and 99 percentiles to
remove the impact of outliers.
CEO SEVERANCE PAY AND CORPORATE TAX PLANNING
Empirical Model
We estimate the following empirical model to assess the association between CEO severance
pay and corporate tax planning:
TPi,t = β0 + β1SERPAYi,t + β2SIZEi,t + β3ROAi,t + β4LEVi,t + β5MBi,t + β6FIRMAGEi,t

+ β7MNEi,t + β8RDi,t + β9CAPXi,t + β10PPEi,t + β11INTANGi,t + β12EQINCi,t
8
These firms disclosed severance payments in narrative form instead of tabular form, resulting in ExecuComp
mistakenly coding a zero amount of severance pay.
9
As the new requirements apply to firms with fiscal year end after December 15, 2006, we also collect data for 2007
to ensure we have at least one full year for all firms in our sample.
12

+ β13NOLi,t + β14DNOLi,t + β15CEOAGEi,t +β16CEOTENUREi,t
+ β17MSHAREi,t + β18VEGAi,t + β19DELTAi,t + ∑YEAR + ∑INDUSTRY + εi,t, (1)
where i and t index firm and year, respectively. The dependent variable is corporate tax planning
(TP). We use effective tax rate as our main proxy for corporate tax planning. This is consistent
with Crocker and Slemrod’s (2005) theoretical argument that “[t]o align incentives, it may be
appropriate for the tax officer’s salary to depend (inversely) on the effective tax rate achieved.”
Thus, effective tax rate is an established proxy for our theoretical construct, especially under the
setting of agency conflicts. In addition, prior research shows that lower ETRs are associated greater
tax uncertainty (Dyreng, Hanlon, and Maydew 2018), and attract unwanted attention from the IRS
(Bozanic et al. 2017), the SEC (Kubick et al. 2016), analysts (Ehinger et al. 2017), and the media
(Chen et al. 2016).10 Such intense public attention is the main source of riskiness of corporate tax
planning that results in managers’ career concerns. Therefore, ETR is an appropriate proxy in our
setting. However, as noted in Hanlon and Heitzman (2010), effective tax rate can be calculated in
a number of ways, and the use of a single effective tax rate measure in isolation may not accurately
capture the extent of corporate tax planning. Specifically, total tax expense that constitutes the
calculation of GAAP effective tax rate includes certain accounting items such as valuation
allowances that are more broadly related to the firm’s profitability in general rather than specific
to tax planning (Dyreng, Hanlon, and Maydew 2008). Cash effective tax rate, defined as cash taxes
paid over pre-tax income, can reduce this impact. For these reasons, we use both GAAP ETR and
Cash ETR to capture the level of corporate tax planning. Both ETR measures are winsorized to be
within 0 and 1. To the extent that ETRs can be noisy in measuring corporate tax planning, we use
two additional tax measures (tax settlements and unrecognized tax benefits) in robustness tests.
10
For example, the New York Times revealed, “The official corporate rate is 35 percent, yet PG&E, the California-
based utility company, has paid zero net taxes since 2007.”
13

The variable of interest is the amount of a CEO’s severance pay. We follow Cadman et al. (2016)
and deflate CEO severance pay by CEO wealth defined as the market value of firm equity shares
held by the CEO. We also use alternative deflators in robustness tests to ensure that our measure
of CEO severance pay is not sensitive to alternative scalers. Our main focus is the coefficient on
SERPAY, which captures the effect of CEO severance pay on corporate tax planning. A negative
coefficient on SERPAY would support the notion that CEO severance pay encourages managers to
engage in corporate tax planning.
We control for other variables shown in the prior literature to be related to corporate tax
planning. We first control for a set of firm characteristics, including firm size (SIZE), firm
profitability (ROA), leverage ratio (LEV), firm maturity (MB and FIRMAGE), foreign operation
(MNE), firm investment (CAPX), capital and R&D intensity (PPE and RD), asset tangibility
(INTANG), equity income in earnings (EQINC), and loss carry forward (NOL and DNOL). Next,
we control for a set of variables that capture managerial characteristics and incentives, including
CEO age and tenure (CEOAGE and CEOTENURE), managerial ownership (MSHARE), and stock
option vega and delta (VEGA and DELTA). Variables are defined in Appendix A. Industry and
year fixed effects are also included.
Descriptive Statistics
Table 1 reports descriptive statistics of our severance data as well as of our regression variables.
Panel A of Table 1 reports CEO compensation information. The average contracted severance pay
of our sample is $6.0 million, while the counterpart reported in ExecuComp is $7.1 million. This
is likely due to the fact that, as previously mentioned, ExecuComp includes as severance pay
amounts that a CEO has already “earned.” We also report the mean values of various components
of severance pay. Cash-based payment and stock-based payment constitute the largest proportion
14

of total severance pay. Also, note that severance pay is significant when compared to alternative
forms of CEO compensation. In particular, contracted severance pay is, on average, 7.63 times
CEO salary, 4.20 times bonus, 3.53 times stock awards, and 4.51 times option awards. The sharp
contrast highlights the economic significance of severance pay in CEO compensation.
Panel B of Table 1 reports descriptive statistics for corporate tax planning, severance pay, firm
characteristics, and CEO characteristics for the full sample. Our two measures of effective tax rate,
GAAP and cash ETR, have mean values ranging from 26.06% to 29.68%, both of which fall below
the statutory tax rate of 35%. This suggests that, on average, firms engage in at least some degree
of tax planning. Table 2 reports Pearson and Spearman correlations for the sample. Total severance
pay (SERPAY) is negatively associated with our two primary measures of tax planning. These
univariate correlations support H1.
Baseline Results
Table 3 presents our baseline results. We focus on the coefficient estimate on SERPAY, which
captures the impact of CEO severance pay on corporate tax planning. The coefficient on SERPAY
is significantly negative for our two primary measures of tax planning. This result is also
economically significant. A one-standard-deviation increase in CEO severance pay relative to
CEO wealth is associated with a 0.90 percentage point reduction in GAAP ETR [4.4898 × (–
0.0020)] and a 0.90 percentage point reduction in Cash ETR [4.4898 × (–0.0020)]. These
correspond to a 3.03% reduction in GAAP ETR [0.0090/0.2968] and a 3.45% reduction in cash
ETR [0.0090/0.2606] for the average firm. These results are consistent with severance pay
encouraging managers to engage in tax planning. They are inconsistent with severance pay
15

reflecting management entrenchment, as entrenched managers would refrain from taking the same
amount of risk.11
The coefficients on control variables are consistent with the findings in prior research. Large
firms report higher ETRs, consistent with larger firms facing a higher political cost that serves as
a deterrent to tax planning (Zimmerman 1983). Profitable firms have higher effective tax rates,
consistent with the notion that such firms often fall into high tax brackets (Gupta and Newberry
1997). Highly leveraged firms have lower effective tax rates, consistent with such firms utilizing
the tax deduction on interest expense to avoid taxes (Mills, Erickson, and Maydew 1998).
Multinational firms avoid more taxes, consistent with multinational firms having more tax
planning opportunities (Rego 2003). Firms with large intangible assets have higher effective tax
rates, and firms with loss carry forwards experience lower effective tax rates, consistent with the
findings in Chen, Chen, Cheng, and Shevlin (2010). Older CEOs and shorter-tenured CEOs engage
less in tax planning, consistent with a manager effect as documented in Dyreng et al. (2008).
We find that the coefficient on vega is significantly negative in one specification. This finding
is consistent with vega increasing corporate tax planning (Rego and Wilson 2012). A one-standard
deviation increase in vega is associated with a 0.8 percentage point increase in GAAP ETR, and
this corresponds to a 2.7% increase in GAAP ETR for the average firm. In terms of economic
significance, severance pay is at least as important as vega in inducing corporate tax planning.
However, we also note that this result is not consistent across both specifications. One possible
reason for this lack of consistency is offered by Kahneman and Tversky (1979), whose prospect
11
According to SFAS No. 146, severance pay is expensed only when the termination arrangement is communicated
to managers. However, it is likely that the severance payment is made in a different year in which severance pay is
tax deductible. This way, severance pay can create a book-tax difference that causes a mechanical association between
severance pay and corporate tax planning. We believe that the mechanical association, if any, is most likely to be
evident around CEO turnover. We examine whether our results are sensitive to CEO turnover, and find that our results
continue to hold when we exclude 22 observations with CEO turnover from our sample. Thus, it is unlikely that our
results are driven by the potential book-tax difference created by severance payment upon CEO termination.
16

theory suggests that people care more about losses than gains. Thus, we might find a stronger
incentive effect for CEO severance pay that protects managers from losses rather than stock
options that only provide CEOs with upside gain potential. The result is also consistent with
Armstrong et al. (2015)’s argument that stock options only provide tax planning incentives for
extreme corporate planning activities. In untabulated quantile regressions, we find that VEGA is
significant at the 1% level at the tails of the distribution of corporate tax planning.
Severance Pay Components
One important feature of our hand-collected severance data is that we are able to examine the
components of severance pay. Through this feature, we are able to determine which component of
severance pay induces the incentive effect. We decompose the total amount of severance pay into
several components, including (1) the cash payment which involves salaries and bonuses, (2)
continuing healthcare and other benefits, (3) the immediate vesting of any unvested stock options
and awards, and (4) the vesting of previously unvested pension payments. For each component,
we scale the amount by the total amount of CEO severance pay in order to compare the relative
importance of each component. Untabulated results show that cash payment and stock
options/awards components positively affect corporate tax planning, whereas the healthcare and
pension benefits do not affect corporate tax planning.
Endogeneity
Propensity Score Matching
Omitted variables can bias our baseline estimation. One form of omitted variable bias arises
from functional form misspecification. More specifically, our baseline model might omit, for
example, a quadratic term of certain independent variables that simultaneously affect CEO
severance pay and corporate tax planning. This functional form misspecification can render our
17

