Accounting Forum

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/racc20

Chief financial officer overconfidence and earnings

management

Lu Qiao, Emmanuel Adegbite & Tam Huy Nguyen

To cite this article: Lu Qiao, Emmanuel Adegbite & Tam Huy Nguyen (28 Apr 2023): Chief
financial officer overconfidence and earnings management, Accounting Forum, DOI:
10.1080/01559982.2023.2196045

To link to this article: https://doi.org/10.1080/01559982.2023.2196045

© 2023 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 28 Apr 2023.

Submit your article to this journal

Article views: 2945

View related articles

View Crossmark data

Citing articles: 5 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=racc20
ACCOUNTING FORUM
https://doi.org/10.1080/01559982.2023.2196045

Chief financial officer overconfidence and earnings

management
Lu Qiaoa, Emmanuel Adegbite b,c
and Tam Huy Nguyenb,d
a
School of Business, University of Leicester, Leicester, United Kingdom; bNottingham University Business
School, University of Nottingham, Nottingham, United Kingdom; cJCUS Business School, James Cook
University, Singapore, Singapore; dUniversity of Economics and Business, Vietnam National University,
Hanoi, Vietnam

ABSTRACT ARTICLE HISTORY

This study explores the relationship between overconfident Chief Received 28 October 2021
Financial Officers (CFOs) and earnings management. Through the Accepted 24 March 2023
lens of upper echelons and overconfidence theories, and using a
KEYWORDS
large sample of 14,156 observations of US firms from 1999 to CFO overconfidence;
2021 inclusive, our study finds that overconfident CFOs are earnings management;
positively associated with earnings management. We show that overconfidence theory;
overconfident CFOs use earnings management to reduce upper echelons theory
earnings volatility, given that a smooth performance can release
their financing pressure. In doing this, we rule out another ACCEPTED BY
possible explanation of overconfident CFOs engaging in earnings Andrei Filip
management to pursue high compensation. Our findings pass a
series of robustness tests, including entropy balancing, the
Difference-in-Differences test based on the propensity score
matching sample (PSM-DID), and alternative measures of main
variables. Our study provides a new determinant of earnings
management that has more explanatory power than CFO
demographic traits – i.e. CFO cognitive biases. Our findings
nonetheless show the “bright” side of CFO overconfidence,
helping investors, regulators, and policymakers understand
overconfident CFOs’ financial reporting decisions.

1. Introduction
Overconfidence, also known as “the root of all cognitive biases”, is considered a man-
ager’s vital psychological trait that significantly affects firm decisions (e.g. investment,
financing, and risk management) (Bazerman & Moore, 2013). Due to Chief Executive
Officers (CEOs) setting the tone at the top, most studies only investigate the influence
of CEO overconfidence (i.e. Chen et al., 2020; Hsieh et al., 2014; Lin et al., 2020; Malmen-
dier & Tate, 2005).
Malmendier et al. (2022) raise one concern, that this singular attention might generate
some biased inferences, as it ignores the role of other managers who are primarily

CONTACT Emmanuel Adegbite [email protected]

Supplemental data for this article can be accessed online at https://doi.org/10.1080/01559982.2023.2196045.
© 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License
(http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any
medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. The terms on which
this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
2 L. QIAO ET AL.

responsible for the outcome of a particular realm. Given that CFOs, rather than CEOs,
are in charge of financing decisions (Mian, 2001), Malmendier et al. (2022) show that
CFO overconfidence dominates CEO overconfidence in the choice of external finance
method and leverage. In addition to Malmendier et al. (2022), a growing number of
studies shed light on CFO overconfidence. After controlling for CEO overconfidence,
these studies remain to discover that CFO overconfidence is positively associated with
cost stickiness (Ben-David et al., 2013; Chen et al., 2022) and hoarding bad news
(Qiao et al., 2022). Prior findings indicate that overconfidence biases CFOs’ forecasting,
risk-taking, and information disclosure. Such cognitive bias might also be associated with
the CFOs’ other decisions where they play a more vital role than other managers.
Earnings management is a practice influencing financial reporting primarily at the
CFOs’ discretion. Specifically, Geiger and North (2006) suggest significant changes in
discretionary accruals surrounding the appointment of a new CFO. Jiang et al. (2010)
find that the CFO equity incentive is stronger than that of the CEO in explaining
firms’ earnings management activities. Peni and Vähämaa (2010) show that earnings
management is affected more by the CFOs’ gender than by the CEOs’. This piece of evi-
dence implies that CFOs have vital roles in earnings management decisions.
Previous studies have competing views on earnings management. On the one hand,
some suggest that earnings management can improve the information value of earnings
through communicating private information, thereby benefiting financial reporting
users’ decision-making (e.g. Arya et al., 2003; Holthausen, 1990; Jiraporn et al., 2008).
On the other hand, much evidence shows that managers use aggressive earnings manage-
ment to pursue personal gains, thereby impairing financial reporting users’ interests (e.g.
Chen et al., 2017; Francis et al., 2016; Nguyen & Soobaroyen, 2019; Schrand & Zechman,
2012). For example, Enron’s aggressive earnings management led to financial fraud,
causing serious damage to stakeholders’ wealth and the American economy (Davis,
2002). In addition to Enron, Lucent, Cendant, and MicroStrategy abused earnings man-
agement, resulting in more than 34 billion dollars in stock market losses in three days
(Magrath & Weld, 2002). The above discussion shows that the impact of earnings man-
agement on financial report users’ decisions and interests is substantial, and has kept
earnings management attracting the constant attention of researchers.
Surprisingly, current earnings management studies overlook the significance of CFO
overconfidence. This study fills this gap. We predict there is a positive relationship
between CFO overconfidence and earnings management. From the perspective of infor-
mation signalling, previous studies show that managers tend to use earnings manage-
ment to convey private information. Specifically, managers would like to present a
smoother earnings stream to signal their firms’ earnings stability, reducing the cost of
borrowing (Gassen & Fülbier, 2015; Jung et al., 2013; Li & Richie, 2016; Minton &
Schrand, 1999; Trueman & Titman, 1988). According to the upper echelons and over-
confidence theories,1 prior studies suggest that overconfident CFOs perceive external
funding to be costly as they deem the market to undervalue their firms’ performance
(Ben-David et al., 2013; Malmendier et al., 2022). In this case, as CFOs are responsible

1
Upper echelons theory states that managers’ cognitive biases can explain their decisions (Hambrick & Mason, 1984).
Overconfidence theory discover four overconfidence manifestations: over-optimism (Weinstein, 1980), the above-
average effect (Svenson, 1981), miscalibration(Alicke, 1985), and the illusion of control (Taylor & Brown, 1988).,
which will be discussed in detail in the next section.
 ACCOUNTING FORUM 3

for financing, we predict that overconfident CFOs might tend to use earnings manage-
ment to smooth earnings, hence reducing the cost of borrowing.
Another potential explanation for the positive relationship between CFO overconfi-
dence and earnings management is that overconfident CFOs manipulate earnings to
seek personal financial gains. Adopting the agency theory framework,2 prior studies
show that CFOs might opportunistically utilise earnings management to seek higher
compensation (Jiang et al., 2010; Watts & Zimmerman, 1978). This incentive might be
stronger for overconfident CFOs since they have incentive-heavy compensation con-
tracts (Humphery-Jenner et al., 2016).
Using a large sample of data (14,156 observations) on US companies from 1999 to
2021 inclusive, we find a positive relationship between CFO overconfidence and earnings
management. The positive association is more profound when the firms have more vola-
tile performance or have external financing needs. These findings indicate that overconfi-
dent CFOs, due to financing concerns, manage earnings to provide smooth earnings. Our
findings pass a series of robustness tests, including entropy balancing, the Difference-in-
Differences test based on the Propensity Score Matching sample (PSM-DID), and
alternative measures of main variables. In an additional test, we rule out an alternative
explanation of the positive relationship between CFO overconfidence and earnings man-
agement, relating to the CFOs seeking personal gains.
This study contributes to the earnings management, CFO, and overconfidence litera-
ture. Although few studies document significant relationships between CFO demo-
graphic traits and earnings management (Barua et al., 2010; Peni & Vähämaa, 2010),
Ge et al. (2011) argue that CFOs’ psychological traits explain their financial reporting
decisions better than demographic traits. Consistently, when we incorporate CFO over-
confidence in testing the determinants of earnings management, most coefficients on
male CFOs become insignificant. Thus, our study provides a new determinant of earn-
ings management that has more explanatory power than CFO demographic traits.
Further, after controlling for CEO overconfidence, the effect of CFO overconfidence
on earnings management remains significant, but that of CEO overconfidence has
become insignificant. This finding provides more fruitful evidence to support the litera-
ture suggesting that the CFO has a dominant effect on earnings management (Geiger &
North, 2006; Jiang et al., 2010; Peni & Vähämaa, 2010).
Moreover, our study closely relates to Ham et al. (2017) which first sheds light on the
relationship between one CFOs’ psychological trait, narcissism, and earnings manage-
ment. They suggest that narcissistic CFOs engage in earnings management to pursue per-
sonal wealth at the expense of others. However, we find that overconfident CFOs, in
contrast to narcissistic CFOs, use earnings management for financing purposes rather
than self-interest, which enriches the study of the psychological characteristics of CFOs.
Also, overconfident CFOs’ earnings management incentives reveal the positive result of
CFO overconfidence, adding to the literature discussing the advantages and disadvantages
of managerial overconfidence (Chen et al., 2022; Malmendier et al., 2022; Qiao et al., 2022).
Our study also makes theoretical contributions. The upper echelons theory implies
that managers’ cognitive biases affect their decisions (Hambrick & Mason, 1984). As

2
According to agency theory, an agency problem occurs when agents, such as the CFO, make a decision that contradicts
the interests of the shareholders (principals) (Jensen & Meckling, 1976).
4 L. QIAO ET AL.

executives’ cognitive biases are difficult to observe, early empirical studies use executives’
demographic characteristics as proxies for cognitive biases based on the suggestion by
Hambrick and Mason (1984). However, this practice is controversial as overconfidence,
rather than demographic traits, is the origin of all cognitive biases (e.g. Ge et al., 2011).
Given that Malmendier and Tate (2005) create a novel overconfidence measurement to
make the empirical capture of overconfidence more valid, we adopt this method modified
by Campbell et al. (2011) and provide significant findings that are consistent with the
overconfidence theory. Therefore, our study provides empirical evidence to bridge the
gap between overconfidence and upper echelons theories.
Finally, our study has strong implications for financial report users. Previous research
only focuses on the overconfident CEOs’ earnings management decisions (Hsieh et al.,
2014; Schrand & Zechman, 2012). Such a singular investigation may undermine
financial report users’ ability to achieve a comprehensive understanding of the forcefulness
of top managers’ cognitive biases on accounting information. Our study can enhance
financial report users’ understanding that the real effect of managerial overconfidence
on earnings management decisions originates from CFO overconfidence. In addition,
our research shows the “bright side” of CFO overconfidence by revealing that overconfi-
dent CFOs engage in earnings management to improve the firm’s financing ability
rather than for their own interests. This finding also partly answers the puzzle of why
boards continue to hire overconfident CFOs, despite some research showing the adverse
consequences of CFO overconfidence (Chen et al., 2022; Qiao et al., 2022).
The remainder of this paper is arranged as follows: Section 2 provides the theoretical
framework, literature review, and hypotheses development. The data and research design
are described in section 3. Section 4 reports the main results. Underpinning mechanisms
are tested in section 5. Several robustness tests are conducted in section 6. Section 7 pre-
sents the results of some additional analyses, and section 8 concludes the study.

