You are on page 1of 10

Academy of Management Journal

2004, Vol. 47, No. 2, 267275.

EARNINGS MANAGEMENT FOLLOWING DUALITY-CREATING


SUCCESSIONS: ETHNOSTATISTICS, IMPRESSION
MANAGEMENT, AND AGENCY THEORY
WALLACE N. DAVIDSON III
Southern Illinois University

PORNSIT JIRAPORN
Texas A&M International University

YOUNG SANG KIM


Quinnipiac University

CAROL NEMEC
Southern Oregon University

We explore impression and earnings management via reasoning and methods


grounded in ethnostatistics and agency theory. We hypothesized that earnings man-
agement occurs more frequently following duality-creating successions than otherwise
because CEO-chairs have greater control of the impressions created by their firms
financial reports and are operating under greater expectations of positive results.
Using a sample of 173 duality-creating succession announcements and 112 non-
duality-creating succession announcements, we obtained empirical results consistent
with these predictions.

We have all been told at one time or another to the true economic conditions of the firm (Neu,
put your best foot forward. In doing so, an indi- 1991). We propose that impression management
vidual attempts to manage the impression he or she can obscure economic results and cloud the ability
gives. Corporations may also attempt to manage the of all stakeholders to make financial decisions. We,
impressions they give, but if doing so leads to ex- therefore, extend agency theory by showing that
cessive earnings management, have corporations impression management in general and earnings
gone too far? We define earnings management as management in particular can be an agency prob-
the use of flexible accounting principles that allow lem. The agency problem occurs because investors
managers to influence reported earnings, thereby and other stakeholders may not be able to make
causing reported income to be larger or smaller optimal decisions concerning a company.
than it would otherwise be. The Enron, Arthur If earnings management is viewed as a corpora-
Anderson, and WorldCom scandals of the early tions attempt to manage its image and the impres-
2000s have highlighted the importance of impres- sions it gives, then it is important to understand
sion management and the fact that impression man- the processes that may drive the phenomenon.
agement may be abused. Recently, ten of Wall As measuring and reporting performance has
Streets most prominent investment banks signed a become increasingly important in our society,
settlement with security regulators after they had ethnostatisticsthe study of how statistics are pro-
been accused of excessive impression management duced and managedmay need to be included in
in overly optimistic reports on their clients. If a management thought (Corvellec, 1997). In relating
corporations impression management hurts share- impression management to an agency problem, we
holders, then it may be an agency cost. shed light on how numbers can be misused.
The theory that we propose is that impression Specifically, we examine the question of whether
management is related to agency problems. Agency more earnings management follows duality-
problems occur when managers do not operate a creating successions than non-duality-creating suc-
company for the shareholders best interests. When cessions. We hypothesize that an executive put into
a firm manages the impression it presents to the the dual position of CEO and chair has both greater
marketplace through earnings management, share- expectations for improved performance and greater
holders and other stakeholders must base financial power to control the impressions a firm makes as it
decisions on numbers that do not, perhaps, reflect releases accounting information. The increased ex-
267
268 Academy of Management Journal April