focus variable SERPAY endogenous. To address this concern, we use a matched sample approach,
which imposes no assumptions on the functional form (Kothari, Leone, and Wasley 2005).
We follow Fang, Tian, and Tice (2014) to implement the propensity score matching. For each
year, we sort firms into terciles based on scaled CEO severance pay (SERPAY). Top tercile firms
are high severance pay firms while bottom tercile firms are low severance pay firms. Through
matching, we make high versus low severance pay firms more comparable on observables. We
construct an indicator variable to capture the likelihood of being a high severance pay firm. In the
first stage, we estimate a Probit model. We select matching variables based on Cadman et al. (2016)
who suggest that CEO severance pay should increase with CEO’s risk and the cost of dismissal.
Based on this theoretical argument, they predict that CEO severance pay should be higher when
(1) a CEO is hired from outside the firm, (2) a CEO owns less of the firm’s equity, (3) a CEO has
shorter tenure, (4) a CEO has signed a non-compete agreement, and (5) a CEO is far away from
retirement. Cadman et al. (2016) also include other CEO characteristics (CEO stock option vega
and delta) and firm characteristics (firm age, leverage, firm size, and the market-to-book assets
ratio) as determinants of CEO severance pay. We include all these variables as well as the
remaining control variables in our baseline regression in the first-stage Probit regression. We
impose a caliper distance of 0.03, which, according to Shipman, Swanquist, and Whited (2017), is
the most commonly used caliper distance in accounting research. We also impose a common
support requirement, restricting our attention to propensity scores falling in the common support
region. We perform the nearest neighbor matching without replacement, i.e., if a control firm is
matched to more than one treatment firm, we retain only the pair with the minimum distance in
propensity score. This matching without replacement results in 322 observations with 161 unique
pairs.
18

Table 4 presents the propensity score matching estimation results. Column (1) of Panel A
reports pre-matching Probit regression results. A CEO hired from outside the firm, holding less of
firm’s shares, and signing a non-compete agreement is more likely to receive higher severance
payment, consistent with Cadman et al. (2016). Some of the CEO and firm characteristics are also
predictors of a high severance payment. The pre-matching Probit regression has a pseudo-R2
equaling 0.34. Also, a χ2 test for model fitness has a p-value below 0.001. These results suggest
that these variables do well in explaining the probability of being a high severance pay firm.
Column (2) of Panel A reports post-matching Probit regression results. The coefficients on
independent variables become largely insignificant. The pseudo-R2 drops to 0.02 and χ2 test for
model fitness shows a p-value near 1. These results indicate a successful matching. Panel B
presents covariate balances. Before matching, we observe significantly large differences in CEO
and firm characteristics between high and low severance paying firms. After matching, high and
low severance paying firms become largely comparable, with all the differences in matching
variables being insignificant. Panel C estimates our baseline regression using the matched sample.
We find that the coefficient on SERPAY remains negative and significant for our effective tax rate
measures. Consequently, our baseline results are likely not influenced by potential functional form
misspecification.
Instrumental Variable Approach
We next use an instrumental variable approach. We use two instruments for CEO severance
pay. The selection of the instruments is based on the community effect of decision making. Prior
studies show that decision-makers located in the same community make similar decisions through
social interactions or information diffusion (Brown, Ivković, Smith, and Weisbenner 2008; Hong,
Kubik, and Stein 2005; Pool, Stoffman, and Yonker 2015). Therefore, we expect that a focal firm’s
19

CEO severance pay is affected by its community CEO severance pay. To capture this community
effect, we first calculate the geographic distance between a focal firm and the local largest
severance payer. We define communities at the city-level (Pool et al. 2015) and use firm
headquarter ZIP codes to pinpoint latitudes and longitudes. The Vincenty formula is applied to
calculate the geographic distance. Closer geographic proximity between a focal firm and its local
largest severance payer leads to a higher likelihood that this focal firm also exhibits a larger amount
of CEO severance pay. The second instrument we use is the community median CEO severance
pay. We expect that a focal firm’s severance pay increases with the community median severance
pay.
Table 5 presents the estimation results. Column (1) reports the first-stage regression results.
The dependent variable is CEO severance pay (SERPAY). The independent variables are the two
instruments (MSERPAY and DISTANCE) and all the controls from our baseline model. We find
that the coefficient on MSERPAY is significantly positive and that the coefficient on DISTANCE
is significantly negative. Therefore, CEO severance pay increases with its community median
severance pay and decreases with the geographic distance from the local largest severance payer,
supporting our predictions.
We next test for weak instruments. Valid instruments need to strongly predict severance pay.
We perform an F-test by excluding the two instruments from the first-stage regression. This yields
an F-statistic equaling 109.79, which is much higher than 10, the critical value for a weak
instrument F-statistic (Staiger and Stock 1997). We also perform a Stock and Yogo (2005) test to
verify the instrument strength. We derive a Cragg-Donald F-statistic equaling 1063.18 which
20

exceeds the critical value of 19.93 for weak instruments. 12 These results suggest that our
instruments are not weak. Thus, the instrumental variable estimates are unlikely to be biased
toward OLS estimates.
We also check whether our instruments satisfy the exclusion restriction condition. Valid
instruments should be exogenous. This requires that our instruments are not correlated with the
error term. As we use more than one instrument, we can check this through an over-identification
test. We derive a Hansen J-statistic equalling 0.132 which corresponds to a p-value of 0.7164.
There is no evidence that our instruments violate the exclusion restriction condition. Thus, our
instruments appear valid. In this sense, the instrumental variable estimates can capture a causal
effect of CEO severance pay on corporate tax planning.
Columns (2) - (3) of Table 5 report the second-stage instrumental variable estimation results.
The dependent variable is corporate tax planning, measured using GAAP and cash effective tax
rates. The main independent variable is the instrumented CEO severance pay, measured as the
predicted value from the first-stage estimation. We find that the coefficient on the instrumented
SERPAY remains significantly negative across both specifications, suggesting that the impact of
CEO severance pay on corporate tax planning is likely not due to correlated omitted variables.
Alternative Measures of Key Variables
We next consider alternative measures of CEO severance pay. In our initial design, we deflate
the total amount of CEO severance pay by a measure of CEO wealth, i.e., the market value of firm
equity shares held by the CEO (as in Cadman et al. 2016), to assess how significant the payment
amount is to the CEO’s overall net worth. Alternative scalars have been used in the prior literature
12
Stock and Yogo (2005) suggest that, for one endogenous regressor (n = 1) and two instruments (K = 2), the critical
value for weak instrument based on the maximum size bias at the 5% significance level is 19.93. Refer to Table 5.2
in Stock and Yogo (2005) for more details.
21

that, instead of focusing on CEO wealth, focus on CEO annual compensation. Consequently, we
assess whether our baseline results are sensitive to using CEO annual cash compensation (salary
and bonus) and annual total compensation as scalers of CEO severance pay. Columns (1) – (4) of
Table 6, Panel A report the results. We find that the coefficient on severance pay remains (at least
marginally) significantly negative for three of the four columns. Therefore, our baseline results
appear relatively robust to alternative scalers.
Our severance pay variable is a continuous variable that covers both severance paying firms
and non-severance paying firms, and this raises the question of whether our results are driven by
a self-selection bias. When we partition only on severance paying firms, we find similar results
(Columns (5) - (6) of Table 6, Panel A). Consistent with Cadman et al. (2016), these results suggest
that it is the amount of severance pay that matters, not simply the existence. These patterns also
suggest that our results are not driven by selection differences between firms that provide
severance pay and firms that do not provide severance pay. Finally, we create a severance pay
indicator variable that equals 1 for severance paying firms to test whether the existence of
severance pay (rather than the amount) has an impact on tax planning activities. We find at least
marginally significant results across both ETR measures (Columns (7)-(8) of Table 6, Panel A),
suggesting that the existence of severance also plays a role in incremental tax planning activities.
In addition, we consider alternative measures to capture the level of corporate tax planning.
First, we use tax settlements with the Internal Revenue Service. Specifically, TAXSETTLE is
defined as the natural logarithm of tax settlements attributable to unrecognized tax benefits plus
one, with missing values set to zero. Column (1) of Table 6, Panel B shows that CEO severance
pay is positively associated with tax settlements, supporting that severance contracts encourage
managers to engage in incremental tax planning. We also use unrecognized tax benefits to measure
22