2. Theoretical framework, literature review, and hypotheses development

Given the high likelihood of overconfidence among top executives (Bernardo & Welch,
2001; Goel & Thakor, 2008), a growing number of studies investigate how this bias affects
managers’ decisions, including financial policies (e.g. dividend payout and cash holding)
(Chen et al., 2020; Deshmukh et al., 2013), investments (e.g. M&A, innovation, and CSR)
(Hirshleifer et al., 2012; Malmendier & Tate, 2005; Sauerwald & Su, 2019) and operations
(e.g. cost and risk management) (Chen et al., 2022; Kim et al., 2016; Qiao et al., 2022),
using overconfidence theory and upper echelons theory.
Upper echelons theory documents that managerial cognitive biases influence their
decisions (Hambrick & Mason, 1984). Cognitive bias stems from overconfidence (Bazer-
man & Moore, 2013). Overconfidence has four manifestations, as suggested by overconfi-
dence theory, encompassing over-optimism, the above-average effect, miscalibration, and
the illusion of control. Specifically, over-optimistic individuals perceive the probability of
fortunate events to be greater than the probability of unfortunate ones (Weinstein, 1980).
The above-average effect is people’s perception that their abilities are superior to others
(Svenson, 1981). Miscalibration means that people might overestimate their ability to
reduce future uncertainty and make correct forecasts (Alicke, 1985), comparable to the
“illusion of control”. The latter implies that people have a false belief in their ability to
affect an incident, even if the event is entirely unintentional (Taylor & Brown, 1988).
 ACCOUNTING FORUM 5

In the context of managerial decisions, financing decisions are significantly deter-

mined by CFO overconfidence. Specifically, Hackbarth (2008) provides theoretical evi-
dence that overconfident managers are more likely to issue new debt and choose
higher debt levels when they need external financing. Overconfident managers overesti-
mate their firm’s future performance and hence regard external funding as excessively
costly. Because variations in views about future cash flows are typically more sensitive
to stock prices, overconfident managers believe that issuing equity is more expensive
than issuing debt. Therefore, firms with overconfident managers tend to issue debt
rather than equity when they require external financing. Consistent with Hackbarth
(2008), Malmendier et al. (2022) show empirical evidence that overconfident CFOs
tend to use debt financing as they perceive equity financing to be more costly. They
further highlight that firms’ financing choices depend on CFO overconfidence rather
than CEO overconfidence.
When overconfident CFOs are confronted with the (overestimated) high cost of exter-
nal funding, they may seek a solution to reduce this cost, hence strengthening the firms’
financing ability. A common method admitted by CFOs to achieve this aim is to restrict
earnings volatilities via manipulating earnings (Dichev et al., 2013). This idea is consist-
ent with survey findings of Graham et al. (2005), which show that CFOs desire stable
earnings to reduce the cost of capital, even at the cost of sacrificing long-term firm
value. Thus, we anticipate that overconfident CFOs have the incentive to decrease earn-
ings volatility by manipulating earnings, thereby lowering what they perceive to be over-
priced financing costs.
One might be concerned that earnings manipulation is easily detected (Cohen et al.,
2008). If aggressive earnings management is detected, the CFO, as the principal account-
ing decision maker, might lose credibility among stakeholders (Goel & Thakor, 2003)
and escalate their turnover risk (Collins et al., 2009). However, overconfident CFOs
have high-risk tolerance (Ben-David et al., 2013; Chen et al., 2022) and are thereby
more likely to ignore or underestimate the detection risk than their rational counterparts.
Therefore, the high detection risks may not lessen the incentives of overconfident CFOs
to use earnings management.
Based on the above discussions, we propose our first hypothesis.
H1: There is a positive association between overconfident CFOs and earnings management.

If overconfident CFOs manage earnings because they are keen to strengthen the ability of
external financing by signalling performance stability, we should expect this incentive to be
stronger if their firms have more volatile performance or have external financing needs.
Volatile performance increases creditors’ perceptions of firms’ risks. Earnings vola-
tility undermines earnings predictability and increases information risk, resulting in
creditors’ inability to accurately forecast the firm’s future performance during the
credit-granting decision-making stage (Graham et al., 2005). Besides, cash flow volatility
exacerbates firms’ default risks, potentially leading to firms’ failure to pay principal and
interest in full and on time (Brogaard et al., 2017).
To restrict creditors’ high perceived risk, managers use earnings smoothing as an
information signal to convey private information (Merton, 1974). This notion is
further supported by Trueman and Titman (1988). They create a model where firms
smooth earnings to lessen creditors’ perceived bankruptcy risk, resulting in lower
6 L. QIAO ET AL.

borrowing costs. Empirical evidence also suggests stable earnings are negatively associ-
ated with the cost of debt (e.g. Gassen & Fülbier, 2015; Jung et al., 2013; Li & Richie,
2016; Minton & Schrand, 1999; Sha et al., 2021). Therefore, we predict that overconfident
CFOs faced with highly volatile performance are more likely to smooth earnings through
earnings management, thereby reducing the cost of debt.
In addition, as overconfident CFOs believe that their equity is undervalued, it is poss-
ible that their earnings smoothing incentives are derived from mitigating the perceived
equity undervaluation and reducing the cost of equity. Specifically, Francis et al. (2004)
document a negative relationship between smooth earnings and the cost of equity.
However, McInnis (2010) finds no significant relationship between earnings volatility
and the cost of equity and argues that the equity market does not drive managers to
favour smooth earnings. Given the debatable relationship between earnings smoothing
and the cost of equity, we estimate that the incentives of overconfident CFOs to manip-
ulate earnings may not be changed significantly by equity issues.
Based on the above discussions, we propose our second and third hypotheses.
H2: The positive relationship between overconfident CFOs and earnings management is
more pronounced when firms have more volatile performance.

H3: The positive relationship between overconfident CFOs and earnings management is
more pronounced when firms have external financing needs.

3. Data and research design

3.1 Sample and data
Our main sample is based on 14,156 observations of US firms from 1999 to 2021.3 We
obtain information about executive compensation from ExecuComp. We define the
CFO position according to executive titles in ExecuComp, including chief financial
officer, CFO, controller, treasurer, finance, and vice president-finance (Jiang et al.,
2010). We collect financial data from the CRSP/Compustat Merged database (CCM).
We gather institutional ownership and governance data from the Thomson-Reuters
13F database and the BoardEx database, respectively. Financial firms (SIC:6000-6999)
and utility firms (SIC:4900-4999) are excluded since their operational and financial struc-
tures vary from those of other businesses. The detailed sample selection construction is
shown in Table 1. Following previous studies (e.g. Hribar & Nichols, 2007), we winsorise
all continuous variables (except for absolute value) at the 1st and 99th percentiles and all
absolute values of continuous variables at the 99th percentiles to eliminate the impact
of outliers.

3.2 Measurement
3.2.1 Earnings management measurement
This study uses accrual-based earnings management as a proxy for earnings manage-
ment. Accrual-based earnings management is a way to manipulate earnings by changing
3
Our sample period starts in 1999 as corporate governance information (control variables) in BoardEx is available begin-
ning in 1999.
 ACCOUNTING FORUM 7

Table 1. Sample selection.

Steps Details Observations
Step 1 Initial observations from CCM database (1999–2021) 141,075
Step 2 Exclude observations without CFOs’ compensation information in the ExecuComp database (101,992)
Step 3 Exclude observations without board information in the BoardEx database (5,403)
Observations with data available in CCM, ExecuComp, and BoardEx databasesa 33,680
Step 4 Exclude observations in the financial industry (SIC:6000-6999) (6,783)
Step 5 Exclude observations in the utility sector (SIC:4900-4999) (1,555)
Observations before deleting the missing values in various regressions 25,342
Step 6 Number of observations after deleting observations with missing values for the main sample 14,156
a
Since Thomson-Reuters 13F database only discloses information that firms have institutional investors, we retain insti-
tutional ownership information in the matched sample (i.e. data available in CCM, ExecuComp and Thomson-Reuters
13F database). We replace institutional ownership with zero for unmatched sample (i.e. data available in CCM and
ExecuComp and unavailable in Thomson-Reuters 13F database). This step is not listed in the Table 1, because no obser-
vation is deleted.

accounting estimates or methods, which are primarily determined at the CFO’s discre-
tion (Baker et al., 2019; Jiang et al., 2010). Dechow et al. (1995) introduce the
modified Jones model, which is widely utilised in the research on accrual-based earnings
management. In line with Hsieh et al. (2014), we use the modified Jones model. At least
15 observations are required for each industry year group (Zang, 2012). The modified
Jones model is as follows:
TAit 1 DREVit − DRECit PPEit
= b0 + b1 + b2 + b3 + 1it, (1)
Ait − 1 Ait − 1 Ait − 1 Ait − 1
where, TAit is the total accrual calculated from the cash flow statement (CCM variable:
IBC–(OANCF–XIDOC)).4 β0 is the unscaled intercept.5 ΔREVit is the change of
revenue (REVT–L. REVT). ΔRECit is the change of accounts receivable (RECT–L.
RECT). PPEit is the property, plant, and equipment (PPEGT). Ait−1 is the lagged
total asset (L. AT). The discretionary accrual (AEMit) is the estimated residual from
Equation (1).
We use the absolute value of AEMit (AbsAEMit) since it captures accrual reversals fol-
lowing AEM (Cohen et al., 2008), and no specific directions for AEM are predicted by
our hypotheses.