pectations and control may lead a dual executive to ered an agency cost. If duality exacerbates the po-
create the impression of improved performance tential for agency problems, then the presence of a
through earnings management. dual CEO-chair may increase the extent of earnings
A firms board of directors is often described as management.
the shareholders first line of defense against in- CEOs may have incentives to manage their com-
competent management (Weisbach, 1988). Yet if panys earnings. CEOs are generally evaluated and
the companys CEO is the board chair, the primary compensated on the basis of their firms financial
monitor of management is management itself. performance. Since it is not highly likely that CEOs
Shareholder groups, Congress, and the Securities can directly control stock market performance, they
and Exchange Commission (SEC) have addressed may seek to control accounting-based measures.
the duality issue (Brickley, Coles, & Jarrell, 1997) The research literature has documented instances
and have argued that duality may lead to subopti- in which companies have engaged in earnings man-
mal managerial performance. Since numerous agement in relation to stock issues (Teoh, Welch, &
studies have indicated that duality does not impact Wong, 1998), management buyouts (Wu, 1997), take-
overall financial performance (Baliga, Moyer, & overs (Easterwood, 1997), auditor changes (DeFond &
Rao, 1996; Brickley et al., 1997; Brickley, Coles, & Subramanyan, 1998), and auditor selection, (Becker,
Jarrell, 1997; Daily & Dalton, 1997), the controversy DeFond, Jiambalvo, & Subramanyam, 1998), among
it generates may be inappropriate. However, we other instances.
refocus the debate about duality away from its re-
lation to overall performance and direct it to a more
Duality, Impression Management, and Agency
situational context, impression management. We
Theory
therefore, extend agency theory by showing that the
agency problems associated with the dual gover- Managers may have personal goals that compete
nance structure may be situational and will not, with those of shareholders. Since managers have
therefore, show up in overall performance results. been empowered by shareholders to make deci-
Using a sample of the announcements of 173 sions, a conflict of interests has potential agency
duality-creating successions and 112 non-duality- costs. Earnings management may be a type of
creating successions from the years 198292, we agency cost if managers release financial reports
obtained results consistent with the idea that dual- that do not present an accurate economic picture of
ity creates situationally relevant agency problems. a firm and shareholders make nonoptimal invest-
We found that, following duality-creating succes- ment decisions as a result. Thus, earnings manage-
sions, income-increasing earnings management ment is related to agency theory because the former
occurs to a greater extent than it does in non- can create or exacerbate agency costs. It has been
duality-creating successions. argued the dual governance structure can lead to or
We also propose that when performance prior to increase agency problems. We therefore propose
a succession has been poor, a successor CEO will be that there may be a link between duality and earn-
under greater pressure to show improved perfor- ings management through their relations with
mance. It may be this poor prior performance that agency theory.
creates the need for the executive to manage earn- A new CEO may feel pressure to show improved
ings and thereby exacerbate agency problems. We performance and to convince the board and share-
present evidence consistent with this hypothesis. holders that their hiring decision is moving the
company in a good direction. Prior research has
documented that new CEOs manage earnings.
THEORETICAL PERSPECTIVE
LaSalle, Jones, and Jain (1993) reported that firms
The desire for performance has gained an in- experiencing CEO changes are more likely than
creasing importance in our society, and this desire other firms to report income in ways that show
may lead to creativity in financial reporting (Cor- either increases or decreases in relation to past
vellec, 1997). Ethnostatistics (Gephart, 1988) is the income. Furthermore, Pourciau (1993) reported
situational context in which we examine a poten- that executive changes led to earnings management
tial agency problem, duality. We argue that when a that increased reported earnings following execu-
company engages in impression management in tive succession. Neither of these studies attempted
regard to earnings, the accounting numbers it re- to address the conflict between ideas based respec-
ports may not reflect its economic condition. If tively on agency and organization theory that is
shareholders make nonoptimal decisions because inherent in the duality literature.
of the managed earnings, then impression manage- The debate on duality has focused on its poten-
ment through earnings management may be consid- tial to create agency problems. A strict interpreta-
2004 Davidson, Jiraporn, Kim, and Nemec 269