corporate tax planning. UTB is defined as the natural logarithm of year-end unrecognized tax
benefits plus one, with missing values set to zero. CEO severance pay is positively associated with
the amount of unrecognized tax benefits (Column (2) of Table 6, Panel B).13 This finding is also
consistent with severance pay being associated with incremental tax planning. Untabulated results
suggest that these alternative tax measures survive both the propensity score matching approach
and the instrumental variable estimation.
TESTING FOR THE MECHANISM
In this section, we test for the mechanism through which we expect CEO severance pay affects
corporate tax planning. If our theoretical argument holds, we would find that CEO severance pay
provides stronger tax planning incentives when managers face greater career concerns.
Accordingly, we conduct cross-sectional tests related to managerial career concerns. Throughout
these cross-sectional tests, we use the severance pay variable derived from the instrumental
variable estimation to quantify CEO severance pay.14
CEO Tenure
CEO tenure can capture managerial career concerns. There is evidence that long-tenured
managers face a low risk of forced turnover (Goyal and Park 2002). One possible reason is that
long-tenured managers have more knowledge about a firm’s environment and are more
experienced in dealing with uncertainty (Simsek 2007). Another reason is that, over time,
managers have accumulated more power and thus are less likely to be fired (Goyal and Park 2002).
Consequently, we use CEO tenure to infer career concerns. We expect that our results are more
13
A one-standard-deviation increase in severance pay is associated with a 5 percentage point increase in log-
transformed tax settlements (0.0119×4.4898). This corresponds to a 10% increase in log-transformed tax settlements
for the average firm (0.0119×4.4898/0.5001). Also, a one-standard-deviation increase in severance pay is associated
with a 9 percentage point increase in log-transformed UTB (0.0211×4.4898). This corresponds to a 5% increase in
log-transformed UTB for the average firm (0.0211×4.4898/1.8808).
14
We also perform OLS and propensity score matching estimations of the cross-sectional tests. Results remain
unchanged.
23

pronounced for short-tenured CEOs who likely face greater career concerns.
We use an indicator variable to capture short-tenured versus long-tenured CEOs. Short-tenured
CEOs are defined as those who are early in their tenure (i.e., bottom quartile of tenure). Long-
tenured CEOs are those who have a long tenure (i.e., top quartile of tenure). We create an indicator
variable LOWTENURE that equals 1 for short-tenured CEOs and 0 for long-tenured CEOs. We
use the interaction term between instrumented CEO severance pay and CEO tenure indicator
(SERPAY*LOWTENURE) to examine how CEO tenure affects the association between CEO
severance pay and corporate tax planning. Panel A of Table 7 reports the results. For brevity, we
only report variables of interest. Consistent with expectations, we find that the coefficient on the
interaction SERPAY*LOWTENURE is significantly negative across both ETR specifications. This
suggests that CEO severance pay motivates short-tenured CEOs to take risky tax positions.
Shareholder Rights
Next, we use shareholder rights to capture managerial career concerns. Strong shareholder
rights often impose intense shareholder monitoring on managers and are thus effective in replacing
managers (Yermack 2006). We use the governance index (G-index) developed by Gompers, Ishii,
and Metrick (2003) to measure shareholder rights, as in Chava, Livdan, and Purnanandam (2008).
The G-index is constructed based on 24 governance provisions. A low G-index indicates strong
shareholder rights and therefore greater managerial career concerns.
We use an indicator variable to capture strong versus weak shareholder rights. Firms with strong
shareholder rights are defined as those with low G-index (i.e., bottom quartile of G-index). Firms
with weak shareholder rights are those with high G-index (i.e., top quartile of G-index). We create
an indicator variable HIGHRIGHTS that equals 1 for firms with strong shareholder rights and 0
for firms with weak shareholder rights. We use the interaction term between instrumented CEO
24

severance pay and shareholder rights (SERPAY*HIGHRIGHTS) to determine how shareholder
rights affect the association between CEO severance pay and corporate tax planning. Panel B of
Table 7 reports the results. Consistent with expectations, we find that the coefficient on the
interaction SERPAY*HIGHRIGHTS is significantly negative for cash ETR. This suggests that
CEO severance pay motivates managers under intense shareholder monitoring to take risky tax
positions.
Idiosyncratic Volatility
Finally, we use firm idiosyncratic volatility as a third proxy for managerial career concerns.
Managers are expected to have greater career concerns when firms’ operations are more volatile
(Rau and Xu 2013). Consistent with this notion, Bushman, Dai, and Wang (2010) find that the
likelihood of CEO turnover increases in firm idiosyncratic volatility. As a result, managers face
greater career concerns when firms exhibit higher idiosyncratic volatility. We measure
idiosyncratic volatility as the standard deviation of the idiosyncratic regression residuals derived
from regressing a firm’s weekly stock returns on value-weighted industry and market weekly
returns (Crawford, Roulstone, and So 2012). A higher value of the standard deviation of the
idiosyncratic regression residuals indicates higher idiosyncratic volatility.
We use an indicator variable to capture high versus low idiosyncratic volatility. Firms with
high idiosyncratic volatility are those with more volatile idiosyncratic regression residuals (i.e.,
top quartile of idiosyncratic volatility). Firms with low idiosyncratic volatility are those with less
volatile idiosyncratic regression residuals (i.e., bottom quartile of idiosyncratic volatility). We use
the interaction term between instrumented CEO severance pay and idiosyncratic volatility
indicator (SERPAY*HIGHIVOL) to determine how idiosyncratic volatility affects the association
between CEO severance pay and corporate tax planning. Panel C of Table 7 reports the results.
25

Consistent with expectations, we find that the coefficient on the interaction SERPAY*HIGHIVOL
is significantly negative in two specifications. This suggests that CEO severance pay encourages
more corporate tax planning when managers operate in firms with high idiosyncratic volatility.
Principal Component Analysis
We conduct a principal component analysis of the above three measures of career concerns. We
multiply CEO tenure and shareholder rights by negative one so that all three measures are
increasing in career concerns. The first principal component explains around 40% of the variation
in three measures of career concerns. The first and the second principle components together
explain more than 70% of the variation in three measures. Both the first and the second principle
components have eigenvalues greater than one. Panels D, E, and F of Table 7 re-estimate the
regression using the first principal component, the second principal component, and the average
of the first and second principal components, respectively. Consistent with expectations, we find
that the coefficients on the three interactions, SERPAY*FIRSTPRIN, SERPAY*SECONDPRIN,
and SERPAY*AVGPRIN, are all significantly negative across both ETR specifications.
To summarize, we find that CEO severance pay provides stronger tax planning incentives to
managers when managers are expected to have greater career concerns (i.e., when they are short-
tenured, when they face strong shareholder monitoring, and when they operate firms with high
idiosyncratic volatility). These results are consistent with career concerns being a mechanism
underlying the association between CEO severance pay and corporate tax planning.
CONCLUSION
We examine the association between CEO severance pay (i.e., payment a CEO would receive
if s/he is involuntarily terminated) and corporate tax planning activities. We find that CEO
severance pay increases corporate tax planning, consistent with CEO severance pay providing
26

contractual protection against managers’ career concerns and thereby inducing otherwise risk-
averse managers to engage in incremental tax planning. This result continues to hold under an
instrumental variable approach and a propensity score matching and survives alternative measures
of CEO severance pay and corporate tax planning. Finally, we find that severance pay provides
stronger tax planning incentives in situations where managers are expected to have greater career
concerns– i.e., when they are short-tenured, when they face strong shareholder monitoring, and
when they operate firms with high idiosyncratic volatility. Overall, our results suggest that CEO
severance pay represents a form of efficient contracting with otherwise risk-averse managers in
that it encourages incremental tax risk-taking.
27

References
Allingham, M. G., and Sandmo, A. 1972. Income tax evasion: A theoretical analysis. Journal of Public
Economics 1: 323–338.
Almazan, A., and Suarez, J. 2003. Entrenchment and severance pay in optimal governance structures. The
Journal of Finance 58: 519–547.
Armstrong, C. S., Blouin, J. L., Jagolinzer, A. D., and Larcker, D. F. 2015. Corporate governance,
incentives, and tax avoidance. Journal of Accounting and Economics 60: 1–17.
Austin, C. R., and Wilson, R. J. 2017. An examination of reputational costs and tax avoidance: Evidence
from firms with valuable consumer brands. The Journal of the American Taxation Association, 39: 67-
93.
Baginski, S. P., Campbell, J. L., Hinson, L. A., and Koo, D. 2018. Do career concerns affect the delay of
bad news disclosure? The Accounting Review 93 (2), 61–95.
Blaylock, B.S. 2016. Is tax avoidance associated with economically significant rent extraction among US
firms? Contemporary Accounting Research 33: 1013-1043.
Bozanic, Z., Hoopes, J. L., Thornock, J. R., and Williams, B. M. 2017. IRS attention. Journal of Accounting
Research 55, 79-114.
Brown, J. R., Ivković, Z., Smith, P. A., and Weisbenner, S. 2008. Neighbors matter: Causal community
effects and stock market participation. The Journal of Finance 63: 1509–1531.
Brown, K., Jha, R., and Pacharn, P. 2015. Ex ante CEO severance pay and risk-taking in the financial
services sector. Journal of Banking and Finance 59: 111–126.
Brown, K., W. Dong, and Y. Ke. 2016. Ex ante severance agreements and tax avoidance. Working paper,
Brock University.
Bushman, R., Dai, Z., and Wang, X. 2010. Risk and CEO turnover. Journal of Financial Economics 96,
381-398.
Cadman, B. D., Campbell, J. L., and Klasa, S. 2016. Are ex-ante CEO severance pay contracts consistent
with efficient contracting? Journal of Financial and Quantitative Analysis 51: 737–769.
Chava, S., Livdan, D., and Purnanandam, A. 2008. Do shareholder rights affect the cost of bank loans? The
Review of Financial Studies 22, 2973-3004.
Chen, K.-P., and Chu, C. Y. C. 2005. Internal control versus external manipulation: A model of corporate
income tax evasion. The RAND Journal of Economics 36: 151–164.
Chen, S., Chen, X., Cheng, Q., and Shevlin, T. 2010. Are family firms more tax aggressive than non-family
firms? Journal of Financial Economics 95: 41–61.
Chen, S., Powers, K., and Stomberg, B. 2016. Examining the role of the media in influencing corporate tax
avoidance and disclosure. SSRN working paper.
Chen, X., Cheng, Q., Lo, A. K., and Wang, X. 2015. CEO contractual protection and managerial short-
termism. The Accounting Review 90: 1871–1906.
Christensen, D. M., Dhaliwal, D. S., Boivie, S., and Graffin, S. D. 2015. Top management conservatism
and corporate risk strategies: Evidence from managers' personal political orientation and corporate tax
avoidance. Strategic Management Journal 36: 1918–1938.
Chyz, J. A. 2013. Personally tax aggressive executives and corporate tax sheltering. Journal of Accounting
and Economics 56: 311–328.
Chyz, J. A., and Gaertner, F. 2018. Can paying “too much” or “too little” tax contribute to forced CEO
turnover? The Accounting Review 93, 103-130.
Crawford, S. S., Roulstone, D. T., and So, E. C. 2012. Analyst initiations of coverage and stock
return synchronicity. The Accounting Review 87: 1527-1553.
Crocker, K. J., and Slemrod, J. 2005. Corporate tax evasion with agency costs. Journal of Public Economics
89: 1593–1610.
Desai, M. A., and Dharmapala, D. 2006. Corporate tax avoidance and high-powered incentives. Journal of
Financial Economics 79: 145–179.
Dyreng, S. D., Hanlon, M., and Maydew, E. L. 2008. Long-run corporate tax avoidance. The Accounting
28