3.2.2 Overconfidence measurement

The option-based technique suggested by Malmendier and Tate (2005) is the broadly uti-
lised managerial overconfidence measure in existing research (Campbell et al., 2011;
Hsieh et al., 2018; Huang et al., 2016) and is the most robust compared to alternative
interpretations (Malmendier & Tate, 2015). Thus, this study adopts the option-based
method to measure CFO and CEO overconfidence.
Malmendier and Tate (2005) use managers’ option exercise behaviours to measure
overconfidence. If managers put off exercising options that are more than 67% in the
money, they are overconfident. Because the ExecuComp database lacks detailed data
for calculating moneyness prior to 2006, we use the average moneyness of the CFOs’
4
In the measurements of AEM, the item names in parentheses are from the CCM.
5
Following recommendations from Kothari et al. (2005), Collins et al. (2017) and Lara et al. (2020), this study includes an
unscaled intercept in the regression of the modified Jones model to incorporate the impact of firm growth and per-
formance. The results remain qualitatively similar when the unscaled intercept is excluded.
8 L. QIAO ET AL.

(CEOs’) option portfolio to measure CFO (CEO) overconfidence following Campbell

et al. (2011). First, we divide the total realisable value of the exercisable options (Execu-
Comp variable: OPT_UNEX_EXER_EST_VAL) by the number of exercisable options
(OPT_UNEX_EXER_NUM) to calculate the average realisable value per option.6 Next,
the average exercise price per option is calculated by extracting the average realisable
value per option from the stock price at the fiscal year-end (PRCC_F). Finally, we use
the average realisable value per option divided by the average exercise price per option
to get the average moneyness of the options.
Holder67CFOit (Holder67CEOit) is an indicator variable that equals one if CFOs
(CEOs) are reluctant to exercise vested options that are no less than 67% in the
money for the first time, and zero otherwise. CFOs (CEOs) would remain in their classifi-
cation until the end of their tenure (Hirshleifer et al., 2012).7 In addition, we change the
threshold to eliminate the bias of choosing the threshold on the overconfidence measure.
Holder100CFOit (Holder100CEOit), an indicator variable, equals one if CFOs (CEOs)
delay exercising vested options that are larger than 100% in the money for the first
time, and zero otherwise. CFOs (CEOs) would remain to be recognised as such until
the end of their tenure (Hirshleifer et al., 2012).

3.3 Main model

For testing the relationship between CFO overconfidence and numerous control factors
on earnings management, our study adopts the following regression:
AbsAEMit = b0+b1OverCFOit + b2OverCEOit + b3MaleCFOit + b4MaleCEOit + b5MTBit
+b6CashFlowit + b7ROAit + b8FirmSizeit + b9Leverageit + b10FirmAgeit
+b11AbsREMit + b12Big4it + b13HighLitigationRiskit + b14InsHoldingsit
+b15CEOdualityit + b16BoardSizeit + b17BoardIndependenceit
+Firmfixedeffects + Yearfixedeffects + 1it,
(2)
where, the dependent variable, the absolute value of accrual-based earnings management
(AbsAEMit), is measured by the modified Jones model (Dechow et al., 1995; Zang, 2012).
The variable of interest is the overconfident CFO (OverCFOit), proxied by Holder67CFOit
and Holder100CFOit. Following Campbell et al. (2011), Hirshleifer et al. (2012), and Chen
et al. (2022), we use managers’ option exercise behaviours to measure overconfidence.
We expect β1 to be positive, confirming the positive association between overconfident
CFOs and earnings management.
We control for CEO overconfidence (OverCEOit), proxied by Holder67CEOit and
Holder100CEOit, as overconfident CEOs are positively associated with earnings man-
agement decisions (Hsieh et al., 2014). Our study also considers including top man-
agers’ gender, as prior studies suggest that CFO gender (MaleCFOit) and CEO
gender (MaleCEOit) sway earnings management practice (e.g. Barua et al., 2010;
Gupta et al., 2020).
6
In the measurement of overconfidence, the variable names in parentheses are from the ExecuComp database.
7
The results remain qualitatively similar when we require CFOs to hold vested options that are no less than 67% in the
money at least twice (Malmendier & Tate, 2005).
 ACCOUNTING FORUM 9

Taking other possible factors into account, our study controls for firm growth oppor-
tunities proxied by the market-to-book ratio (MTBit), firm performance proxied by cash
flow (CashFlowit), and return on asset (ROAit), which have a significant link to discre-
tionary accruals (Kothari et al., 2005; Zang, 2012). We control for some other firm
characteristics that are regarded as determinants of earnings management practice,
such as firm size (FirmSizeit) (Dechow & Dichev, 2002; Zang, 2012), leverage (Leverageit)
(DeFond & Jiambalvo, 1994; Hsieh et al., 2014), and firm age (FirmAgeit) (Capalbo et al.,
2018; Gul et al., 2009). Additionally, our study controls real earnings management
(AbsREMit) because managers strategically use different methods (Cohen et al., 2008;
Zang, 2012).
We also control for the external monitoring, including the Big Four auditing firms
(Big4it), litigation risk (HighLitigationRiskit), and institutional ownership (InsHoldingsit)
(Boahen & Mamatzakis, 2020; Egan & Xu, 2020; Fraser et al., 2004; Koh, 2007; Matsu-
moto, 2002; Roychowdhury, 2006) and internal governance, encompassing CEO
duality (CEOdualityit), board size (BoardSizeit) and board independence
(BoardIndependenceit) (Man & Wong, 2013). The CEO duality (CEOdualityit) also can
reflect CEO power (DeBoskey et al., 2019). We control this effect, as Friedman (2014)
finds that powerful CEOs can pressure CFOs to manipulate earnings. Finally, we
include firm and year dummies to control the time-invariant firm and year effects.
Detailed variable measurements are outlined in Appendix A.

4. Main results
4.1 Descriptive statistics
Table 2 shows the descriptive statistics. We find a similar mean value (0.049) of AbsAEM
as reported in prior studies (Cohen et al., 2008). Further, 55.2% of CFOs indicated by the
mean value of Holder67CFO are overconfident, which implies that overconfidence is a
common characteristic among CFOs.8 This proves the point made by Black and Galle-
more (2013) and Malmendier et al. (2022) that only examining CEO overconfidence
and ignoring CFO overconfidence might underestimate the significance of overconfi-
dence. The summary statistics for the control variables are in line with previous
studies. For instance, the mean values of the MTB, InsHoldings, and Leverage are
similar to the findings of Ali and Zhang (2015).

4.2 Pairwise correlations

We conduct the Pearson correlation test. As shown in Online Appendix A, the corre-
lations of Holder67CFO (Holder100CFO) with the other control variables do not
exceed 0.5, and the variance-inflating factors (VIFs) (not tabulated for brevity) of all
independent variables are less than the threshold of 10 (Gujarati et al., 2012), implying
that multicollinearity is not an issue when interpreting the regression results. The
Pearson correlation coefficient between CFO and CEO overconfidence is high, which
means that both should be taken into account to mitigate the omitted variable bias.
8
Chen et al. (2022) identify 51% of CFOs as overconfident CFOs, using the same overconfidence measurement for the
period from 1992 to 2015 inclusive.
10 L. QIAO ET AL.

Table 2. Descriptive statistics.

N Mean SD p25 Median p75
AbsAEM 14,156 0.049 0.054 0.015 0.034 0.062
Holder67CFO 14,156 0.552 0.497 0 1 1
Holder67CEO 14,156 0.654 0.476 0 1 1
Holder100CFO 14,049 0.430 0.495 0 0 1
Holder100CEO 14,080 0.544 0.498 0 1 1
MaleCFO 14,156 0.906 0.292 1 1 1
MaleCEO 14,156 0.966 0.182 1 1 1
MTB 14,156 3.143 4.232 1.500 2.364 3.867
CashFlow 14,156 0.105 0.081 0.063 0.103 0.150
ROA 14,156 0.041 0.107 0.018 0.054 0.091
FirmSize 14,156 7.439 1.583 6.308 7.318 8.440
Leverage 14,156 0.214 0.184 0.041 0.197 0.325
FirmAge 14,156 3.151 0.660 2.639 3.135 3.738
AbsREM 14,156 0.274 0.262 0.089 0.196 0.373
Big4 14,156 0.905 0.294 1 1 1
HighLitigationRisk 14,156 0.094 0.291 0 0 0
InsHoldings 14,156 0.537 0.404 0 0.706 0.876
CEOduality 14,156 0.516 0.500 0 1 1
BoardSize 14,156 2.174 0.245 1.946 2.197 2.303
BoardIndependence 14,156 0.808 0.168 0.714 0.833 0.900
VolEarnings 9,923 0.045 0.055 0.012 0.025 0.053
VolCFO 9,923 0.039 0.033 0.017 0.029 0.049
EFN 11,026 0.424 0.494 0 0 1
DebtIssuer 13,653 0.384 0.486 0 0 1
EquityIssuer 13,236 0.372 0.483 0 0 1
Notes: This table presents summary statistics. The sample consists of 14,156 firm-year observations from 1999 to 2021 in
US firms with financial, executive compensation, and board data available from CCM, ExecuComp and BoardEx. See
Appendix A for variable definitions. Variables of interest are marked in bold.

4.3 Regression results – CFO overconfidence and earnings management (H1)

Table 3 displays the results of the estimating Equation (2) which show the association
between CFO overconfidence and earnings management. The full sample estimations
are listed in columns (1) and (2), while the analyses based on the entropy balancing
sample are shown in columns (3) and (4).
Holder67CFO (Holder100CFO) has a positive coefficient at a 1% (1%) significance level
in column (1) ((2)), indicating a statistically significant positive association between over-
confident CFOs and earnings management. These results also have economic signifi-
cance. Holder67CFO (Holder100CFO) will lead to a 12.2% increase in AbsAEM at the
mean.9 Therefore, we show statistically and economically significant evidence that
overconfident CFOs are positively associated with earnings management, supporting
hypothesis 1 (H1).
To improve the robustness of the initial results, we use a more restrictive approach:
entropy balancing, as it essentially recalibrates the observation weights by adjusting for
systematic and random differences in the control variables.10 Specifically, to construct
a sample to test earnings management, all observations with overconfident CFOs are
reweighted to match observations with non-overconfident CFOs, based on the control
variables used in Equation (2). The samples are matched on the mean, variance, and
skewness of covariate distribution. Columns (3) and (4) show the results of using the
9
12.2% = 0.006/0.049 × 100%, where 0.006 is the coefficient of Holder67CFO (Holder100CFO) and 0.049 is the mean value
of AbsAEM.
10
We thank the anonymous referee for this suggestion.
 ACCOUNTING FORUM 11