tion of agency theory and duality is that the dual mance but does seem to mitigate agency problems
governance structure can allow management to co- in certain situations (Hermalin & Weisbach, 2000).
opt board power (Jensen, 1993). Brickley, Coles, Duality may be one of many variables that influ-
and Jarrell (1997) took a different perspective. ence a firms financial condition, and it may have a
Their approach embodies ideas from both the trans- similar situational importance even if it does not
action cost and social theories of the firm as well as impact overall financial performance. Measures of
agency theory ideas. They maintained that duality the overall financial impact of duality could be
may have costs, as outlined under agency theory: confounded by the many forces that influence fi-
nonoptimal decisions made by a dual executive. nancial outcomes. This problem would be most
But they also saw benefits of duality in its preven- severe when researchers are addressing dualitys
tion of costs associated with splitting the positions impact on overall or long-run performance, and it
of chair and CEO; these include the costs of infor- may, therefore, explain the lack of impact in the
mation, shared authority, succession processes, studies noted above. The confounding problem
and two large salaries, as well as reputational costs. may not be as severe when a researcher is measur-
Because both forms of leadership have costs and ing the impact of duality in a specific situation.
benefits, it is not theoretically obvious which lead- When an agency situation arises, a dual leader may
ership structure is optimal. Under this view, not be as likely to support shareholders and may
agency theory costs and social theory/transaction pursue outcomes that support the leaders own in-
costs and benefits are not mutually exclusive. The terests and position in the firm. Examining the spe-
decision to use the dual or nondual leadership cific situations in which the agency costs of duality
structure is a matter of comparing costs to benefits occur should lead to a better understanding of the
in each situation. costs and benefits of the various leadership struc-
The empirical evidence in Brickley, Coles, and tures.
Jarrell (1997) suggests that the costs of separating Since board independence can be negatively in-
the CEO and chair positions may be larger than the fluenced under the dual leadership structure, du-
agency costs of duality. Other research has also ality may make earnings management more likely
documented that the agency costs of duality may in a CEO-chairs first year, for several reasons. Gen-
not outweigh its benefits (Dalton, Daily, Ellstrand, erally, in the first year of a new appointment, a
& Johnson, 1998). leader will want to accomplish considerable
change. New leaders may be given more authority
to make changes and to set the tone for their peri-
Overall Performance versus Situational
ods of leadership. The greater authority given the
Outcomes
dual leader may be a double-edged sword, how-
From prior research, it is difficult to conclude ever, in that expectations about fast and positive
that the costs of duality outweigh its benefits and effects of the change in leadership might also be
that duality is detrimental to stakeholders. How- high. Given these high expectations, the ability of a
ever, there is research suggesting that in certain new CEO-chair to report better earnings in the
contexts, duality may be inconsistent with an inde- initial year may be important. These expectations
pendent monitoring role for a board. For example, may not be as strong in nondual governance struc-
McWilliams and Sen (1997) found that stock mar- tures, since no single executive would have to
ket investors reacted more negatively to the adop- shoulder the blame for poor performance. Since the
tion of antitakeover amendments by companies income-increasing effects of earnings management
with dual governance, implying that they viewed are temporary and will be reversed later (Jones,
the amendments as more likely to entrench execu- 1991), a nondual CEO with sights set on the chair
tives under dual leadership. Similarly, Mallette position may not be willing to sacrifice longer-term
and Fowler (1992) found that dual leadership made performance for immediate increased income. Fur-
adoption of poison pills more likely. Cannella and thermore, income-increasing earnings management
Lubatkin (1993) found that the likelihood that a with separate leaders would require collusion or at
CEO will be fired is lower in firms with dual lead- least acquiescence by the two leaders, thus reduc-
ership structure, which may imply that duality en- ing its probability. On the other hand, managers
trenches managers. incentives and compensation are often based on
The idea that duality may not impact long-run their companies financial performance, so in gen-
firm performance but may be situationally impor- eral it may be in managements own interest to give
tant is analogous to views forwarded in a related the impression of better performance.
line of research on board composition. Board com- This combination of managements discretion
position has limited impact on overall firm perfor- over reported earnings, the effect of earnings on
270 Academy of Management Journal April