Review 83: 61–82.
Dyreng, S. D., Hanlon, M., and Maydew, E. L. 2010. The effects of executives on corporate tax avoidance.
The Accounting Review 85: 1163–1189.
Dyreng, S. D., Hanlon, M., and Maydew, E. L. 2018. When does tax avoidance result in tax
uncertainty? The Accounting Review, forthcoming.
Ehinger, A., Lee, J. A., Stomberg, B., and Towery, E. 2017. Let's talk about tax: The determinants and
consequences of income tax mentions during conference calls. SSRN working paper.
Fang, V. W., Tian, X., and Tice, S. 2014. Does stock liquidity enhance or impede firm innovation? The
Journal of Finance 69: 2085–2125.
Fiegerman, S. 2010. “Outrageous CEO severance payouts.” Published on August 16, 2010. Accessed on
September 15, 2018 at: https://www.thestreet.com/slideshow/12816598/1/infamous-ceo-severance-
payouts.html.
Gaertner, F. 2014. CEO after-tax compensation incentives and corporate tax avoidance. Contemporary
Accounting Research 31: 1077–1102.
Gallemore, J., and E. Labro. 2015. The Importance of the Internal Information Environment for Tax
Avoidance. Journal of Accounting and Economics 60: 149–167.
Gallemore, J., Maydew, E. L., and Thornock, J. R. 2014. The reputational costs of tax
avoidance. Contemporary Accounting Research 31, 1103-1133.
GMI Rating (GMI). 2013. 2013 CEO pay survey.
Goh, B., Lee, J., Lim, C., and Shevlin, T. 2016. The effect of corporate tax avoidance and cost of equity.
The Accounting Review 91: 1647–1670.
Gompers, P., Ishii, J., and Metrick, A. 2003. Corporate governance and equity prices. Quarterly Journal of
Economics 118: 107–155.
Goyal, V. K., and Park, C. W. 2002. Board leadership structure and CEO turnover. Journal of Corporate
Finance 8: 49–66.
Graham, J.R., Hanlon, M., Shevlin, T., and Shroff, N. 2014. Incentives for tax planning and avoidance:
Evidence from the field. The Accounting Review 89: 991–1023.
Gupta, S., and Newberry, K. 1997. Determinants of the variability in corporate effective tax rates: Evidence
from longitudinal data. Journal of Accounting and Public Policy 16: 1–34.
Hanlon, M., and Heitzman, S. 2010. A review of tax research. Journal of Accounting and Economics 50:
127–178.
Hanlon, M., and Slemrod, J. 2009. What does tax aggressiveness signal? Evidence from stock price
reactions to news about tax shelter involvement. Journal of Public Economics, 93(1-2), 126-141.
Hirshleifer, D. 1993. Managerial reputation and corporate investment decisions. Financial Management,
145-160.
Hong, H., Kubik, J. D., and Stein, J. C. 2005. Thy neighbor's portfolio: Word of mouth effects in the
holdings and trades of money managers. The Journal of Finance 60: 2801–2824.
Internal Revenue Service (IRS). 2014. Internal Revenue Service Data Book, 2014. Internal Revenue Service:
Washington, DC.
Jensen, M. C., and Meckling, W. H. 1976. Theory of the firm: managerial behavior, agency costs and
ownership structure. Journal of Financial Economics 3: 305–360.
Ju, N., Leland, H., and Senbet, L. W. 2014. Options, option repricing in managerial compensation: Their
effects on corporate risk. Journal of Corporate Finance 29: 628–643.
Kahneman, D., and Tversky, A. 1979. Prospect theory: An analysis of decision under risk. Econometrica
47:263-91.
Kothari, S. P., Leone, A. J., and Wasley, C. E. 2005. Performance matched discretionary accrual measures.
Journal of Accounting and Economics 39: 163–197.
Kubick, T. R., Lynch, D. P., Mayberry, M. A., and Omer, T. C. 2016. The effects of regulatory scrutiny on
tax avoidance: An examination of SEC comment letters. The Accounting Review 91, 1751-1780.
Lennox, C., Lisowsky P., and Pittman, J. 2013. Tax aggressiveness and accounting fraud. Journal of
29

Merton, R. C. 1973. Theory of rational option pricing. The Bell Journal of Economics and Management
Science 4: 141–183.
Mills, L. F., Erickson, M. M., and Maydew, E. L. 1998. Investments in tax planning. Journal of the
American Taxation Association 20: 1–20.
Nessa, M., C. Schwab, B. Stomberg, and E. Towery. 2016. How do IRS resources affect the tax enforcement
process? Working paper, Indiana University.
Phillips, J. 2003. Corporate tax planning effectiveness: The role of compensation-based incentives. The
Accounting Review 78: 491–521.
Pool, V. K., Stoffman, N., and Yonker, S. E. 2015. The people in your neighborhood: Social interactions
and mutual fund portfolios. The Journal of Finance 70: 2679–2732.
Powers, K., Robinson, J. R., and Stomberg, B. 2016. How do CEO incentives affect corporate tax planning
and financial reporting of income taxes? Review of Accounting Studies 21: 672–710.
Rau, P. R., and Xu, J. 2013. How do ex ante severance pay contracts fit into optimal executive incentive
schemes? Journal of Accounting Research 51: 631-671.
Rego, S. O. 2003. Tax-avoidance activities of U.S. multinational corporations. Contemporary Accounting
Research 20: 805–833.
Rego, S. O., and Wilson, R. 2012. Equity risk incentives and corporate tax aggressiveness. Journal of
Robinson, J.R., Sikes, S.A. and Weaver, C.D. 2010. Performance measurement of corporate tax
departments. The Accounting Review 85:1035–1064.
Ross, S. A. 2004. Compensation, incentives, and the duality of risk aversion and riskiness. The Journal of
Finance 59: 207–225.
Shevlin, T. 2007. The future of tax research: from an Accounting professor’s perspective. Journal of the
American Taxation Association 29: 87–93.
Shipman, J. E., Swanquist, Q. T., and Whited, R. L. 2017. Propensity score matching in accounting
research. The Accounting Review 92, 213-244.
Simsek, Z. 2007. CEO tenure and organizational performance: An intervening model. Strategic
Management Journal 28: 653–662.
Slemrod, J. 2004. The economics of corporate tax selfishness. National Tax Journal 57: 877–899.
Staiger, D., and Stock, J. H. 1997. Instrumental variables regression with weak instruments. Econometrica
65: 557–586.
Stock, J. H., and Yogo, M. 2005. Testing for Weak Instruments in Linear IV Regression: in Donald Andrews
and James Stocks, eds: Identification and inference for econometric models, Cambridge University Press,
New York, NY.
Wilson, R. J. 2009. An examination of corporate tax shelter participants. The Accounting Review 84: 969–
999.
Yermack, D. 2006. Golden handshakes: Separation pay for retired and dismissed CEOs. Journal of
Accounting and Economics 41, 237-256.
Zimmerman, J. L. 1983. Taxes and firm size. Journal of Accounting and Economics 5: 119–149.
30