Table 3. The relationship between CFO overconfidence and earnings management activities.
(1) (2) (3) (4)
Sample Full sample Entropy balancing sample
Holder67CFO 0.006*** 0.004***
(0.001) (0.001)
Holder67CEO 0.001 0.002
(0.001) (0.001)
Holder100CFO 0.006*** 0.005***
(0.001) (0.002)
Holder100CEO 0.002 0.003*
(0.001) (0.002)
MaleCFO 0.003 0.003 0.003 0.005*
(0.002) (0.002) (0.003) (0.003)
MaleCEO 0.001 0.001 −0.002 −0.004
(0.004) (0.004) (0.005) (0.006)
MTB 0.000 0.000 0.000 0.000
(0.000) (0.000) (0.000) (0.000)
CashFlow −0.010 −0.009 −0.066*** −0.057**
(0.019) (0.019) (0.023) (0.023)
ROA −0.167*** −0.168*** −0.090*** −0.101***
(0.016) (0.016) (0.023) (0.022)
FirmSize 0.001 0.001 0.001 0.001
(0.002) (0.002) (0.002) (0.002)
Leverage −0.015** −0.015** −0.012 −0.012*
(0.006) (0.006) (0.007) (0.007)
FirmAge −0.021*** −0.021*** −0.022*** −0.020***
(0.005) (0.005) (0.006) (0.005)
AbsREM 0.061*** 0.061*** 0.069*** 0.069***
(0.005) (0.005) (0.007) (0.006)
Big4 −0.001 −0.001 −0.003 −0.002
(0.004) (0.004) (0.005) (0.004)
HighLitigationRisk 0.003 0.003 0.003 0.001
(0.002) (0.002) (0.003) (0.003)
InsHoldings 0.004 0.003 −0.001 −0.004
(0.004) (0.004) (0.004) (0.004)
CEOduality −0.001 −0.001 −0.002 −0.002
(0.002) (0.002) (0.002) (0.002)
BoardSize −0.007 −0.007 −0.009* −0.003
(0.004) (0.004) (0.005) (0.005)
BoardIndependence 0.001 0.002 0.005 0.002
(0.005) (0.005) (0.006) (0.006)
Constant 0.111*** 0.112*** 0.119*** 0.100***
(0.019) (0.018) (0.022) (0.021)
Observations 14,062a 13,892 14,062 13,892
Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.276 0.277 0.309 0.315
a
The reason why the observations in the regression are different from those in the original sample is that 94 singleton
observations were deleted from the regression (Stata code: reghdfe).
Notes: The results of the relationship between overconfident CFOs and earnings management are shown in this table. The
dependent variable is AbsAEM. The standard errors clustering at the firm level are displayed in parentheses. The sig-
nificance levels of 10%, 5%, and 1% are indicated by *, **, and ***, respectively. See Appendix A for variable definitions.
Variables of interest are marked in bold.

entropy-balancing sample. The coefficients of Holder67CFO and Holder100CFO are stat-

istically and economically positive, indicating that our findings are robust.11
Turning to the result of the control variables, after considering Holder67CFO (Hol-
der100CFO), the coefficient on Holder67CEO (Holder100CEO) becomes insignificant or

11
Since the entropy balance method is more restrictive, we only show the results of the entropy balance method in the
following analyses.
12 L. QIAO ET AL.

marginally significant and has a smaller magnitude than Holder67CFO (Holder100CFO). These
findings suggest that CFO overconfidence outweighs CEO overconfidence in earnings manage-
ment, consistent with prior studies’ findings (Baker et al., 2019; Jiang et al., 2010). The results of
other control variables are generally consistent with prior studies. For example, MaleCFO has a
statistically significant positive coefficient in column (4). A male CFO will have a 10.2% (=0.005/
0.049 × 100%) increase in AbsAEM at the mean, consistent with the findings of Barua et al.
(2010). FirmAge has a significantly negative coefficient in columns (1) to (4), implying that
firms with a long history use fewer earnings management (Gul et al., 2009).

5. Underpinning mechanisms
5.1 CFO overconfidence, earnings management, and firm performance volatility
(H2)
We conjecture that overconfident CFOs aim to reduce earnings volatility by manipulating
earnings. This section conducts subsample analyses to explore this prediction further. If
overconfident CFOs tend to smooth earnings, this incentive should be stronger when
firms have higher performance volatility. Empirically, we use earnings volatility (VolEarn-
ings) and cash flow volatility (VolCFO) to capture firm performance volatility following pre-
vious studies (Jayaraman, 2008).12 The higher value of VolEarnings and VolCFO means
more volatile performance. We separate the sample based on the median value of VolEarn-
ings (VolCFO) into two subsamples, high and low VolEarnings (VolCFO). As shown in
Panel A of Table 4, we find that the coefficients on Holder67CFO and Holder100CFO are
significant, with higher economic magnitude, only in firms with high VolEarnings. We
find similar results (Panel B of Table 4) when we split the sample based on the alternative
proxy of performance volatility, VolCFO. These findings are consistent with our expectation
that the positive relationship between overconfident CFOs and AEM is more pronounced in
firms with more volatile performance, supporting hypothesis 2 (H2).

5.2 CFO overconfidence, earnings management, and external financing needs (H3)
Overconfident CFOs deem that their firms are undervalued by the capital market, result-
ing in costly external funding (Ben-David et al., 2013; Malmendier et al., 2022). We
predict that if overconfident CFOs have to use external funds, they tend to use more
earnings management to provide the stable earnings that creditors prefer.
Empirically, first, we separate the sample into two subsamples based on the dummy
variable, external financing needs (EFN),13 and test whether overconfident CFOs are
more likely to use earnings management when firms need external funding. As shown
in Panel A of Table 5, the coefficients on Holder67CFO and Holder100CFO are only sig-
nificant in columns (1) and (3) when firms need external financing. Next, we conduct two
subsample analyses based on issuing debt (DebtIssuer) and issuing equity (EquityIssuer)
to further investigate this.14 In the first test, we find that Holder67CFO and Hol-
der100CFO only have a significantly positive and larger coefficient in columns (1) and
12
See Appendix A for detailed VolEarnings and VolCFO measures.
13
See Appendix A for detailed EFN measure.
14
See Appendix A for detailed DebtIssuer and EquityIssuer measures.
 ACCOUNTING FORUM 13

Table 4. CFO overconfidence, earnings management, and firm performance volatility.

Panel A: Subsample – earnings volatility
(1) (2) (3) (4)
Group variable: VolEarnings High Low High Low
Holder67CFO 0.006* −0.001
(0.003) (0.002)
Holder100CFO 0.008** 0.001
(0.003) (0.002)
Constant 0.083 0.065** 0.067 0.080***
(0.058) (0.030) (0.058) (0.030)
Controls Yes Yes Yes Yes

Observations 4,788 4,735 4,717 4,656

Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.252 0.349 0.265 0.345
Panel B: Subsample – cash flow volatility
(1) (2) (3) (4)
Group variable: VolCFO High Low High Low
Holder67CFO 0.005* −0.000
(0.003) (0.002)
Holder100CFO 0.007** 0.001
(0.003) (0.002)
Constant 0.038 0.138*** 0.060 0.132***
(0.050) (0.035) (0.048) (0.033)
Controls Yes Yes Yes Yes

Observations 4,799 4,694 4,735 4,606

Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.257 0.298 0.263 0.310
Notes: The results of the subsample analyses based on earnings volatility and cash flow volatility are shown in this table. If
the value of earnings volatility (cash flow volatility) is higher than the sample median, it belongs to the high earnings
volatility (cash flow volatility) group; otherwise, it belongs to the low earnings volatility (cash flow volatility) group. The
dependent variable is AbsAEM and controls are consistent with Equation (2). The standard errors clustering at the firm
level are displayed in parentheses. The significance levels of 10%, 5%, and 1% are indicated by *, **, and ***, respect-
ively. See Appendix A for variable definitions. Variables of interest are marked in bold.

(3) of Panel B (Table 5) when firms have debt issues. In the second test, we find that both
Holder67CFO and Holder100CFO have significantly positive coefficients in columns (1)
to (4) of Panel C (Table 5), but the coefficients between the two groups, equity issuers and
non-equity issuers, are insignificant, indicating that overconfident CFOs’ earnings man-
agement decisions are not significantly affected by equity issue incentives. This finding
might be because the relationship between earnings smoothness and the cost of equity
is unclear (McInnis, 2010) and undermines overconfident CFOs’ incentive to manipulate
earnings when they issue equity.
Overall, we find that the positive relationship between overconfident CFOs and earn-
ings management is more pronounced when firms have external financing needs, par-
ticularly debt financing, supporting hypothesis 3 (H3).

6. Robustness tests
We conduct numerous tests to mitigate certain potentially endogenous problems. We use
the PSM-DID to mitigate the problem of reverse causation. We also change the measure-
ment of earnings management to further mitigate mismeasurement.
14 L. QIAO ET AL.

Table 5. CFO overconfidence, earnings management, and external financing needs.

Panel A: Subsample – external financing needs
(1) (2) (3) (4)
Group variable: EFN Yes No Yes No
Holder67CFO 0.007*** −0.000
(0.003) (0.002)
Holder100CFO 0.008** 0.004
(0.003) (0.002)
Constant 0.086* 0.125*** 0.092** 0.123***
(0.047) (0.033) (0.046) (0.034)
Controls Yes Yes Yes Yes
Observations 4,339 6,105 4,260 6,015
Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.386 0.243 0.379 0.251
Panel B: Subsample – debt issuer
(1) (2) (3) (4)
Group variable: DebtIssuer Yes No Yes No
Holder67CFO 0.005** 0.003
(0.002) (0.002)
Holder100CFO 0.006** 0.003
(0.003) (0.002)
Constant 0.155*** 0.099*** 0.147*** 0.095***
(0.032) (0.028) (0.033) (0.027)
Controls Yes Yes Yes Yes
Observations 4,930 8,197 4,859 8,090
Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.353 0.328 0.366 0.328
Panel C: Subsample – equity issuer
(1) (2) (3) (4)
Group variable: EquityIssuer Yes No Yes No
Holder67CFO 0.008** 0.003**
(0.004) (0.002)
Holder100CFO 0.007* 0.004***
(0.004) (0.002)
Constant 0.111** 0.140*** 0.091* 0.122***
(0.053) (0.028) (0.047) (0.027)
Controls Yes Yes Yes Yes
The coefficient difference on overconfidence CFO proxies (Yes-No)a p-value = 0.155 p-value = 0.404
Observations 4,255 9,015 4,208 8,889
Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.339 0.302 0.357 0.287
a
We only present the coefficient difference on overconfidence CFO when the coefficient on overconfidence CFO is sig-
nificant in two groups.
Notes: The results of the subsample analysis based on external financing needs, debt issuers, and equity issuers are shown
in this table. If the value of the group variable (external financing needs, debt issuers, and equity issuers) is equal to one,
it belongs to the “Yes” group; otherwise, it belongs to the “No” group. The dependent variable is AbsAEM and controls
are consistent with Equation (2). The standard errors clustering at the firm level are displayed in parentheses. The sig-
nificance levels of 10%, 5%, and 1% are indicated by *, **, and ***, respectively. See Appendix A for variable definitions.
Variables of interest are marked in bold.