management compensation, and expectations for when they are incurred, regardless of when cash is
dual managers may lead to impression manage- received or disbursed. Accrual accounting gives
ment through earnings management. We propose: managers discretion in determining reported earn-
ings. Management has a great deal of control over
Hypothesis 1. Earnings management will be
the timing of actual expenditures and also has some
greater following succession announcements
ability to alter the timing of recognition of revenues
that put one individual into a dual leadership
and expensesfor example, by advancing recogni-
role than following succession announcements
tion of sales revenue through credit sales, or delay-
that do not.
ing recognition of losses by waiting to establish loss
We would also expect that the incidence of earn- reserves (Teoh, Welch, & Wong, 1998).
ings management after succession announcements There are several ways to measure earnings man-
will be greater when prior performance has been agement. We used the modified Jones (1991) model
poor. The intuitive relation between prior perfor- to detect earnings management. Although this ap-
mance and succession has led to a large number of proach may not measure sales-based earnings ma-
studies whose results suggest that poor perfor- nipulations, it has been shown to have the most
mance is related to CEO turnover and succession power in detecting earnings management (De-
(e.g., Chung, Lubatkin, Rogers, & Owers, 1987; Dal- chow, Sloan, & Sweeney, 1995: 193) and only the
ton & Kesner, 1985; Datta & Guthrie, 1994; David- Jones and modified Jones models appear to have
son, Nemec, Worrell, & Lin, 2002; Puffer & Wein- the potential to provide reliable estimates of discre-
trop, 1991). tionary accruals (Guay, Kothari, & Watts, 1996:
Since poor prior performance often leads to turn- 104). Our results are, of course, limited by model
over, the successor CEO of a poorly performing firm selection, and we must recognize that no model can
will likely be under pressure to show positive re- detect the most blatant forms of earnings manage-
sults. Affecting real economic change often takes ment, the outright fabrication of results. A discus-
time, yet a new CEO may need to exhibit positive sion of our approach is available from the authors
results immediately to be allowed to continue. If upon request.
positive change does not occur naturally, the new Our approach divides current accruals into
CEO may feel pressure and have incentives to cre- discretionary and nondiscretionary parts. Accruals
ate the appearance of improved financial perfor- include both accrued assets and accrued liabilities,
mance through impression (earnings) management. which are respectively, amounts to be received and
When prior performance has not been poor, there amounts to be paid in the future. To identify non-
will be less pressure to demonstrate immediate discretionary current accruals, we regressed the
results. current accruals of firms that were not in our sam-
The poor prior performance may be the genesis of ple on their changes in sales. These firms were
the impression management that follows succes- matched to sample firms by two-digit SIC code and
sions and may, therefore, be the genesis of the had data on COMPUSTAT. (Using four-digit SIC
resulting agency problems. Without poor prior per- codes produced qualitatively similar but slightly
formance, there would be less need to improve weaker results.) To reduce heteroskedasticity, we
reported performance. Thus, poor performance deflated all variables by the lagged value of total
may be the causal link between impression man- assets. The resulting regression parameters were
agement and agency problems. We therefore applied to the change in sales less the change in
propose: accounts receivable of the sample firms to predict
the level of nondiscretionary current accruals that
Hypothesis 2. Earnings management will be
would be normal in each industry. The difference
greater following succession announcements
between the predicted nondiscretionary current
preceded by poor prior performance than it
accruals and the actual reported accruals is called
will be following succession announcements
discretionary current accruals (DCA). When its
not preceded by poor performance.
value is significantly different from zero, the
DCA is considered to be evidence of earnings
METHODS management.
Definition and Measurement of Earnings
Management Sample
Under generally accepted accounting principles, Our sample was composed of corporations and
companies use accrual accounting, whereby in- their CEOs included in Business Weeks 1992 re-
come is recognized when it is earned and expenses, port, The Corporate EliteThe Chief Executives
2004 Davidson, Jiraporn, Kim, and Nemec 271

of the 1000 Most Valuable Publicly Held U.S. Com- lessen earnings management. We obtained informa-
panies, covering a large cross-section of indus- tion about the boards from the proxy statements
tries. This report is a compilation of large firms in that immediately preceded the successions. We
the public eye, whose shareholders, both institu- classified an insider as a person who was employed
tional and noninstitutional, would question or cen- by a firm as well as sitting on its board. We classi-
sure a firm in the marketplace if executive duality fied as an affiliated outsider a director not directly
were a valid concern. employed by a firm yet having very close or strong
Starting in 1992, and going back to 1982, we ties to the firm or to its top executive officers. This
searched the financial press to find succession an- category includes, but is not limited to, outside
nouncements for the executives in our sample. We legal counsel, retired executives, other members of
categorized each announcement as indicating ei- the founding or controlling family of a focal firm,
ther duality or nonduality. We obtained complete directors employed by other firms that hold mate-
information on 285 succession announcements. rial voting shares of the firms securities, and em-
This included 173 duality announcements and 112 ployees of material capital providers. Finally, we
nonduality (separate CEO and chair) announce- classified directors who did not fall into one of the
ments. Brickley, Coles, and Jarrell (1997) argued two previous categories as independent outsiders.
that the dual leadership structure is the most com- Other measures of board independence were the
mon organizational form. For our sample of 285 proportional voting power of insiders and board
successions, we expected there to be 142.5 of each members, and board member tenure. Greater voting
type of governance structure if they occurred power through stock ownership provides motiva-
equally often. We compared actual to expected fre- tion for a board to act on behalf of shareholders.
quency, and the resulting chi-square was 13.06 (sig- Board tenure may also reflect board power; it was
nificant at .01). The observed frequency of duality measured as the average number of years an inde-
was thus greater than expected by chance. pendent outside director has served on the board.
We also classified each succession as an inside or
outside one, drawing on the time elapsed between
Performance Data
an individuals initial hiring by a firm and the date
To determine prior firm performance, we com- of appointment as CEO. An executive not originally
puted cumulative abnormal returns (the summed employed by the firm receiving one or more posi-
daily abnormal returns) over a 200-trading-day pe- tions at the same time as his initial hire is clearly an
riod extending from 350 days prior to an announce- outsider, just as an executive who has been with
ment date to 150 days prior to the announcement the firm for a long period of years is clearly an
date as measures of prior firm performance. We insider. An executive who had been with a firm for
used the standard event study methodology and the two years or less before the CEO appointment was
market model whose parameters come from the classified as an outsider; the two-year span was
period 600 to 351 days, as described in Fama, considered a time of grooming for the appoint-
Fisher, Jensen, and Roll (1969). This procedure ad- ment to the leadership.
justs stock returns for overall market movement
and individual firm risk. If this variable is positive
RESULTS
(negative) it means that the firms stock has per-
formed well (poorly) relative to the stock market. Table 1 shows the correlation coefficients for our
We used a stock market measure of performance variables. There is considerable correlation; hence,
because use of an accounting measure would have potential multicollinearity had to be considered.
biased our attempt to determine if there has been We computed variance inflation factors, which in-
accounting manipulation. dicated that multicollinearity was not a problem for
the variables in the regressions whose results ap-
pear in Table 2. As shown in Table 1, the discre-
Controls
tionary current accruals are, on average, negative.
We controlled for board independence, defined Large positive (negative) discretionary current ac-
in terms of the representation of firm insiders (em- cruals would imply income-increasing (-decreas-
ployees) and outsiders, because an independent ing) earnings management.
board (one with a greater proportion of unaffiliated Table 2 presents the regression results. Although
outsiders) may offset some of the agency problems the amount of variance explained (the adjusted R2s)
created in a dual governance structure. Empirical seems low, we do not think this is particularly
findings in Xie, Davidson, and DaDalt (2003) dem- problematic. Lev (1989) found that a majority of
onstrated that independent and active boards earnings studies have exhibited low values on this
272 Academy of Management Journal April

