Can External Monitoring Affect Corporate

Accounting Horizons American Accounting Association
Vol. 28, No. 3 DOI: 10.2308/acch-50771

2014
pp. 529–559
Can External Monitoring Affect Corporate

Financial Reporting and Disclosure?
Evidence from Earnings and
Expectations Management
Alfred Zhu Liu
SYNOPSIS: This study examines whether financial analysts and institutional investors
play a disciplinary role in monitoring corporate financial reporting and disclosure. Using
a sample of firms that meet or marginally beat analysts’ forecasts, likely through
upward earnings management and downward expectations management, this study
shows that managers’ use of the two tactics is associated with monitoring measures,
including analyst following, analyst experience and independence, institutional
ownership, and institutional investors’ experience and investment style. Managers
under more effective monitoring by analysts and institutional investors are more likely
to manage expectations downward than to manage earnings upward. Overall, the
findings are consistent with financial analysts and institutional investors playing a
monitoring role in constraining distortions in reported earnings and inducing timely
disclosure of bad news.
Keywords: external monitoring; earnings management; expectations management;

ﬁnancial analysts; institutional investors.
Data Availability: All data are from public sources.
Alfred Zhu Liu is an Assistant Professor at the University at Albany, SUNY.
I appreciate the valuable comments from two anonymous referees, Ming-Hui Chen, Amy Dunbar, Paul Griffin, Bengt
Holmstrom, Yinhua Li, Morton Pincus, Gim Seow, Siew Hong Teoh, David Weber, and Michael Willenborg. I also
thank workshop participants at the 2010 American Accounting Association Annual Meeting, McGill University,
University of California, Irvine, and University of Connecticut for comments on earlier versions. I thank the support of
the University of Connecticut and Thomson Reuters for providing the analyst data. All errors are my own.
Submitted: September 2012
Accepted: March 2014
Published Online: April 2014
Corresponding author: Alfred Zhu Liu
Email: zliu3@albany.edu
529
530 Liu
INTRODUCTION
T
his study investigates whether financial analysts and institutional investors play a
disciplinary role in monitoring corporate financial reporting and disclosure. Prior research
shows that managers engage in earnings and expectations management to meet or beat
analyst forecasts (Bartov, Givoly, and Hayn 2002; Matsumoto 2002; Burgstahler and Eames
2006).1 Recent studies report that determinants of managers’ use of the two tactics for meeting or
beating analyst forecasts (hereinafter, MBE) include the rigor of the financial reporting process,
regulations and enforcements, and balance sheet constraints on earnings management (Brown and
Pinello 2007; Koh, Matsumoto, and Rajgopal 2008; Das, Kim, and Patro 2011). The present study
adds to this line of research by exploring the potential monitoring role of financial analysts and
institutional investors.
Financial analysts and institutional investors arguably are motivated and able to play a
monitoring role in financial reporting. They demand high-quality earnings for valuation or
performance assessment and, thus, have incentives to encourage managers to minimize distortions
in reported earnings and improve disclosure quality (Dechow and Schrand 2004). In addition,
financial analysts are sophisticated financial information users and have access to a broad set of
resources, leading to a comparative advantage in monitoring managers’ reporting and disclosure
decisions (Jensen and Meckling 1976; Dyck, Morse, and Zingales 2010). They provide the market
with their own research and incremental information about firms’ fundamentals, which helps reveal
‘‘true’’ earnings (or distortions in reported earnings). From this perspective, their activities increase
the observability of the distortions in financial reporting and disclosure (Holmstrom 1979).2
There is scattered, but increasing, evidence on the monitoring role of analysts and institutional
investors. Analysts and institutional investors ask probing questions about abnormal performance
measures in conference calls (Yu 2008) and express their concerns about ‘‘less reliable’’ earnings in
public media (Donnelly 1990). Sometimes, analysts provide detailed analyses of components of
earnings to reveal the impact of managers’ accounting discretions, especially when reported
earnings differ from analysts’ own forecasts. Some institutions describe how they detect earnings
management or publicly challenge managers’ accounting practices.3 Not surprisingly, Dyck et al.
(2010) report that analysts and equity holders are the largest group of outsiders that bring (alleged)
corporate frauds to light.
This study incorporates the monitoring role of analysts and institutional investors into
managers’ decisions in managing earnings and expectations to meet or beat analysts’ expectations. I
consider the following factors to be related to their monitoring effect: (1) monitoring intensity, (2)
1
Earnings management in this study refers to accruals management. Managers can manipulate real economic
activities to manage GAAP earnings (Roychowdhury 2006; Gunny 2010). An important difference between
accruals or expectations management and real earnings management is that the former does not directly affect the