6.1 PSM-DID
One may argue that our empirical findings have a reverse causation problem. That is,
firms engaged in more earnings management activities tend to hire overconfident
CFOs. To address this concern, we perform PSM-DID. Specifically, according to prior
research (e.g. Ahmed & Duellman, 2013; Lin et al., 2020), we use CFO turnover as the
exogenous shock. In line with Huang and Kisgen (2013) and Fu and Zhang (2019), we
 ACCOUNTING FORUM 15

first restrict our sample to the three years preceding and following a CFO turnover,
excluding the transition year. Second, we require the firm to have at least two years of
available data prior to the CFO turnover. Third, we require a new CFO to be in
charge for at least two years after the turnover year to ensure that the CFO has
enough time to influence firm choices. To exclude the impact of CEO turnover, we
exclude samples of simultaneous CEO and CFO turnover. We define Treat, an indicator
variable, as equal to one if a firm changes its CFO from a non-overconfident one to an
overconfident one, and it equals zero if the firm replaces a non-overconfident CFO
with another non-overconfident one.15 Post, an indicator variable, equals one in the
first three years after CFO turnover and equals zero in the last three years before CFO
turnover.16 To select a control group, we use kernel matching. We match the treatment
and control groups using the control variables in Equation (2). This procedure leaves
2,092 observations. The empirical model for DID estimation is as follows:
AbsAEMit = b0 + b1Treati × Postit + Controls + Firmfixedeffects
+ Yearfixedeffects + 1it, (3)
where, the interaction term (Treati × Postit) is our variable of interest. Given that we
include firm and year dummies, we exclude Treati and Postit to avoid collinearity. We
predict β1 to be positive, indicating that the new overconfident CFOs tend to do more
earnings management than the new CFOs who are non-overconfident. We include the
same control variables as in Equation (2). Detailed variable measurements are summar-
ised in Appendix A.
Panel A of Table 6 displays the results of PSM. As shown, p-values of all control vari-
ables are greater than 0.1 in the matched sample, indicating that these controls are not
significantly different in the matched sample between the treatment and control
groups. The results of DID are presented in panel B of Table 6. The interaction term
(Treati × Postit) has a significantly positive coefficient, implying that firms with overconfi-
dent CFOs engage in more earnings management, consistent with our main finding.

6.2 Alternative measures

To eliminate the endogeneity concern from measurement errors, we change the
measurement of earnings management. Some studies suggest a non-linear relationship
between firm growth (or firm performance or sale growth) and accrual (e.g. Collins
et al., 2017; Kothari et al., 2005). The modified Jones model, however, does not
address these non-linear relationships. To capture them, Collins et al. (2017) add quintile
dummies of firm performance and growth to the modified Jones model. Banker et al.
(2019) argue that the quintile dummy has discontinuous jumps at the quintile boundary,
resulting in bias inference. Banker et al. (2019), therefore, use a spline correction to
resolve the weakness of the quantile dummy method, as the spline would catch variance
both inside and outside of the quantile, improving the precision of the non-linear
15
For brevity in the PSM-DID test, our proxies for overconfident CFOs and CEOs are Holder67CFO and Holder67CEO,
respectively. We find similar results when we use Holder100CFO and Holder100CEO.
16
Consistent with Ge et al. (2020), the turnover year is excluded from the regression analysis in this study because both
departing and replacement CFOs may impact earnings management in the turnover year.
16 L. QIAO ET AL.

Table 6. PSM-DID.
Panel A: PSM
Variables Sample Type Mean %Reduct T-test
Treated Control %Bias |Bias| t p>t
(1) (2) (3) (4) (5) (6) (7)
Holder67CEO Unmatched 0.650 0.406 50.300 10.440 0.000
Matched 0.649 0.647 0.400 99.200 0.070 0.942
MaleCFO Unmatched 0.916 0.905 3.900 0.810 0.417
Matched 0.916 0.916 −0.200 94.600 −0.040 0.970
MaleCEO Unmatched 0.948 0.965 −8.300 −1.810 0.070
Matched 0.948 0.949 −0.800 90.100 −0.130 0.896
MTB Unmatched 2.602 2.678 −1.800 −0.350 0.724
Matched 2.610 2.586 0.600 69.500 0.100 0.918
CashFlow Unmatched 0.094 0.088 6.600 1.410 0.158
Matched 0.093 0.092 1.700 74.800 0.270 0.784
ROA Unmatched 0.027 0.018 7.500 1.590 0.111
Matched 0.027 0.025 1.500 79.800 0.260 0.796
FirmSize Unmatched 7.147 7.796 −40.400 −8.270 0.000
Matched 7.155 7.175 −1.200 97.000 −0.220 0.827
Leverage Unmatched 0.217 0.241 −12.800 −2.730 0.006
Matched 0.216 0.214 1.200 90.500 0.200 0.840
FirmAge Unmatched 3.155 3.343 −30.000 −6.190 0.000
Matched 3.155 3.157 −0.200 99.200 −0.040 0.968
AbsREM Unmatched 0.262 0.254 3.200 0.690 0.488
Matched 0.262 0.266 −1.300 58.300 −0.220 0.823
Big4 Unmatched 0.874 0.939 −22.400 −5.060 0.000
Matched 0.877 0.874 0.900 96.100 0.130 0.896
HighLitigationRisk Unmatched 0.089 0.111 −7.300 −1.500 0.134
Matched 0.090 0.091 −0.400 95.100 −0.070 0.948
InsHoldings Unmatched 0.521 0.514 1.700 0.360 0.723
Matched 0.521 0.521 −0.100 96.900 −0.010 0.993
CEOduality Unmatched 0.497 0.482 2.900 0.610 0.543
Matched 0.498 0.482 3.300 −12.000 0.560 0.575
BoardSize Unmatched 2.132 2.230 −40.800 −8.750 0.000
Matched 2.133 2.138 −2.100 94.900 −0.360 0.722
BoardIndependence Unmatched 0.830 0.850 −12.400 −2.610 0.009
Matched 0.830 0.832 −0.900 92.800 −0.150 0.881
Panel B: DID
(1)
Treati × Postit 0.014***
(0.005)
Constant 0.139**
(0.062)
Controls Yes
Observations 2,092
Firm fixed effects Yes
Year fixed effects Yes
Adj. R 2 0.349
Notes: Panel A presents the results for PSM. Panel B presents the results of DID regression based on the PSM-matched
sample. The dependent variable is AbsAEM, and controls are consistent with Equation (2). In the DID test, for brevity, our
proxies for overconfident CFOs and CEOs are Holder67CFO and Holder67CEO, respectively. We find similar results when
we use Holder100CFO and Holder100CEO. The standard errors clustering at the firm level are displayed in parentheses.
The significance levels of 10%, 5%, and 1% are indicated by *, **, and ***, respectively. See Appendix A for variable
definitions. Variables of interest are marked in bold.

relationship being captured. We use a spline-corrected accrual-based earnings manage-

ment measure (AbsAEM_Alter) provided by Banker et al. (2019).17 As shown in Table 7,
Holder67CFO and Holder100CFO have a positive coefficient that remains economically
and statistically significant.

17
See Online Appendix C.2 for detailed AbsAEM_Alter measure.
 ACCOUNTING FORUM 17

In addition to using accrual-based earnings management, managers might also use

real earnings management (Baker et al., 2019; Zang, 2012). Thus, we change our earnings
management proxy (dependent variable) to real earnings management (AbsREM).18
When testing AbsREM, we control for top management characteristics (Overconfident
CEOs proxies, MaleCFO, and MaleCEO), firm growth opportunities (MTB), firm charac-
teristics (FirmSize, Leverage, FirmAge, MarketShare), external monitoring (Big4, InsHold-
ings), corporate governance (CEOduality, BoardSize, BoardIndependence) and accrual-
based earnings management (AbsAEM_Alter), following previous studies (e.g. Hsieh
et al., 2014; Roychowdhury, 2006; Zang, 2012). As presented in Table 7, the coefficients
of Holder67CFO and Holder100CFO are positive and significant. Overall, this evidence
suggests that different measures of earnings management do not significantly alter our
inferences.

7. Additional tests
7.1 CFO overconfidence, earnings management, and earnings smoothness
To further consolidate the notion that overconfident CFOs use earnings management to
smooth earnings, we investigate the relationship between overconfident CFOs’ earnings
management and earnings smoothness directly. We split the sample into high and low-
earnings management based on the median value of AbsAEM. Then, we regress earnings
smoothness (proxied by Smooth1 and Smooth2) (Bouwman, 2014) on CFO overconfi-
dence and controls.19,20 We expect the relationship between CFO overconfidence and
earnings smoothness to be stronger when firms use a high level of earnings management.
The results are shown in Online Appendix B.1, Panels A and B. We find that earnings are
smoother when overconfident CFOs use earnings management, which is consistent with
our prediction.

7.2 Alternative mechanism

Although we find strong evidence that the incentive of overconfident CFOs using earn-
ings management is to smoothen earnings and increase firms’ financing abilities,
someone may argue that overconfident CFOs use earnings management to pursue
their own interests (i.e. higher compensation), such as from the agency perspective.
Overconfident CFOs have incentive-heavy compensation contracts (Humphery-Jenner
et al., 2016), which might give them more incentives to use earnings management to
gain higher compensation (Jiang et al., 2010; Watts & Zimmerman, 1978).
To investigate this argument, we use two different measures to gauge the level of CFO
compensation incentives: equity incentives (CFOEquityIncentive) (Bergstresser & Philip-
pon, 2006) and equity intensity compensation (CFO_PEBC) (Humphery-Jenner et al.,
2016).21 The higher values mean stronger compensation incentives. Then, we divide
our sample into two groups based on the median value of CFO compensation incentives
18
See Online Appendix C.1 for detailed AbsREM measure.
19
See Appendix A for detailed Smooth1 and Smooth2 measures. Lower values mean smoother earnings.
20
Controls are consistent with equation (2).
21
See Appendix A for detailed CFOEquityIncentive and CFO_PEBC measures.
18 L. QIAO ET AL.

Table 7. Alternative measures.