TABLE 1
Correlations and Descriptive Statistics for a Sample of 285 Successionsa

Variable Mean s.d. 1 2 3 4 5 6 7

1. Discretionary current accrualsb 0.59% 0.16%


2. CEO/chair dualityc 0.63 0.48 .12*
3. Prior firm performance 0.80% 0.21% .12* .05
4. Outside successionc 0.13 0.34 .01 .07 .08
5. Proportion of outsiders on board 66.72% 16.34% .00 .08 .05 .09
6. Outside director tenure 8.13 3.93 .03 .11* .00 .10* .06
7. Blockholder ownership 11.02% 17.88% .00 .00 .01 .03 .16*** .12*
8. Insider and board ownership 6.80% 12.05% .02 .10 .02 .00 .34*** .12* .10*

a
The successions include 173 dual (CEO and chair) and 112 nondual (CEO-only) successions.
For the duality subsample, the discretionary current accruals average 1.14 percent; for the nonduality subsample, they average 3.12
b

percent. This difference is significant at the .05 level (t 1.99). For firms with prior performance above (below) the median, the
discretionary current accruals average 2.22 percent (1.07%). This difference is significant at .05 (t 1.98).
c
Dichotomous (0, 1) variables where 1 represents outside succession or duality.

p .10
* p .05
*** p .001

statistic. The amounts of variance explained (R2s) ory. Impression management (through earnings
are comparable to those in other earnings studies. management) can be considered an agency cost if it
In model 1, we present results for the five control leads to nonoptimal decision making by sharehold-
variables. None of the estimated coefficients are ers. Our first hypothesis was that earnings manage-
statistically significant. ment will be greater following duality-creating suc-
For the firms with duality succession announce- cessions. The unambiguous dual-position leader
ments, average discretionary current accruals are may be operating under greater expectations, and
1.14 percent, a positive value. For the firms with the dual position may give this individual more
nonduality announcements, average discretionary authority to adjust financial reports. Given high
current accruals are negative, 3.12 percent. The expectations for good performance and the strong
difference is significant at the .05 level (t 2.05). authority provided by dual leadership, a CEO-chair
Model 2 of Table 2 shows the results of our test of may engage in impression (earnings) management
Hypothesis 1. We added a binary variable taking to show better performance. We argue that although
the value of 1 when a succession created a dual much of the large body of literature on duality does
position and 0 otherwise. The estimated coefficient not show it to be harmful to overall firm perfor-
for the duality variable is positive and significant at mance, under specific conditions duality may ex-
the 5 percent level. This finding is consistent with acerbate agency problems between managers and
the predictions of Hypothesis 1. shareholders.
For firms with prior performance below the me- We found some evidence consistent with the
dian, the discretionary current accruals average possibility that the agency costs of duality are sit-
1.07 percent, and for firms with prior performance uational. Specifically, we found that companies
above the median, the accruals average 2.22 per- whose succession announcements created a dual
cent. This difference is significant at the 5 percent leadership structure had greater measurable earn-
(t 1.98). In model 3 in Table 2, we included the ings management than those with successions cre-
variable measuring prior performance to test Hy- ating a nondual structure. If earnings management
pothesis 2. The estimated coefficient for the prior exacerbates agency problems, then our results are
firm performance variable is negative and signifi- consistent with the agency perspective on duality,
cant (at the 5 percent level), consistent with the since they show greater earnings management for
predictions of Hypothesis 2. Postsuccession earn- dual than for nondual CEO successions.
ings management is greater when there has been In our second hypothesis, we propose that in-
relatively poor prior firm performance. come-increasing earnings management will be
more likely to occur following poor performance.
Companies that have experienced poor perfor-
DISCUSSION
mance may have pressure to produce results. Thus,
The theoretical contribution of this paper is that poor prior performance may be the causal link be-
we link impression management and agency the- tween impression management and postsuccession
2004 Davidson, Jiraporn, Kim, and Nemec 273

TABLE 2 the impact of duality, but also into other outcomes


Results of Ordinary Least Square Regression for which agency-based explanations have been