APPENDIX A
Variable Definition
Variable Definition Compustat
Panel A CEO severance pay and corporate tax planning
Dependent variables
GAAP GAAP effective tax rate, measured as total tax expense over pre-tax book income TXT/(PI-SPI)
less special items.
CASH Cash effective tax rate, measured as cash taxes paid over pre-tax book income less TXPD/(PI-SPI)
special items.
Main independent variable
SERPAY CEO severance pay, measured as the contracted total severance payment amount to Hand-collected
a firm's CEO as disclosed in the 10-K report over the market value of firm equity severance/
shares held by the CEO. (SHROWN_EXCL
_OPTS*PRCC_F)
Control variables at the firm level
SIZE Firm size, measured as the logarithm transformed total assets. AT
ROA Firm profitability, measured as earnings before extraordinary items over total assets. IB/AT
LEV Firm leverage, measured as current debt plus long-term debt over total assets. (DLC+DLTT) /AT
MB Market-to-book ratio, measured as the market value of equity over the book value of (CSHO×PRCC_F)/
equity. CEQ
FIRMAGE Firm age, measured as the difference between the current year and the year a firm
first appears on Compustat.
MNE Multinationality, measured as an indicator variable that equals 1 for firms with pre- PIFO
tax foreign income and 0 otherwise.
RD R&D intensity, measured as R&D expenditure over sales. XRD/SALE
CAPX Firm investment, measured as capital expenditure over net property, plant, and CAPX/PPENT
equipment.
PPE Capital intensity, measured as net property, plant, and equipment over total assets. PPENT/AT
INTANG Asset tangibility, measured as intangible assets over total assets. INTAN/AT
EQINC Equity income, measured as equity income in earnings over total assets. ESUB/AT
NOL Loss carry forward, measured as an indicator variable that equals 1 for firms with TLCF
positive loss carry forward and 0 otherwise.
DNOL Change in loss carry forward, measured as changes in loss carry forward over total TLCF/AT
assets.
Control variables at the manager level
CEOAGE CEO age, measured as the age of the CEO of a firm. ExecuComp
CEOTENURE CEO tenure, measured as the number of years since a CEO has been in office. ExecuComp
MSHARE Managerial stock ownership, measured as the CEO shareholding percentage. ExecuComp
VEGA Option Vega, measured as change in the value of the CEO's stock options and
holdings for a one percent increase in stock return volatility, scaled by 1,000.
DELTA Option Delta, measured as change in the value of the CEO's stock options and
holdings for a one percent increase in stock price, scaled by 1,000.
Panel B Sensitivity tests and further tests
Instruments for CEO severance pay
DISTANCE Geographic distance to community largest severance payer, measured as the
geographic distance in miles between a firm and its city’s largest severance payer.
MSERPAY Community median severance pay, measured as the city’s median CEO severance
pay.
Propensity score matching variables
OUTSIDE Outside CEO, an indicator variable equal to 1 if a CEO is hired outside of a firm and Hand-collected
0 otherwise.
NONCOMPETE Non-compete contract, an indicator variable equal to 1 if a severance package Hand-collected
contains a non-compete agreement, and 0 otherwise.
31

Alternative measures of corporate tax planning
TAXSETTLE Tax settlements with tax authority, measured as the natural logarithm of one plus Log (1 +
tax settlements. Missing values are set to zero. TXTUBSETTLE)
UTB Unrecognized tax benefits, measured as the natural logarithm of one plus year-end Log (1 +
unrecognized tax benefits. Missing values are set to zero. TXTUBEND)
Alternative measures of CEO severance pay
SERPAY1 CEO severance pay, measured as the contracted total severance payment amount to Hand-collected
a firm's CEO as disclosed in the 10-K report over CEO total cash compensation.
SERPAY2 CEO severance pay, measured as the contracted total severance payment amount to Hand-collected
a firm's CEO as disclosed in the 10-K report over CEO total compensation.
Components of CEO severance pay
SERCASH Cash-related severance pay, measured as the cash component of severance payment Hand-collected
to a firm's CEO as disclosed in the 10-K report over total assets.
SERBENEFIT Healthcare-related severance pay, measured as the healthcare and other benefits of Hand-collected
severance payment to a firm's CEO as disclosed in the 10-K report over total assets.
SEROPTION Equity-related severance pay, measured as the stock options and stock awards of Hand-collected
severance payment to a firm's CEO as disclosed in the 10-K report over total assets.
SERPENSION Pension-related severance pay, measured as the pension component of severance Hand-collected
payment to a firm's CEO as disclosed in the 10-K report over total assets.
Panel C Cross-sectional tests
Managerial career concern
LOWTENURE Managers who are short-tenured. CEO tenure is defined as the number of years ExecuComp
since a CEO has been in office. Short-tenured CEOs (long-tenured CEOs) have
tenure below (above) the bottom (top) quartile of CEO tenure.
HIGHRIGHTS Managers who face strong shareholder monitoring. Shareholder monitoring is IRRC
defined as shareholder rights captured by Gompers et al (2003)’s governance index.
The G-index summarizes 24 governance provisions with a lower score indicating
stronger shareholder rights.
HIGHIVOL Idiosyncratic volatility is estimated by the following regression model: CRSP
𝑅𝐸𝑇𝑖,𝑡 = 𝛼0 + 𝛼1 𝑀𝐴𝑅𝐾𝐸𝑇𝑡 + 𝛼2 𝐼𝑁𝐷𝑅𝐸𝑇𝑡 ,
where RET is firm i’s weekly stock returns. MARKET is weekly value-weighted
stock market returns excluding firm i. INDRET is weekly value-weighted two-digit
SIC industry returns excluding firm i. Idiosyncratic volatility is the standard
deviation of the idiosyncratic regression residuals.
FIRSTPRIN The first principal component of the above three measures of career concerns.
SECONDPRIN The second principal component of the above three measures of career concerns.
AVGPRIN Average of the first and second principal components of the three measures.
32

APPENDIX B
Regression Results using ExecuComp Severance Pay
(1) (2)
GAAP CASH
Intercept 0.2703*** 0.2767***
(6.47) (6.24)
SERPAY -0.0006 -0.0014
(-0.57) (-1.60)*
SIZE 0.0034 0.0007
(1.07) (0.21)
ROA 0.8057*** 0.5030***
(17.29) (12.23)
LEV -0.0328 -0.0856***
(-1.46) (-4.10)
MB -0.0055*** -0.0037***
(-4.47) (-3.74)
FIRMAGE -0.0004 -0.0004
(-1.27) (-1.32)
FI -0.0140* 0.0165*
(-1.66) (1.79)
PPE -0.2655*** -0.3398***
(-3.21) (-4.61)
RD 0.0220 -0.0522*
(0.80) (-1.90)
CAPX 0.0009 -0.0459**
(0.04) (-2.35)
INTANG 0.0767*** 0.0559***
(4.00) (2.76)
EQINC -0.0827 0.0046
(-0.12) (0.01)
NOL 0.0040 -0.0155**
(0.62) (-2.29)
DNOL -0.1341** 0.0289
(-1.97) (0.55)
CEOAGE -0.0001 0.0003
(-0.29) (0.50)
CEOTENURE 0.0000 0.0003
(0.00) (0.96)
MSHARE -0.0197 -0.0378**
(-1.31) (-2.14)
VEGA -0.0961*** -0.0315
(-4.24) (-1.09)
DELTA 0.0053 0.0048
(1.53) (1.29)
YEAR YES YES
INDUSTRY YES YES
R2 0.15 0.12
N 6,880 6,880
We re-estimate our baseline regression using ExecuComp severance pay data (TERM_PYMT). We use a sample period from 2006
to 2014. We use the same variable definitions and sample selection procedure and obtain 6,880 firm-year observations. The
coefficient on SERPAY is negative and only marginally significant when we use cash ETR as the dependent variable, whereas the
coefficient is not significant when using GAAP ETR. The magnitudes of these coefficients are also smaller than our baseline
estimates. These results using a larger sample from ExecuComp severance data suggest merit in hand collected data on severance
pay to ensure accuracy – as ExecuComp severance data has been shown to be noisy and inaccurate (Cadman et al. 2016).
33

Figure 1. Utility function of a risk-averse agent
Figure 2. Payoff function of stock options
Figure 3. Payoff function of severance pay
34

TABLE 1
Descriptive Statistics
Panel A: CEO compensation (in thousands of dollars)
N Mean Median
CEO severance pay (hand-collected)
CEO severance pay 1,413 5960.61 1634.16
CEO severance pay components
Cash salary and bonus continuation benefits 1,413 2591.70 1000.00
Health care or related continuation benefits 1,413 64.17 0.00
Stock options accelerated vesting benefits 1,413 2667.70 0.00
Pension continuation benefits 1,413 437.06 0.00
CEO severance pay (ExecuComp)
CEO severance pay 1,413 7134.91 1987.45
CEO compensation components
Salary 1,413 781.71 745.00
Bonus 1,413 1419.78 802.56
Stock awards 1,413 1688.42 684.38
Option awards 1,413 1320.53 600.07
Panel B: Descriptive statistics
N Mean STD Q1 Median Q3
GAAP 1,413 0.2968 0.1587 0.2343 0.3232 0.3694
CASH 1,413 0.2606 0.1922 0.1290 0.2614 0.3490
TAXSETTLE 1,413 0.5002 1.1076 0.0000 0.0000 0.3443
UTB 1,413 1.8808 2.0830 0.0000 1.3161 3.4438
SERPAY 1,413 1.4952 4.4898 0.0000 0.2283 1.1165
SIZE 1,413 7.9195 1.6064 6.7588 7.7712 8.8996
ROA 1,413 0.0985 0.1054 0.0379 0.0874 0.1542
LEV 1,413 0.2107 0.2217 0.0442 0.1751 0.3005
MB 1,413 3.0482 2.6526 1.7270 2.5517 3.7262
FIRMAGE 1,413 33.3914 17.0922 20.0000 28.0000 51.0000
MNE 1,413 0.6341 0.4818 0.0000 1.0000 1.0000
RD 1,413 0.0320 0.0706 0.0000 0.0000 0.0238
CAPX 1,413 0.2404 0.1447 0.1360 0.2066 0.3125
PPE 1,413 0.2900 0.2702 0.0806 0.2034 0.4347
INTANG 1,413 0.2073 0.2308 0.0196 0.1279 0.3253
EQINC 1,413 0.0012 0.0043 0.0000 0.0000 0.0000
NOL 1,413 0.3907 0.4881 0.0000 0.0000 1.0000
DNOL 1,413 0.0036 0.0520 0.0000 0.0000 0.0000
CEOAGE 1,413 55.7466 6.5784 51.0000 56.0000 60.0000
CEOTENURE 1,413 6.3489 6.5347 2.0000 5.0000 8.0000
MSHARE 1,413 0.1156 0.2815 0.0101 0.0276 0.0863
VEGA 1,413 0.1313 0.1722 0.0222 0.0651 0.1584
DELTA 1,413 0.6635 1.0932 0.1031 0.2875 0.7135
Panel A reports CEO compensation information. Panel B reports descriptive statistics of variables used in our main analysis.
Variables are defined in the Appendix A.
35