(1) (2) (3) (4)
Dependent variable AbsAEM_Alter AbsAEM_Alter AbsREM AbsREM
Holder67CFO 0.002* 0.028***
(0.001) (0.006)
Holder100CFO 0.003*** 0.027***
(0.001) (0.007)
Constant 0.078*** 0.076*** 0.671*** 0.702***
(0.015) (0.015) (0.125) (0.130)
Controls Yes Yes Yes Yes
Observations 13,591 13,426 13,591 13,426
Firm fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Adj. R 2 0.254 0.263 0.728 0.728
Notes: The results of the alternative measures of the main variables are shown in this table. When the dependent variable
is AbsAEM_Alter, controls are consistent with Equation (2). When the dependent variable is AbsREM, we control top
management characteristics (Overconfident CEOs proxies, MaleCFO, and MaleCEO), firm growth opportunities (MTB),
firm characteristics (FirmSize, Leverage, FirmAge, MarketShare), external monitoring (Big4, InsHoldings), and corporate
governance (CEOduality, BoardSize, BoardIndependence) and accrual-based earnings management (AbsAEM_Alter).
The standard errors clustering at the firm level are displayed in parentheses. The significance levels of 10%, 5%,
and 1% are indicated by *, **, and ***, respectively. See Appendix A for variable definitions. Variables of interest
are marked in bold.

proxies. The subsample results are presented in Online Appendix B.2. We find that the
positive relationship between CFO overconfidence and earnings management is signifi-
cant in both high- and low-compensation incentive groups, but the coefficients between
the two groups are insignificant. Thus, we discover that overconfident CFOs are not
motivated by high compensation to manage earnings.

8. Conclusion
Our study focuses on the link between CFO overconfidence and earnings management.
In the main, we find a significantly positive relationship between CFO overconfidence
and earnings management. We further test the channels through which overconfident
CFOs affect earnings management via a series of subsample analyses. This positive
relationship between CFO overconfidence and earnings management is profound
when firms have highly volatile performance and high external financing needs, particu-
larly when they issue debt. These findings suggest that overconfident CFOs engage in
earnings management to smoothen earnings and increase firm financing abilities. In
further analysis, we rule out one potential channel – overconfident CFOs use earnings
management due to compensation incentives. Our findings remain consistent after a
series of robustness tests, including using entropy balancing, PSM-DID estimation,
and changing the measurements of main variables.
Our study extends and contributes to the earnings management literature by adding a
new determinant, CFO overconfidence. We also find a bright side to managerial over-
confidence in that overconfident CFOs use earnings management due to financing con-
cerns rather than personal compensation incentives, thus enriching the overconfidence
literature. Our findings add more empirical evidence that CFOs’ cognitive bias affects
their decisions, thus supporting the upper echelons and overconfidence theories. In
addition, our research can assist financial reporting users in comprehending the signifi-
cance of CFO overconfidence in financial reporting decisions and explain why boards
 ACCOUNTING FORUM 19

continue to hire overconfident CFOs despite evidence demonstrating the negative con-
sequences of CFO overconfidence.

Acknowledgement
The authors would like to thank the editor and the anonymous reviews for their constructive feed-
back on earlier versions of the article.

Disclosure statement
No potential conflict of interest was reported by the author(s).

ORCID
Emmanuel Adegbite http://orcid.org/0000-0002-7370-2818

References
Ahmed, A. S., & Duellman, S. (2013). Managerial overconfidence and accounting conservatism.
Journal of Accounting Research, 51(1), 1–30. https://doi.org/10.1111/j.1475-679X.2012.00467.x
Ali, A., & Zhang, W. (2015). CEO tenure and earnings management. Journal of Accounting and
Economics, 59(1), 60–79. https://doi.org/10.1016/j.jacceco.2014.11.004
Alicke, M. D. (1985). Global self-evaluation as determined by the desirability and controllability of
trait adjectives. Journal of Personality and Social Psychology, 49(6), 1621. https://doi.org/10.
1037/0022-3514.49.6.1621
Arya, A., Glover, J. C., & Sunder, S. (2003). Are unmanaged earnings always better for share-
holders? Accounting Horizons, 17(s-1), 111–116. https://doi.org/10.2308/acch.2003.17.s-1.111
Badertscher, B. A. (2011). Overvaluation and the choice of alternative earnings management
mechanisms. The Accounting Review, 86(5), 1491–1518. https://doi.org/10.2308/accr-10092
Baker, T. A., Lopez, T. J., Reitenga, A. L., & Ruch, G. W. (2019). The influence of CEO and CFO
power on accruals and real earnings management. Review of Quantitative Finance and
Accounting, 52(1), 325–345. https://doi.org/10.1007/s11156-018-0711-z
Banker, R. D., Byzalov, D., Fang, S., & Jin, B. (2019). Operating asymmetries and non-linear spline
correction in discretionary accrual models. Review of Quantitative Finance and Accounting, 54,
1–48. https://doi.org/10.1007/s11156-019-00808-5
Barua, A., Davidson, L. F., Rama, D. V., & Thiruvadi, S. (2010). CFO gender and accruals quality.
Accounting Horizons, 24(1), 25–39. https://doi.org/10.2308/acch.2010.24.1.25
Bazerman, M. H., & Moore, D. A. (2013). Judgment in managerial decision making. Wiley.
Ben-David, I., Graham, J. R., & Harvey, C. R. (2013). Managerial miscalibration. Quarterly Journal
of Economics, 128(4), 1547–1584. https://doi.org/10.1093/qje/qjt023
Bergstresser, D., & Philippon, T. (2006). CEO incentives and earnings management. Journal of
Financial Economics, 80(3), 511–529. https://doi.org/10.1016/j.jfineco.2004.10.011
Bernardo, A. E., & Welch, I. (2001). On the evolution of overconfidence and entrepreneurs.
Journal of Economics & Management Strategy, 10(3), 301–330. https://doi.org/10.1162/
105864001316907964
Black, D. E., & Gallemore, J. (2013). Bank executive overconfidence and delayed expected loss rec-
ognition [Working paper]. Dartmouth and University of Chicago.
Boahen, E. O., & Mamatzakis, E. C. (2020). The impact of religion on classification shifting in the
presence of corporate governance and BIG 4 audit. Accounting Forum, 44(2), 103–131. https://
doi.org/10.1080/01559982.2019.1573404
Bouwman, C. H. (2014). Managerial optimism and earnings smoothing. Journal of Banking &
Finance, 41, 283–303. https://doi.org/10.1016/j.jbankfin.2013.12.019
20 L. QIAO ET AL.

Bradshaw, M. T., Richardson, S. A., & Sloan, R. G. (2006). The relation between corporate
financing activities, analysts’ forecasts and stock returns. Journal of Accounting and
Economics, 42(1-2), 53–85. https://doi.org/10.1016/j.jacceco.2006.03.004
Brogaard, J., Li, D., & Xia, Y. (2017). Stock liquidity and default risk. Journal of Financial
Economics, 124(3), 486–502. https://doi.org/10.1016/j.jfineco.2017.03.003
Campbell, T. C., Gallmeyer, M., Johnson, S. A., Rutherford, J., & Stanley, B. W. (2011). CEO opti-
mism and forced turnover. Journal of Financial Economics, 101(3), 695–712. https://doi.org/10.
1016/j.jfineco.2011.03.004
Capalbo, F., Frino, A., Lim, M. Y., Mollica, V., & Palumbo, R. (2018). The impact of CEO narcis-
sism on earnings management. Abacus, 54(2), 210–226. https://doi.org/10.1111/abac.12116
Chen, C., Kim, J.-B., & Yao, L. (2017). Earnings smoothing: does it exacerbate or constrain stock
price crash risk? Journal of Corporate Finance, 42, 36–54. https://doi.org/10.1016/j.jcorpfin.
2016.11.004
Chen, C. X., Gores, T., Nasev, J., & Wu, S. (2022). Cfo overconfidence and cost behavior. Journal of
Management Accounting Research, 34(2), 117–135. https://doi.org/10.2308/JMAR-18-055
Chen, Y.-R., Ho, K.-Y., & Yeh, C.-W. (2020). CEO overconfidence and corporate cash holdings.
Journal of Corporate Finance, 62, 101577. https://doi.org/10.1016/j.jcorpfin.2020.101577
Cohen, D. A., Dey, A., & Lys, T. Z. (2008). Real and accrual-based earnings management in the
pre- and post-Sarbanes-Oxley periods. The Accounting Review, 83(3), 757–787. https://doi.
org/10.2308/accr.2008.83.3.757
Collins, D., Masli, A., Reitenga, A. L., & Sanchez, J. M. (2009). Earnings restatements, the
Sarbanes-Oxley Act, and the disciplining of chief financial officers. Journal of Accounting,
Auditing & Finance, 24(1), 1–34. https://doi.org/10.1177/0148558X0902400103
Collins, D. W., Pungaliya, R. S., & Vijh, A. M. (2017). The effects of firm growth and model spe-
cification choices on tests of earnings management in quarterly settings. The Accounting Review,
92(2), 69–100. https://doi.org/10.2308/accr-51551
Davis, M. (2002). WorldCom to cost $64 billion. Australian Financial Review, Monday 19 August
2002.
DeBoskey, D., Luo, Y., & Zhou, L. (2019). CEO power, board oversight, and earnings announce-
ment tone. Review of Quantitative Finance and Accounting, 52(2), 657–680. https://doi.org/10.
1007/s11156-018-0721-x
Dechow, P. M., & Dichev, I. D. (2002). The quality of accruals and earnings: The role of accrual
estimation errors. The Accounting Review, 77(s-1), 35–59. https://doi.org/10.2308/accr.2002.77.
s-1.35
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The
Accounting Review, 70(2), 193–225.
DeFond, M. L., & Jiambalvo, J. (1994). Debt covenant violation and manipulation of accruals.
Journal of Accounting and Economics, 17(1-2), 145–176. https://doi.org/10.1016/0165-4101
(94)90008-6
Demerjian, P., Lewis-Western, M., & McVay, S. (2020). How does intentional earnings smoothing
vary with managerial ability? Journal of Accounting, Auditing & Finance, 35(2), 406–437.
https://doi.org/10.1177/0148558X17748405
Deshmukh, S., Goel, A. M., & Howe, K. M. (2013). CEO overconfidence and dividend policy.
Journal of Financial Intermediation, 22(3), 440–463. https://doi.org/10.1016/j.jfi.2013.02.003
Dichev, I. D., Graham, J. R., Harvey, C. R., & Rajgopal, S. (2013). Earnings quality: Evidence from
the field. Journal of Accounting and Economics, 56(2-3), 1–33. https://doi.org/10.1016/j.jacceco.
2013.05.004
Egan, M., & Xu, Y. W. (2020). The true and fair view: Exploring how managers, directors and audi-
tors engage in practice. Accounting Forum, 44(4), 398–420. https://doi.org/10.1080/01559982.
2020.1727177
El Diri, M., Lambrinoudakis, C., & Alhadab, M. (2020). Corporate governance and earnings man-
agement in concentrated markets. Journal of Business Research, 108, 291–306. https://doi.org/
10.1016/j.jbusres.2019.11.013
 ACCOUNTING FORUM 21