a, b
Analysis made.
Our results suggest that the debate over the im-
Model Model Model
pact of duality on company performance is not
Variable 1 2 3
over. Since our results are not unequivocal, our
Constant 0.00 0.01 0.00 interpretation of them is somewhat provisional and
( 0.06) (0.21) (0.08) may require additional research. Future research
may be necessary to fully measure whether our
Outside successionc 0.00 0.01 0.00
( 0.12) (0.32) (0.11) findings are robust when other control variables are
Proportion of outsiders 0.00 0.00 0.00 added to analyses. In addition, the impact of dual-
on board ity in specific situations with agency-cost implica-
( 0.00) (0.15) ( 0.16) tions may deserve attention.
Outside director tenure 0.00 0.00 0.00
Financial reports are constructed by individuals
(0.58) (0.59) (0.53)
Blockholder ownership 0.00 0.00 0.00 who may be trying to create images of both their
(0.11) (0.21) (0.14) organizations and of themselves as leaders. This
Insider and board 0.00 0.00 0.00 creativity can become problematic (as in the cases
ownership of Enron/Arthur Anderson and WorldCom), and
( 0.35) ( 0.44) ( 0.42)
CEO/chair dualityc 0.04 Congress has even recognized its importance in its
( 2.02)* passage of the Sarbanes-Oxley Act. This act is de-
Prior firm performance 0.00 signed to help reduce earnings management by re-
(2.29)* quiring board and audit committee members to
have financial literacy.
Adjusted R2 0.2% 1.7% 1.6%
Our work is related to the field of ethnostatistics,
a
The dependent variable equals discretionary current accru- the study of how statistics are produced and used
als for a sample of large firms from 1982 through 1992. The (Gephart, 1988). In the case of reporting (and pos-
numbers in parentheses are t-statistics. A variance inflation sibly managing) earnings, it is the study of the
factor (VIF) of 10 or more would indicate serious multicollinear-
ity problems. All of the independent variables in the table have production and use of financial reports for inter-
VIFs of 1.3 or lower. preting firm performance. Financial reports may be
b
Because our data spanned several years, we ran alternative used in an attempt to transform corporate images.
regressions in which a potential year effect was considered. The Corvellec (1997) argued that when there is a ban on
results remained qualitatively similar.
c
Dichotomous (0, 1) variables where 1 indicated outside suc-
outright fabrication of performance results, there is
cession and duality. always room for creativity in recounting those re-
* p .05 sults. He stated, Creativity is an integral part of the
work of developing performance indicators, and we
impression management. We found evidence con- know that financial accounting can prove to be a
sistent with this hypothesis, suggesting that follow- very creative endeavor (Corvellec, 1997). Our
ing firms poor performance, successors may want study is an early step in documenting whether or-
to give the impression of improved performance. ganizational processes and structure relate to cre-
Our findings complement those of Lane, Can- atively reporting performance. It appears that poor
nella, and Lubatkin, who tested agency theory in prior performance and a succession creating a dual
another context. Those authors argued that con- CEO and chair may provide motive and opportu-
flicts of interest between shareholders and manag- nity for managing performance stories in financial
ers do arise in certain situations (1998: 574) but do reports.
not dominate day-to-day management of firms, as Given the relation of our work to ethnostatistics
agency theory would suggest. To measure the costs and our findings that managers may use financial
and benefits of a dual governance structure, future numbers to manage the impressions they give the
research may be needed to examine the impact of financial marketplace, future research into ethnos-
duality in specific situations. For example, Conyon tatistics and financial reporting may produce inter-
and Murphy (2000) found that pay-performance esting findings. Ethnostatistics has three levels that
sensitivity was greater in companies with dual may define an agenda for future research in this
leader governance structures and suggested that area (Gephart, 1988). Level 1 is the study of the
this situational perspective on duality could be ex- production of numbers. Researchers could focus on
tended to other situations. A situational perspec- motivations behind earnings management. Level 2
tive may provide important insights, not only into is the experimental simulation of reporting prac-
274 Academy of Management Journal April