TABLE 2
Correlation Matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
(1) GAAP 0.23 -0.09 0.03 0.23 0.01 0.05 -0.03 -0.16 -0.19 0.03
(2) CASH 0.31 -0.06 0.09 0.10 -0.07 0.03 0.07 -0.01 -0.14 -0.11
(3) SERPAY -0.12 -0.09 -0.01 -0.10 0.03 -0.02 0.03 0.04 0.03 0.02
(4) SIZE -0.03 0.10 -0.02 -0.10 0.12 -0.10 0.39 -0.04 -0.20 -0.30
(5) ROA 0.28 0.19 -0.18 -0.16 -0.14 0.38 -0.07 0.14 -0.20 0.19
(6) LEV -0.01 -0.08 0.06 0.26 -0.17 -0.11 0.03 -0.06 -0.08 -0.20
(7) MB 0.08 0.07 -0.12 -0.09 0.55 -0.12 -0.14 0.08 0.10 0.24
(8) FIRMAGE -0.09 0.07 -0.01 0.39 -0.08 0.14 -0.14 0.02 -0.18 -0.31
(9) MNE -0.26 -0.03 0.08 -0.04 0.17 -0.06 0.13 0.02 0.20 0.10
(10) RD -0.31 -0.16 0.04 -0.21 0.03 -0.15 0.18 -0.06 0.42 0.20
(11) CAPX 0.10 -0.08 0.00 -0.29 0.29 -0.28 0.30 -0.31 0.13 0.13
(12) PPE 0.12 0.05 -0.09 0.01 0.21 0.33 0.05 0.20 -0.07 -0.18 -0.25
(13) INTANG -0.03 0.04 0.13 -0.01 0.10 0.20 0.07 -0.08 0.38 0.24 0.08
(14) EQINC -0.01 0.01 0.02 0.18 0.03 0.11 -0.02 0.19 0.11 -0.04 -0.13
(15) NOL -0.12 -0.15 0.08 -0.11 0.00 0.06 0.03 -0.07 0.26 0.23 0.12
(16) DNOL -0.08 0.02 0.05 0.01 -0.09 0.02 0.01 -0.01 0.04 -0.01 -0.03
(17) CEOAGE 0.04 0.06 -0.11 0.05 -0.04 0.07 -0.11 0.11 -0.03 -0.05 -0.11
(18) CEOTENURE 0.06 -0.05 -0.21 -0.08 0.05 0.07 0.06 -0.10 -0.03 0.01 0.04
(19) MSHARE 0.14 -0.01 -0.35 -0.34 0.01 -0.03 -0.02 -0.25 -0.09 -0.14 0.13
(20) VEGA -0.10 0.05 0.01 0.46 0.09 0.00 0.18 0.16 0.20 0.18 0.04
(21) DELTA 0.05 0.07 -0.27 0.43 0.32 0.00 0.32 0.06 0.16 0.00 0.12
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
(1) GAAP 0.10 0.04 -0.01 -0.08 -0.12 0.04 0.01 0.01 -0.05 0.02
(2) CASH -0.01 0.01 0.00 -0.11 -0.03 0.07 -0.06 -0.02 0.02 0.00
(3) SERPAY -0.04 0.07 -0.01 0.04 0.08 -0.10 -0.13 -0.12 0.00 -0.10
(4) SIZE 0.06 -0.02 0.14 -0.12 -0.04 0.06 -0.13 -0.21 0.49 0.27
(5) ROA 0.14 0.06 0.11 -0.05 -0.18 -0.02 0.01 0.09 0.07 0.19
(6) LEV 0.31 0.30 0.04 0.06 0.01 0.03 0.05 -0.06 -0.05 -0.03
(7) MB -0.01 0.02 0.02 -0.02 0.07 -0.10 0.03 0.07 0.12 0.20
(8) FIRMAGE 0.17 -0.13 0.16 -0.08 -0.04 0.11 -0.11 -0.16 0.18 0.01
(9) MNE -0.18 0.29 0.11 0.26 0.02 -0.02 -0.04 -0.02 0.16 0.12
(10) RD -0.24 0.12 -0.06 0.17 0.12 -0.07 0.04 -0.03 0.16 -0.01
(11) CAPX -0.23 0.06 -0.11 0.10 0.03 -0.15 0.02 0.09 -0.01 0.05
(12) PPE -0.23 0.11 -0.09 0.00 0.07 0.09 0.00 -0.09 -0.02
(13) INTANG -0.20 -0.03 0.17 0.03 -0.02 0.01 -0.08 0.09 0.04
(14) EQINC 0.17 -0.01 0.00 0.00 0.03 -0.06 0.03 0.08 0.07
(15) NOL -0.04 0.23 0.00 -0.03 -0.02 -0.03 0.04 0.02 0.00
(16) DNOL -0.01 0.03 0.01 -0.12 -0.03 0.00 -0.03 0.02 0.01
(17) CEOAGE 0.09 -0.01 0.04 -0.03 -0.06 0.37 0.10 0.02 0.10
(18) CEOTENURE 0.08 0.00 -0.04 -0.03 0.00 0.29 0.33 0.03 0.23
(19) MSHARE -0.06 -0.04 -0.06 0.00 -0.02 0.11 0.35 -0.11 0.39
(20) VEGA -0.08 0.16 0.07 0.04 0.03 -0.01 0.08 -0.23 0.39
(21) DELTA -0.03 0.11 0.09 -0.01 -0.03 0.10 0.30 0.17 0.59
This table reports Pearson (below diagonal) and Spearman (above diagonal) correlation. Variables are defined in the Appendix A. Coefficients in bold indicate significance below the 10% level.
36

TABLE 3
(1) (2)
GAAP CASH
Intercept 0.2410*** 0.1017
(4.27) (1.44)
SERPAY -0.0020** -0.0020**
(-2.13) (-1.92)
SIZE 0.0054 0.0114**
(1.33) (2.18)
ROA 0.3154*** 0.1607**
(5.21) (2.31)
LEV -0.0300 -0.0831***
(-0.76) (-2.75)
MB -0.0006 0.0026
(-0.37) (1.07)
FIRMAGE -0.0004 0.0001
(-1.16) (0.28)
MNE -0.0457*** 0.0144
(-4.22) (1.05)
RD -0.0776 -0.1494
(-0.65) (-1.17)
CAPX 0.0577* -0.1152***
(1.80) (-3.02)
PPE 0.0242 -0.0201
(1.13) (-0.75)
INTANG 0.1024*** 0.0699**
(4.28) (2.49)
EQINC -0.4007 -1.2859
(-0.44) (-1.21)
NOL -0.0008 -0.0251**
(-0.08) (-2.18)
DNOL -0.2114** -0.0395
(-2.37) (-0.33)
CEOAGE 0.0008 0.0021**
(1.08) (2.30)
CEOTENURE -0.0000 -0.0018*
(-0.04) (-1.89)
MSHARE -0.0091 0.0179
(-0.57) (0.97)
VEGA -0.0467* -0.0206
(-1.81) (-0.58)
DELTA -0.0009 -0.0088**
(-0.27) (-2.04)
YEAR YES YES
INDUSTRY YES YES
R2 0.12 0.08
N 1,413 1,413
This table reports our baseline regression results of Equation (1), which regresses corporate tax planning activity on the amount of CEO severance
pay. In Columns (1) and (2), the dependent variables are corporate tax planning measured as GAAP effective tax rate (GAAP) and cash effective
tax rate (CASH). The main independent variable is CEO severance pay (SERPAY), measured as the contracted total severance payment amount to
a firm's CEO divided by CEO wealth defined as the market value of firm equity shares held by the CEO. Other variables are defined in the Appendix
A. Regressions include year and industry fixed effects. Standard errors are clustered by firm. T-statistics are in the parentheses. ***, **, and *
represent significance at 1%, 5% and 10% levels using one-tailed p-values when a directional prediction is made (two-tailed otherwise).
37