Francis, B., Hasan, I., & Li, L. (2016). Abnormal real operations, real earnings management, and
subsequent crashes in stock prices. Review of Quantitative Finance and Accounting, 46(2), 217–
260. https://doi.org/10.1007/s11156-014-0468-y
Francis, J., LaFond, R., Olsson, P. M., & Schipper, K. (2004). Costs of equity and earnings attri-
butes. The Accounting Review, 79(4), 967–1010. https://doi.org/10.2308/accr.2004.79.4.967
Fraser, I. A., Hatherly, D. J., & Henry, W. M. (2004). Illegal acts and the auditor. Accounting
Forum, 28(2), 99–118. https://doi.org/10.1016/j.accfor.2004.04.008
Friedman, H. L. (2014). Implications of power: When the CEO can pressure the CFO to bias
reports. Journal of Accounting and Economics, 58(1), 117–141. https://doi.org/10.1016/j.
jacceco.2014.06.004
Fu, X., & Zhang, Z. (2019). CFO cultural background and stock price crash risk. Journal of
International Financial Markets, Institutions and Money, 62, 74–93. https://doi.org/10.1016/j.
intfin.2019.05.001
Gao, H., Li, K., & Ma, Y. (2021). Stakeholder orientation and the cost of debt: Evidence from state-
level adoption of constituency statutes. Journal of Financial and Quantitative Analysis, 56(6),
1908–1944. https://doi.org/10.1017/S0022109020000605
Gassen, J., & Fülbier, R. U. (2015). Do creditors prefer smooth earnings? Evidence from European
private firms. Journal of International Accounting Research, 14(2), 151–180. https://doi.org/10.
2308/jiar-51130
Ge, L., Jamil, T., & Yu, J. (2020). Ceo overconfidence and the choice of debt issuance [Working
paper]. Monash University.
Ge, W., Matsumoto, D., & Zhang, J. L. (2011). Do CFOs have style? An empirical investigation of
the effect of individual CFOs on accounting practices. Contemporary Accounting Research, 28
(4), 1141–1179. https://doi.org/10.1111/j.1911-3846.2011.01097.x
Geiger, M. A., & North, D. S. (2006). Does hiring a new CFO change things? An investigation of
changes in discretionary accruals. The Accounting Review, 81(4), 781–809. https://doi.org/10.
2308/accr.2006.81.4.781
Goel, A. M., & Thakor, A. V. (2003). Why do firms smooth earnings? The Journal of Business, 76
(1), 151–192. https://doi.org/10.1086/344117
Goel, A. M., & Thakor, A. V. (2008). Overconfidence, CEO selection, and corporate governance.
Journal of Finance, 63(6), 2737–2784. https://doi.org/10.1111/j.1540-6261.2008.01412.x
Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate
financial reporting. Journal of Accounting and Economics, 40(1-3), 3–73. https://doi.org/10.
1016/j.jacceco.2005.01.002
Gujarati, D. N., Porter, D. C., & Gunasekar, S. (2012). Basic econometrics. Tata McGraw-Hill
Education.
Gul, F. A., Fung, S. Y. K., & Jaggi, B. (2009). Earnings quality: Some evidence on the role of auditor
tenure and auditors’ industry expertise. Journal of Accounting and Economics, 47(3), 265–287.
https://doi.org/10.1016/j.jacceco.2009.03.001
Gupta, V. K., Mortal, S., Chakrabarty, B., Guo, X., & Turban, D. B. (2020). CFO gender and
financial statement irregularities. Academy of Management Journal, 63(3), 802–831. https://
doi.org/10.5465/amj.2017.0713
Hackbarth, D. (2008). Managerial traits and capital structure decisions. Journal of Financial and
Quantitative Analysis, 43(4), 843–881. https://doi.org/10.1017/S002210900001437X
Ham, C., Lang, M., Seybert, N., & Wang, S. (2017). CFO narcissism and financial reporting quality.
Journal of Accounting Research, 55(5), 1089–1135. https://doi.org/10.1111/1475-679X.12176
Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organization as a reflection of its top
managers. Academy of Management Review, 9(2), 193–206. https://doi.org/10.2307/258434
Hirshleifer, D., Low, A., & Teoh, S. H. (2012). Are overconfident CEOs better innovators? The
Journal of Finance, 67(4), 1457–1498. https://doi.org/10.1111/j.1540-6261.2012.01753.x
Holthausen, R. W. (1990). Accounting method choice: Opportunistic behavior, efficient contract-
ing, and information perspectives. Journal of Accounting and Economics, 12(1-3), 207–218.
https://doi.org/10.1016/0165-4101(90)90047-8
22 L. QIAO ET AL.

Hribar, P., & Nichols, D. (2007). The use of unsigned earnings quality measures in tests of earnings
management. Journal of Accounting Research, 45(5), 1017–1053. https://doi.org/10.1111/j.1475-
679X.2007.00259.x
Hsieh, T. S., Bedard, J. C., & Johnstone, K. M. (2014). CEO overconfidence and earnings manage-
ment during shifting regulatory regimes. Journal of Business Finance & Accounting, 41(9-10),
1243-1268. https://doi.org/10.1111/jbfa.12089.
Hsieh, T.-S., Wang, Z., & Demirkan, S. (2018). Overconfidence and tax avoidance: The role of
CEO and CFO interaction. Journal of Accounting and Public Policy, 37(3), 241–253. https://
doi.org/10.1016/j.jaccpubpol.2018.04.004
Huang, J., & Kisgen, D. J. (2013). Gender and corporate finance: Are male executives overconfident
relative to female executives? Journal of Financial Economics, 108(3), 822–839. https://doi.org/
10.1016/j.jfineco.2012.12.005
Huang, R., Tan, K. J. K., & Faff, R. W. (2016). CEO overconfidence and corporate debt maturity.
Journal of Corporate Finance, 36, 93–110. https://doi.org/10.1016/j.jcorpfin.2015.10.009
Humphery-Jenner, M., Lisic, L. L., Nanda, V., & Silveri, S. D. (2016). Executive overconfidence and
compensation structure. Journal of Financial Economics, 119(3), 533–558. https://doi.org/10.
1016/j.jfineco.2016.01.022
Jayaraman, S. (2008). Earnings volatility, cash flow volatility, and informed trading. Journal of
Accounting Research, 46(4), 809–851. https://doi.org/10.1111/j.1475-679X.2008.00293.x
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs
and ownership structure. Journal of Financial Economics, 3(4), 305–360. https://doi.org/10.
1016/0304-405X(76)90026-X
Jiang, J. X., Petroni, K. R., & Wang, I. Y. (2010). CFOs and CEOs: Who have the most influence on
earnings management? Journal of Financial Economics, 96(3), 513–526. https://doi.org/10.1016/
j.jfineco.2010.02.007
Jiraporn, P., Miller, G. A., Yoon, S. S., & Kim, Y. S. (2008). Is earnings management opportunistic
or beneficial? An agency theory perspective. International Review of Financial Analysis, 17(3),
622–634. https://doi.org/10.1016/j.irfa.2006.10.005
Jung, B., Soderstrom, N., & Yang, Y. S. (2013). Earnings smoothing activities of firms to manage
credit ratings. Contemporary Accounting Research, 30(2), 645–676. https://doi.org/10.1111/j.
1911-3846.2012.01170.x
Kim, I., & Skinner, D. J. (2012). Measuring securities litigation risk. Journal of Accounting and
Economics, 53(1-2), 290–310. https://doi.org/10.1016/j.jacceco.2011.09.005
Kim, J. B., Wang, Z., & Zhang, L. (2016). CEO overconfidence and stock price crash risk.
Contemporary Accounting Research, 33(4), 1720–1749. https://doi.org/10.1111/1911-3846.
12217
Koh, P.-S. (2007). Institutional investor type, earnings management and benchmark beaters.
Journal of Accounting and Public Policy, 26(3), 267–299. https://doi.org/10.1016/j.jaccpubpol.
2006.10.001
Kothari, S. P., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary accrual
measures. Journal of Accounting and Economics, 39(1), 163–197. https://doi.org/10.1016/j.
jacceco.2004.11.002
Krishnan, G. V., Raman, K. K., Yang, K., & Yu, W. (2011). CFO/CEO-board social ties, Sarbanes-
Oxley, and earnings management. Accounting Horizons, 25(3), 537–557. https://doi.org/10.
2308/acch-50028
Lara, J. M. G., Osma, B. G., & Penalva, F. (2020). Conditional conservatism and the limits to earn-
ings management. Journal of Accounting and Public Policy, 39(4), 106738. https://doi.org/10.
1016/j.jaccpubpol.2020.106738
Li, S., & Richie, N. (2016). Income smoothing and the cost of debt. China Journal of Accounting
Research, 9(3), 175–190. https://doi.org/10.1016/j.cjar.2016.03.001
Lin, C.-Y., Chen, Y., Ho, P.-H., & Yen, J.-F. (2020). CEO overconfidence and bank loan contract-
ing. Journal of Corporate Finance, 64, 101637. https://doi.org/10.1016/j.jcorpfin.2020.101637
Magrath, L., & Weld, L. G. (2002). Abusive earnings management and early warning signs. The
CPA Journal, 72(8), 50–54.
 ACCOUNTING FORUM 23