tices. Researchers could study the impact of these 1998. Meta-analytic reviews of board composition,
managed earnings on the meaning assigned to the leadership structure, and financial performance. Stra-
numbers. Finally, level 3 of ethnostatistics con- tegic Management Journal, 19: 219 290.
cerns the rhetoric contained in the numbers. An Dalton, D. R. & Kesner, I. F. 1985. Organizational perfor-
interesting question would be whether rhetoric mance as an antecedent of inside/outside chief ex-
changes when there have been earnings increases ecutive replacement. Academy of Management
caused by earnings management rather than by im- Journal, 28: 749 762.
proved performance. Datta, D. K., & Guthrie, J. P. 1994. Executive succession:
As in all research, our sampling procedures and Organizational antecedents of CEO characteristics.
statistical methodologies affect our findings. Our Strategic Management Journal, 15: 569 577.
sample consisted of succession announcements in Davidson, W. N., Nemec, C., Worrell, D. L., & Lin, J. 2002.
large U.S. corporations. Future research may need Industrial origin of CEOs in outside succession:
to be directed at this issue with a sample of smaller Board preference and stockholder reaction. Journal
firms and international firms to determine if the of Management and Governance, 6: 295321.
results are generalizable. Further, our sample pe- DeChow, P., Sloan, R., & Sweeney, A. 1995. Detecting earn-
riod ended in 1992. To determine if our results are ings management. Accounting Review, 70: 193225.
generalizable to later time periods, we tested earn-
DeFond, M. L., & Subramanyam, K. R. 1998. Auditor
ings management on a sample of 81 successions in changes and discretionary accruals. Journal of Ac-
the years 199799. While the duality successions counting and Economics, 25: 35 67.
still had greater earnings management than the
Easterwood, C. 1997. Takeovers, and incentives for earn-
nondual, the difference was not as significant. The
ings management: An empirical analysis. Journal of
sensitivity of the results to time period may be due Applied Business Research, 14: 29 48.
to the increasing importance placed on corporate
governance since the mid-1990s. Fama, E. F., Fisher, L., Jensen, M., & Roll, R. 1969. The
adjustment of stock prices to new information. In-
ternational Economic Review, 10: 121.
Gephart, R. P. 1988. Ethnostatistics: Qualitative founda-
REFERENCES tions for quantitative research. Newbury Park, CA:
Baliga, B. R., Moyer, R. C., & Rao, R. S. 1996. CEO duality Sage.
and firm performance: Whats the fuss? Strategic Guay, W. R., Kothari, S. P., & Watts, R. 1996. A market
Management Journal, 17: 4153. based evaluation of discretionary accrual models.
Becker, C. L., DeFond, M. L., Jiambalvo, J., & Subra- Journal of Accounting Research, 34: 83105.
manyam, K. R. 1998. The effect of audit quality on Hermalin, B. E., & Weisbach, M. S. 2000. Boards of
earnings management. Contemporary Accounting directors as an endogenously determined institu-
Research, 15: 124. tion: A survey of the economic literature. Working
Brickley, J. A., Coles, J. L., & Jarrell, G. 1997. Leadership paper, University of California, Berkeley.
structure: Separating the CEO and chairman of the Jensen, M. C. 1993. The modern industrial revolution,
board. Journal of Corporate Finance, 3: 189 220. exit, and the failure of internal control systems. Jour-
Cannella, A. A., & Lubatkin, M. 1993. Succession as a nal of Finance, 48: 831 880.
sociopolitical process: Internal impediments to out- Jones, J. J. 1991. Earnings management during impact
sider selection. Academy of Management Journal, relief investigations. Journal of Accounting Re-
36: 763793. search, 29: 193228.
Chung, K. H., Lubatkin, R. C., Rogers, R. C., & Owers, J. E. Lane, P. J., Cannella, A. A., & Lubatkin, M. H. 1998.
1987. Do insiders make better CEOs than outsiders? Agency problems as antecedents to unrelated merg-
Academy of Management Executive, 3: 323329. ers and diversification: Ahimud and Lev reconsid-
Conyon, M. J. & Murphy, K. J. 2000. The prince and the ered. Strategic Management Journal, 19: 555578.
pauper? CEO pay in the United States and United LaSalle, R. E., Jones, S. K., & Jain, R. 1993. The associa-
Kingdom. Economics Journal, 110: 640 671. tion between executive succession and discretionary
Corvellec, H. 1997. Stories of achievements. New Bruns- accounting changes: Earnings management or differ-
wick, NJ: Transaction. ent perspectives? Journal of Business Finance &
Daily, C. M., & Dalton, D. R. 1997. CEO and board chair Accounting, 20: 653 671.
roles held jointly or separately: Much ado about Lev, B. 1989. On the usefulness of earnings and earnings
nothing. Academy of Management Executive, research. Journal of Accounting Research, 27: 153
11(3): 1120. 192.
Dalton, D. R., Daily, C. M., Ellstrand, A. E., & Johnson, J. L. Mallette, P., & Fowler, K. L. 1992. Effects of board com-
2004 Davidson, Jiraporn, Kim, and Nemec 275