TABLE 4
Propensity Score Matching
Panel A First-Stage Probit Regression
(1) (2)
Pre-matcing Post-matching
Intercept -6.9655 4.7474
(-1.05) (0.54)
OUTSIDE 1.2757*** -0.2134
(6.48) (-0.75)
MSHARE -14.5446*** -0.3437
(-7.34) (-0.14)
CEOTENURE -0.0187 0.0050
(-1.01) (0.21)
NONCOMPETE 1.6624*** -0.0425
(8.80) (-0.16)
RETIREAGE 0.1698 -0.0540
(1.63) (-0.39)
VEGA 1.1637* 0.8800
(1.74) (0.90)
DELTA -0.2704* -0.1535
(-1.89) (-0.79)
FAGE -0.0059 -0.0032
(-0.93) (-0.37)
LEVERAGE -0.0167 0.3703
(-0.03) (0.52)
SIZE -0.2953*** -0.0514
(-3.56) (-0.44)
MB 0.0065 -0.0374
(0.17) (-0.73)
ROA -5.7226*** 1.5798
(-5.04) (1.01)
MNE 0.4142* 0.0846
(1.88) (0.27)
RD -4.8341*** 1.2167
(-2.94) (0.55)
CAP 1.0582 -0.8958
(1.39) (-0.86)
PPE -0.8027* 0.5022
(-1.71) (0.75)
INTANG 0.6480 -0.3113
(1.40) (-0.45)
EQINC -1.5887 -9.1585
(-0.08) (-0.29)
NOL 0.2992 0.0213
(1.58) (0.08)
DNOL 0.8150 0.0650
(0.41) (0.02)
CEOAGE 0.1453 -0.0726
(1.45) (-0.55)
YEAR YES YES
INDUSTRY YES YES
R2 0.34 0.02
N 938 322
Panel B Covariate Balance
(1) (2)
Pre-matching Post-matching
Treat Control Difference Treat Control Difference
Mean Mean t-value Mean Mean t-value
OUTSIDE 0.4331 0.1803 8.77*** 0.2484 0.2671 -0.38
MSHARE 0.0272 0.1958 -9.03*** 0.0354 0.0400 -0.75
CEOTENURE 4.6752 7.7338 -7.28*** 5.5093 5.8012 -0.43
NONCOMPETE 0.5505 0.2008 11.85*** 0.2919 0.2981 -0.12
RETIREAGE 10.2612 8.8155 3.65*** 10.0497 9.3478 0.99
VEGA 0.1244 0.1309 -0.59 0.1361 0.1318 0.22
DELTA 0.3566 0.8833 -7.93*** 0.4400 0.5164 -0.82
FAGE 33.6263 34.8658 -1.13 35.2298 35.5839 -0.18
LEVERAGE 0.2245 0.2027 1.53 0.2110 0.2023 0.39
SIZE 7.8916 7.9167 -0.24 7.9875 8.0849 -0.54
MB 2.7465 3.2472 -2.82*** 3.0748 3.1534 -0.25
ROA 0.0763 0.1200 -6.31*** 0.1001 0.0958 0.38
MNE 0.6752 0.6080 2.16** 0.6211 0.6211 0.00
38

RD 0.0328 0.0329 -0.02 0.0375 0.0339 0.42
CAP 0.2431 0.2337 1.02 0.2303 0.2374 -0.44
PPE 0.2682 0.3173 -2.85*** 0.3147 0.2840 1.01
INTANG 0.2220 0.1792 2.90*** 0.1908 0.2044 0.56
EQINC 0.0011 0.0013 -0.67 0.0010 0.0011 -0.32
NOL 0.4268 0.3522 2.36** 0.3602 0.3478 0.23
DNOL 0.0066 0.0006 1.80* 0.0018 0.0008 0.21
CEOAGE 54.8662 56.5367 -3.91*** 55.0994 55.9255 1.10
Panel C Baseline results using matched sample
(1) (2)
GAAP CASH
Intercept 0.3264*** 0.3025**
(2.87) (2.09)
SERPAY -0.0028** -0.0025**
(-1.81) (-1.73)
SIZE -0.0093 -0.0054
(-1.10) (-0.54)
ROA 0.3657*** 0.1171
(2.66) (0.62)
LEV 0.0427 0.0626
(0.68) (0.62)
MB -0.0046 0.0059
(-0.97) (0.79)
FIRMAGE -0.0004 0.0006
(-0.63) (0.85)
MNE -0.0201 -0.0022
(-0.88) (-0.08)
RD 0.2132 0.0845
(0.73) (0.35)
CAPX 0.0607 -0.1477*
(0.70) (-1.66)
PPE 0.0030 0.0220
(0.07) (0.37)
INTANG 0.0029 0.0526
(0.06) (0.74)
EQINC 3.1881* -1.0342
(1.69) (-0.35)
NOL -0.0425* -0.0283
(-1.79) (-1.08)
DNOL -0.0318 -0.1627
(-0.13) (-0.68)
CEOAGE 0.0005 0.0006
(0.38) (0.32)
CEOTENURE -0.0006 0.0008
(-0.31) (0.35)
MSHARE 0.1495 0.0540
(0.95) (0.21)
VEGA 0.0079 0.0584
(0.14) (0.81)
DELTA 0.0010 -0.0157
(0.14) (-1.18)
YEAR YES YES
INDUSTRY YES YES
R2 0.12 0.08
N 322 322
This table reports a propensity score matching technique to estimate Equation (1), which regresses corporate tax planning activity on CEO severance
pay. Panel A reports the first-stage Probit regression. Column (1) reports the pre-matching first-stage Probit regression and Column (2) reports the
post-matching first-stage Probit regression. Panel B compares matching variables. Column (1) compares pre-matching variables and Column (2)
compares post-matching variables. Panel C reports baseline regression results estimated using matched sample. In Columns (1) and (2), the
dependent variables are corporate tax planning measured as GAAP effective tax rate (GAAP) and cash effective tax rate (CASH). The main
independent variable is CEO severance pay (SERPAY), measured as the contracted total severance payment amount to a firm's CEO divided by
CEO wealth defined as the market value of firm equity shares held by the CEO. Other variables are defined in the Appendix A. Regressions include
year and industry fixed effects. Standard errors are clustered by firm. T-statistics are in the parentheses. ***, **, and * represent significance at 1%,
5% and 10% levels using one-tailed p-values when a directional prediction is made (two-tailed otherwise).
39

TABLE 5
Instrumental Variable Approach
First Stage Second Stage
(1) (2) (3)
SERPAY GAAP CASH
Intercept 0.2325 0.2400*** 0.0904
(0.25) (4.32) (1.29)
DISTANCE -0.2092***
(-3.48)
MSERPAY 0.9075***
(45.61)
PSERPAY -0.0038*** -0.0025**
(-3.07) (-1.77)
SIZE 0.0603 0.0060 0.0126**
(0.89) (1.48) (2.44)
ROA -2.3189*** 0.3086*** 0.1735**
(-2.69) (5.11) (2.53)
LEV -0.2719 -0.0311 -0.0777***
(-0.69) (-0.79) (-2.60)
MB 0.0060 -0.0005 0.0027
(0.19) (-0.32) (1.08)
FIRMAGE 0.0050 -0.0004 0.0001
(0.95) (-1.24) (0.19)
NNE -0.2357 -0.0459*** 0.0149
(-1.27) (-4.25) (1.09)
RD -0.9960 -0.0681 -0.1332
(-0.71) (-0.58) (-1.04)
CAPX 1.4989** 0.0675** -0.1142***
(2.54) (2.13) (-3.03)
PPE 0.2379 0.0211 -0.0290
(0.61) (0.98) (-1.08)
INTANG 0.6546* 0.1034*** 0.0683**
(1.70) (4.35) (2.46)
EQINC -12.2505 -0.3672 -1.5282
(-0.71) (-0.40) (-1.46)
NOL 0.1377 -0.0010 -0.0240**
(0.87) (-0.10) (-2.11)
DNOL 5.2947*** -0.1761** -0.0124
(3.67) (-1.98) (-0.10)
CEOAGE -0.0045 0.0008 0.0021**
(-0.38) (1.07) (2.33)
CEOTENURE -0.0226* -0.0001 -0.0017*
(-1.75) (-0.18) (-1.85)
MSHARE -0.4570 -0.0093 0.0198
(-1.45) (-0.59) (1.07)
VEGA -0.8489 -0.0417 -0.0173
(-1.49) (-1.60) (-0.49)
DELTA -0.0955 -0.0021 -0.0102**
(-1.08) (-0.58) (-2.29)
YEAR YES YES YES
INDUSTRY YES YES YES
R2 0.63 0.12 0.08
N 1,397 1,397 1,397
This table reports an instrumental variable approach to estimate Equation (1), which regresses corporate tax planning activity on CEO severance
pay. Column (1) reports first-stage regression results. Instruments are DISTANCE and MSERPAY, denoting geographic distance to local largest
SERPAY and local median SERPAY, respectively. Columns (2)-(3) report second-stage results. The dependent variables are corporate tax
planning measured as GAAP effective tax rate (GAAP) and cash effective tax rate (CASH). The main independent variable is instrumented CEO
severance pay (SERPAY), measured as the regression residuals estimated from the first-stage. Other variables are defined in the Appendix A. T-
40

statistics are in the parentheses. Standard errors are clustered by firm. ***, **, and * represent significance at 1%, 5% and 10% levels using one-
tailed p-values when a directional prediction is made (two-tailed otherwise).
41