Malmendier, U., Pezone, V., & Zheng, H. (2022). Managerial duties and managerial biases. (in
Press).
Malmendier, U., & Tate, G. (2005). CEO overconfidence and corporate investment. The Journal of
Finance, 60(6), 2661–2700. https://doi.org/10.1111/j.1540-6261.2005.00813.x
Malmendier, U., & Tate, G. (2015). Behavioral CEOs: The role of managerial overconfidence.
Journal of Economic Perspectives, 29(4), 37–60. https://doi.org/10.1257/jep.29.4.37
Malmendier, U., Tate, G., & Yan, J. (2011). Overconfidence and early-life experiences: The effect of
managerial traits on corporate financial policies. The Journal of Finance, 66(5), 1687–1733.
https://doi.org/10.1111/j.1540-6261.2011.01685.x
Man, C. K., & Wong, B. (2013). Corporate governance and earnings management: A survey of lit-
erature. Journal of Applied Business Research (JABR), 29(2), 391–418. https://doi.org/10.19030/
jabr.v29i2.7646
Matsumoto, D. A. (2002). Management’s incentives to avoid negative earnings surprises. The
Accounting Review, 77(3), 483–514. https://doi.org/10.2308/accr.2002.77.3.483
McInnis, J. (2010). Earnings smoothness, average returns, and implied cost of equity capital. The
Accounting Review, 85(1), 315–341. https://doi.org/10.2308/accr.2010.85.1.315
Merton, R. C. (1974). On the pricing of corporate debt: The risk structure of interest rates. The
Journal of Finance, 29(2), 449–470. https://doi.org/10.2307/2978814
Mian, S. (2001). On the choice and replacement of chief financial officers. Journal of Financial
Economics, 60(1), 143–175. https://doi.org/10.1016/S0304-405X(01)00042-3
Minton, B. A., & Schrand, C. (1999). The impact of cash flow volatility on discretionary investment
and the costs of debt and equity financing. Journal of Financial Economics, 54(3), 423–460.
https://doi.org/10.1016/S0304-405X(99)00042-2
Nguyen, T., & Soobaroyen, T. (2019). Earnings management by non-profit organisations:
Evidence from UK charities. Australian Accounting Review, 29(1), 124–142. https://doi.org/
10.1111/auar.12242
Peni, E., & Vähämaa, S. (2010). Female executives and earnings management. Managerial Finance,
36(4), 629–645. https://doi.org/10.1108/03074351011050343
Qiao, L., Adegbite, E., & Nguyen, T. H. (2022). Chief financial officer overconfidence and stock
price crash risk. International Review of Financial Analysis, 84, 102364. https://doi.org/10.
1016/j.irfa.2022.102364
Roychowdhury, S. (2006). Earnings management through real activities manipulation. Journal of
Accounting and Economics, 42(3), 335–370. https://doi.org/10.1016/j.jacceco.2006.01.002
Sauerwald, S., & Su, W. (2019). CEO overconfidence and CSR decoupling. Corporate Governance:
An International Review, 27(4), 283–300. https://doi.org/10.1111/corg.12279
Schrand, C. M., & Zechman, S. L. (2012). Executive overconfidence and the slippery slope to
financial misreporting. Journal of Accounting and Economics, 53(1-2), 311–329. https://doi.
org/10.1016/j.jacceco.2011.09.001
Sha, Y., Qiao, L., Li, S., & Bu, Z. (2021). Political freedom and earnings management. Journal of
International Financial Markets, Institutions and Money, 75, 101443. https://doi.org/10.1016/
j.intfin.2021.101443
Svenson, O. (1981). Are we all less risky and more skillful than our fellow drivers? Acta
Psychologica, 47(2), 143–148. https://doi.org/10.1016/0001-6918(81)90005-6
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: a social psychological perspective on
mental health. Psychological Bulletin, 103(2), 193. https://doi.org/10.1037/0033-2909.103.2.193
Trueman, B., & Titman, S. (1988). An explanation for accounting income smoothing. Journal of
Accounting Research, 127–139. https://doi.org/10.2307/2491184
Watts, R. L., & Zimmerman, J. L. (1978). Towards a positive theory of the determination of
accounting standards. The Accounting Review, 112–134.
Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and
Social Psychology, 39(5), 806–820. https://doi.org/10.1037/0022-3514.39.5.806
Zang, A. Y. (2012). Evidence on the trade-off between real activities manipulation and accrual-
based earnings management. The Accounting Review, 87(2), 675–703. https://doi.org/10.2308/
accr-10196
24 L. QIAO ET AL.

Appendix A Variable definitionsa

Variables Definition
AbsAEMit Modified Jones model (Dechow et al., 1995). See Appendix 3.2.1 for details.
Holder67CFOit An indicator variable that equals one if CFOs are reluctant to exercise vested options that are no
less than 67% in the money for the first time, and zero otherwise (Campbell et al., 2011). See
Section 3.2.2 for details.
Holder67CEOit An indicator variable that equals one if CEOs are reluctant to exercise vested options that are no
less than 67% in the money for the first time, and zero otherwise (Campbell et al., 2011). See
Section 3.2.2 for details.
Holder100CFOit An indicator variable that equals one if CFOs are reluctant to exercise vested options that are no
less than 100% in the money for the first time, and zero otherwise (Hirshleifer et al., 2012). See
Section 3.2.2 for details.
Holder100CEOit An indicator variable that equals one if CEOs are reluctant to exercise vested options that are no
less than 100% in the money for the first time, and zero otherwise (Hirshleifer et al., 2012). See
Section 3.2.2 for details.
MaleCFOit An indicator variable equals one if the CFO is a male, and zero otherwise. (Data source: ExecuComp
database)
MaleCEOit An indicator variable equals one if the CEO is a male, and zero otherwise. (Data source: ExecuComp
database)
MTBit Divide equity market value (PRCC_F × CSHO) by equity book value (SEQ) (Demerjian et al., 2020).
CashFlowit Cash flow (OANCF) divided by total assets (AT) (Hsieh et al., 2014).
ROAit Income Before Extraordinary Items (IB) divided by total assets (AT)
FirmSizeit The natural logarithm of total assets (AT) (Jiang et al., 2010).
Leverageit The sum of long-term debt (DLTT) and short-term debt (DLC) over total assets (AT) (Hsieh et al.,
2014).
FirmAgeit The natural logarithm of the number of years, starting with the first year that firm information was
available from the Compustat database (Capalbo et al., 2018). (Data source: Compustat
database)
AbsREMit The method of Roychowdhury (2006). See Online Appendix C.1 for details.
Big4it An indicator variable equals one if the audit firm (AU) belongs to PwC, EY, KPMG, or Deloitte, and
zero otherwise (Hsieh et al., 2014).
HighLitigationRiskit An indicator variable that equals one if firms’ litigation risk is in the top decile of the sample, and
zero otherwise (Gao et al., 2021). The coefficients from model (2) in Table 7 of Kim and Skinner
(2012) are used to calculate the litigation risk. (Data source: CCM and CRSP databases)
InsHoldingsit The percentage of ownership owned by institutional investors (INSTOWN_PERC) (Zang, 2012).
(Data source: Thomson-Reuters 13F database)
CEOdualityit An indicator variable equals one if the CEO is also the board chair, and zero otherwise (Krishnan
et al., 2011). (Data source: BoardEx database)
BoardSizeit The number of board members (Krishnan et al., 2011). (Data source: BoardEx database)
BoardIndependenceit The number of independent non-executive directors divided by the total number of directors (El
Diri et al., 2020). (Data source: BoardEx database)
VolEarningsit The standard deviation of earnings (IB) divided by total assets (AT) computed over the years t-4 to
t (Jayaraman, 2008).
VolCFOit The standard deviation of cash flow (OANCF) divided by total assets (AT) computed over the years
t-4 to t (Jayaraman, 2008).
EFNit-1 An indicator variable that equals one if firms have external financing need at the beginning of the
year, and zero otherwise (Malmendier et al., 2011). See Online Appendix C.3 for details.
DebtIssuerit ChangeDebt is net debt financing measured as the cash proceeds from the issuance of long-term
debt (DLTIS) less cash payments for long-term debt reductions (DLTR) less the net changes in
current debt (DLCCH) (Bradshaw et al., 2006). DebtIssuerit equals one if ChangeDebt is greater
than 0, and zero otherwise.
EquityIssuerit ChangeEquity is net equity financing measured as the proceeds from the sale of common and
preferred stock (SSTK) less cash payments for the purchase of common and preferred stock
(PRSTKC) less cash payments for dividends (DV) (Bradshaw et al., 2006). EquityIssuerit equals one
if ChangeEquity is greater than 0, and zero otherwise.
Treati An indicator variable, it equals one if a firm changes from a non-overconfident CFO to an
overconfident one, and it equals zero if a firm replaces a non-overconfident CFO with another
non-overconfident one.
Postit An indicator variable, equals one in the first three years after CFO turnover and equals zero in the
last three years before CFO turnover.

(Continued)
 ACCOUNTING FORUM 25

Continued.
Variables Definition
AbsAEM_Alterit The spline correction method (Banker et al., 2019). See Online Appendix C.2 for details.
MarketShareit The percentage of the firm’s total sales (SALE) to the total industry sales (Badertscher, 2011).
Smooth1it The standard deviation of residuals from regressions of change in net income (NI) divided by the
total asset (AT) (i.e. deltaNI/AT) on six control variables, including sales growth (percentage
annual growth), size (logarithm of the market value of equity), leverage (total liabilities divided
by total assets), debt issuance (percentage change in total liabilities), equity issuance
(percentage change in shares outstanding), and annual asset turnover (sales divided by total
assets) (Bouwman, 2014).
Smooth2it The standard deviation of residuals from regressions of change in net income (NI) divided by the
total asset (AT) (i.e. deltaNI/AT) on the above six control variables over the standard deviation of
residuals from regressions of change in cash flow (OANCF) divided by the total asset (AT) (i.e.
delta OANCF/AT) on the above six control variables (Bouwman, 2014).
CFOEquityIncentiveit ONEPCT/(ONEPCT + Salary + Bonus). The variable ONEPCT represents the dollar change in the
value of the CFOs’ stock and option holdings as a result of a 1% increase in the firm stock price
(Bergstresser & Philippon, 2006). (Data source: ExecuComp database)
CEOEquityIncentiveit ONEPCT/(ONEPCT + Salary + Bonus). The variable ONEPCT represents the dollar change in the
value of the CEOs’ stock and option holdings as a result of a 1% increase in the firm stock price
(Bergstresser & Philippon, 2006). (Data source: ExecuComp database)
CFO_PEBCit The percentage of annual CFO compensation comes from option grants and stocks. Annual option
awards (ExecuComp item OPTION_AWARDS_BLK_VALUE before FAS 123R (the year 2006) and
OPTION_AWARDS_FV after FAS 123R) plus the annual stock grants (STOCK_AWARDS_FV)
divided by total annual compensation (TDC1) (Humphery-Jenner et al., 2016) (Data source:
ExecuComp database)
CEO_PEBCit The percentage of annual CEO compensation comes from option grants and stocks. Annual option
awards (ExecuComp item OPTION_AWARDS_BLK_VALUE before FAS 123R (the year 2006) and
OPTION_AWARDS_FV after FAS 123R) plus the annual stock grants (STOCK_AWARDS_FV)
divided by total annual compensation (TDC1) (Humphery-Jenner et al., 2016)
a
Unless otherwise noted, the source of the data is CCM.