position and stock ownership on the adoption of


poison pills. Academy of Management Journal,
35: 1010 1035.
Wallace N. Davidson III (davidson@cba.siu.edu) is the
McWilliams, V. B., & Sen, N. 1997. Bold monitoring and
Rehn Distinguished Research Professor of Finance at
anti takeover amendments. Journal of Financial
Southern Illinois University at Carbondale. He received
and Quantitative Analysis, 32: 491505.
his Ph.D. from The Ohio State University. His research
Neu, D. 1991. Trust, impression management and the interests include executive succession and corporate
public accounting profession. Critical Perspectives control.
on Accounting, 2: 295213.
Pornsit Jiraporn is an assistant professor of finance at
Pourciau, S. 1993. Earnings management and nonroutine Texas A&M International University in Laredo, Texas. He
executive changes. Journal of Accounting and Eco- received his Ph.D. in finance from Southern Illinois Uni-
nomics, 16: 317336. versity at Carbondale. His current research focuses on
Puffer, S. M., & Weintrop, J. B. 1991. Corporate performance corporate governance, earnings management, and corpo-
and CEO turnover: The role of performance expecta- rate restructurings.
tions. Administrative Science Quarterly, 36: 119.
Young Sang Kim is an assistant professor of finance at
Teoh, S. H., Welch, I., & Wong, T. J. 1998. Earnings man- Quinnipiac University. He received his Ph.D. in finance
agement and the underperformance of seasoned equity from Southern Illinois University at Carbondale. His cur-
offerings. Journal of Financial Economics, 50: 6399. rent research interests include agency theory, financial
Weisbach, M. S. 1988. Outside directors and CEO turn- management, corporate diversification, and risk manage-
over. Journal of Financial Economics, 20: 431 460. ment.

Wu, W. Y. 1997. Management buyouts and earnings man- Carol Nemec is an associate professor at the Southern
agement. Journal of Accounting, Auditing & Fi- Oregon University School of Business. She received her
nance, 12: 373389. DBA in finance from Southern Illinois University. Her
Xie, B., Davidson, W. N., & DaDalt, P. 2003. Earnings research interests include corporate governance and the
management and corporate governance: The role of roles of minorities and outsiders on boards of directors.
the board and the audit committee. Journal of Cor-
porate Finance, 9: 295316.

You might also like