TABLE 6
Alternative Measures of Key Variables
Panel A Alternative measures of severance pay
CEO cash compensation CEO total compensation
(1) (2) (3) (4)
GAAP CASH GAAP CASH
Intercept 0.2274*** 0.0773 0.2380*** 0.0952
(3.96) (1.10) (4.14) (1.35)
SERPAY -0.0004* -0.0008** -0.0060*** -0.0027
(-1.38) (-1.72) (-2.17) (-0.67)
Controls YES YES YES YES
YEAR YES YES YES YES
INDUSTRY YES YES YES YES
R2 0.12 0.08 0.12 0.07
N 1,408 1,408 1,411 1,411
Only severance paying firms Severance pay indicator
(5) (6) (7) (8)
GAAP CASH GAAP CASH
Intercept 0.2389*** 0.0831 0.2479*** 0.1068
(3.52) (0.95) (4.33) (1.50)
SERPAY -0.0021*** -0.0021** -0.0182*** -0.0164*
(-2.52) (-1.85) (-2.08) (-1.36)
Controls YES YES YES YES
YEAR YES YES YES YES
INDUSTRY YES YES YES YES
R2 0.13 0.07 0.12 0.08
N 972 972 1,413 1,413
Panel B Alternative measures of tax planning
(1) (2)
TAXSETTLE UTB
Intercept -1.8209 -3.2407
(-4.38) (-4.93)
SERPAY 0.0119** 0.0211**
(1.68) (1.88)
Controls YES YES
YEAR YES YES
INDUSTRY YES YES
2
R 0.25 0.37
N 1,413 1,413
This table presents regression results using alternative measures of key variables. To be concise, we only report variables of interest. Panel A checks
the robustness of the regression results to alternative measures of severance pay. Columns (1) - (2) scale CEO severance pay by CEO total cash
compensation (SERPAY). Columns (3) - (4) scale CEO severance pay by CEO total compensation. Columns (5) - (6) focus on severance paying
firms only. Columns (7) - (8) use a severance pay indicator to capture the existence of CEO severance pay. Panel B checks the robustness of
regression results to alternative tax measures. Column (1) measures tax planning as the natural logarithm of tax settlements with the IRS. Column
(2) measures tax planning as the natural logarithm of unrecognized tax benefits. All the regressions include control variables, year fixed effects,
and industry fixed effects. Standard errors are clustered by firm. T-statistics are in the parentheses. ***, **, and * represent significance at 1%, 5%
and 10% levels using one-tailed p-values when a directional prediction is made (two-tailed otherwise).
42

TABLE 7
Cross-Sectional Tests: The Role of Career Concerns
Panel A CEO tenure as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.2278*** 0.1379
(2.97) (1.49)
SERPAY 0.0007 0.0013
(0.23) (0.30)
SERPAY*LOWTENURE -0.0070*** -0.0084**
(-2.16) (-1.87)
LOWTENURE 0.0029 0.0192
(0.23) (1.18)
YEAR YES YES
INDUSTRY YES YES
R2 0.16 0.10
N 677 677
Panel B Shareholder rights as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.1544* 0.0629
(1.88) (0.57)
SERPAY -0.0011 0.0014
(-0.31) (0.41)
SERPAY*HIGHRIGHTS -0.0030 -0.0069**
(-0.74) (-1.83)
HIGHRIGHTS 0.0125 0.0038
(0.97) (0.22)
YEAR YES YES
INDUSTRY YES YES
R2 0.14 0.07
N 600 600
Panel C Idiosyncratic volatility as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.2636*** 0.0964
(2.94) (0.84)
SERPAY 0.0003 0.0034
(0.18) (0.77)
SERPAY*HIGHIVOL -0.0078*** -0.0108***
(-3.48) (-2.30)
HIGHIVOL 0.0323 0.0180
(1.63) (0.65)
YEAR YES YES
INDUSTRY YES YES
R2 0.10 0.06
N 710 710
Panel D First principal component as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.1785** 0.1933
(2.06) (1.56)
SERPAY -0.0025 0.0031
(-0.98) (0.56)
SERPAY*FIRSTPRIN -0.0043* -0.0100**
43

(-1.37) (-1.79)
FIRSTPRIN 0.0213 0.0123
(1.17) (0.47)
YEAR YES YES
INDUSTRY YES YES
R2 0.14 0.04
N 619 619
Panel E Second principal component as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.2376** 0.1995*
(2.54) (1.90)
SERPAY -0.0015 -0.0003
(-0.55) (-0.08)
SERPAY*SECONDPRIN -0.0044* -0.0069**
(-1.38) (-1.80)
SECONDPRIN 0.0095 0.0074
(0.69) (0.40)
YEAR YES YES
INDUSTRY YES YES
R2 0.13 0.06
N 619 619
Panel F Average principal component as a proxy for career concerns
(1) (2)
GAAP CASH
Intercept 0.2447** 0.1394
(2.50) (1.08)
SERPAY 0.0005 0.0010
(0.28) (0.32)
SERPAY*AVGPRIN -0.0068*** -0.0076***
(-2.75) (-2.25)
AVGPRIN 0.0190 0.0201
(1.06) (0.87)
YEAR YES YES
INDUSTRY YES YES
R 2 0.13 0.05
N 619 619
This table reports cross-sectional regression results (based on the level of managerial career concerns) for Equation (1), which regresses corporate
tax planning activity on CEO severance pay. In Columns (1) and (2), the dependent variables are corporate tax planning measured as GAAP
effective tax rate (GAAP) and cash effective tax rate (CASH). The main independent variable is instrumented CEO severance pay (SERPAY),
measured as the regression residuals estimated from the first-stage regression in the instrumental variable estimation (the results are unchanged if
instead the model is based on ordinary least-squares regression). In Panels A to F, managerial career concerns are measured with CEO tenure (Panel
A), shareholder rights (Panel B), idiosyncratic volatility (Panel C), first principal component of the three proxies (Panel D), second principal
component of the three proxies (Panel E), average of the first and second components of the three proxies (Panel F). Across all the panels, our focus
variable is the interaction between severance variable and career concern variable. Variable definitions are in the Appendix A. Regressions include
year and industry fixed effects. Standard errors are clustered by firm. T-statistics are in the parentheses. ***, **, and * represent significance at 1%,
5% and 10% levels using one-tailed p-values when a directional prediction is made (two-tailed otherwise).
44

SSRN Id2802426

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SSRN Id2802426

Uploaded by

Copyright:

Available Formats

CEO Severance Pay and Corporate Tax Planning

Jenny Xinjiao Guan

Running Head: CEO Severance Pay and Corporate Tax Planning

Electronic copy available at: https://ssrn.com/abstract=2802426

Keywords: corporate tax planning; CEO severance pay; risk-taking incentives

Electronic copy available at: https://ssrn.com/abstract=2802426

thus, is analogous to value-increasing investment projects favored by risk-neutral shareholders

managers to take more risk (Crocker and Slemrod 2005).

job termination (Chyz and Gaertner 2018).

Electronic copy available at: https://ssrn.com/abstract=2802426

Severance pay represents a significant portion of CEOs’ compensation package, as over 75

shareholders and managers (Ju et al. 2014; Cadman et al. 2016).

Electronic copy available at: https://ssrn.com/abstract=2802426

identified. Our result continues to hold.

sensitive to the incentive effect of CEO severance pay.

Electronic copy available at: https://ssrn.com/abstract=2802426

the large variance in corporate tax planning activities.

and Gaertner 2018).

Electronic copy available at: https://ssrn.com/abstract=2802426

averse managers to take value-increasing risks.

LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT

Corporate Tax Planning within an Agency Framework

Electronic copy available at: https://ssrn.com/abstract=2802426

detection, and the extent to which an individual is risk averse.

induce otherwise risk-averse managers to engage in corporate tax planning.

than using earnings metrics.

Electronic copy available at: https://ssrn.com/abstract=2802426

CEO Severance Pay and Corporate Tax Planning

Electronic copy available at: https://ssrn.com/abstract=2802426

Electronic copy available at: https://ssrn.com/abstract=2802426

corporate tax planning.

and thus was compensated for losses due to his dismissal.

Electronic copy available at: https://ssrn.com/abstract=2802426

with corporate tax planning.

DATA and SAMPLE

Hand-collected Data on CEO Severance Pay

Electronic copy available at: https://ssrn.com/abstract=2802426

in a comprehensive manner (Cadman et al. 2016).

benefits that vest upon dismissal, and other uncategorized portions.

Electronic copy available at: https://ssrn.com/abstract=2802426

remove the impact of outliers.

CEO SEVERANCE PAY AND CORPORATE TAX PLANNING

pay and corporate tax planning:

TPi,t = β0 + β1SERPAYi,t + β2SIZEi,t + β3ROAi,t + β4LEVi,t + β5MBi,t + β6FIRMAGEi,t

Electronic copy available at: https://ssrn.com/abstract=2802426

Electronic copy available at: https://ssrn.com/abstract=2802426

engage in corporate tax planning.

year fixed effects are also included.

Electronic copy available at: https://ssrn.com/abstract=2802426

contrast highlights the economic significance of severance pay in CEO compensation.

univariate correlations support H1.

economically significant. A one-standard-deviation increase in CEO severance pay relative to

Electronic copy available at: https://ssrn.com/abstract=2802426

Electronic copy available at: https://ssrn.com/abstract=2802426

Severance Pay Components

pension benefits do not affect corporate tax planning.

Propensity Score Matching

Electronic copy available at: https://ssrn.com/abstract=2802426

Electronic copy available at: https://ssrn.com/abstract=2802426

Instrumental Variable Approach

Electronic copy available at: https://ssrn.com/abstract=2802426

supporting our predictions.

that the coefficients on the three interactions, SERPAYFIRSTPRIN, SERPAYSECONDPRIN,