operating, investing, or financing activities of the MBE firms in that particular quarter. This paper focuses on
managers’ decisions in financial reporting and disclosure, i.e., the reporting and disclosure efforts in the
‘‘numbers game’’ addressed in Levitt (1998) and, thus, does not investigate the alternative of using real activities
to meet analysts’ forecasts. See Shih (2014) for discussions on managers’ trade-offs between real activities
management and accruals management. The findings of this paper are not qualitatively affected by whether
managers engage in real earnings management.
2
I am grateful to Bengt Holmstrom for helpful comments.
3
The equity research department of Credit Suisse Americas/United States, for example, publishes a special
quantitative analysis, ‘‘Earnings Quality: Practical Applications’’ (Patel, Barefoot, Yao, and Carlson 2007), to
help assess earnings quality and whether companies are managing earnings. Other research firms also publish
industry-specific rankings based on the quality of reported earnings on an annual basis (e.g., Solid Waste
Earnings Quality Analysis by The First Boston Corporation [1994], and EMS Earnings Quality Analysis by
Kaufman Bros., LP [2004]). Cases of analysts or institutional investors challenging earnings quality are often
available in the financial press, including companies like American Express, Pitney Bowes, Tyco International,
and Qwest, to name just a few (Pulliam 1999; Young 2001).
Accounting Horizons
September 2014
Can External Monitoring Affect Corporate Financial Reporting and Disclosure? 531
monitors’ expertise, and (3) monitors’ other incentives that may impair their monitoring role (e.g.,
analysts’ business affiliation and institutional investors’ investment strategy). I investigate whether
these factors are related to managers’ choices between earnings and expectations management; in
particular, whether monitoring by analysts and institutional investors curbs distortions in reported
earnings and compels managers to disclose bad news in a more timely way, hence shifting
managers’ preference toward expectations management over earnings management.
The sample consists of firms that succeeded in meeting or beating analysts’ forecasts over the
1988–2006 period, and likely did so through earnings and/or expectations management.4 Following
prior studies (Phillips, Pincus, and Rego 2003; Ayers, Jiang, and Yeung 2006; Burgstahler and
Eames 2006), I focus on firms that meet or just beat analysts’ quarterly earnings forecasts by a
small margin (less than or equal to two cents per share). The results show that managers are more
likely to engage in downward expectations management than upward earnings management when
(1) their firms are followed by more analysts, (2) analysts are more experienced, (3) independent
analysts follow their firms, (4) their firms are held by more experienced institutional investors, and
(5) the institutional investors are less short-term-oriented.5
A series of robustness tests show that the findings are not driven by analysts’ and institutional
investors’ selection of firms based on managers’ historical use of MBE tactics, MBE margin, measure
of expectations management, or managers’ use of real earnings management. Additional tests also
show that (1) analysts better understand and discount the effect of upward earnings management
when analyst following and analyst experience are high and when independent analysts follow the
firm, and (2) short-term-oriented institutional investors reduce their holdings when managers manage
expectations downward to achieve MBE. These findings provide further support for the impact of
analysts’ and institutional investors’ monitoring on managers’ use of MBE tactics.
This study contributes to the literature in several ways. First, it provides additional evidence
about analysts’ and institutional investors’ monitoring role in financial reporting and disclosure.
The results suggest that external monitoring by analysts and institutional investors helps reduce
distortions in reported earnings and, at the same time, induces managers to disclose bad news in a
more timely way. The results complement evidence in other studies indicating that analyst coverage
curbs earnings management (Yu 2008; Degeorge, Ding, Jeanjean, and Stolowy 2013), and shed
light on how analysts and institutional investors affect expectations management.6
Second, this study explores the determinants of analysts’ and institutional investors’ effectiveness
in monitoring managers’ financial reporting and disclosure. Prior research documents that analysts, on
average, fail to incorporate the predictable future earnings declines associated with high accruals in a
timely manner when forecasting earnings (Bradshaw, Richardson, and Sloan 2001), which is
mitigated by analysts’ experience (Drake and Myers 2011). This study shows that how well financial
analysts understand and discount the effect of accruals management is related to their monitoring
intensity, experience, and independence. The results also suggest that it is important to differentiate
institutional investors by their investment strategies when investigating their monitoring role.
Third, this study extends our understanding of managers’ decisions on financial reporting and
disclosure in the context of meeting or beating earnings forecasts. Doyle, Jennings, and Soliman
4
The sample period ends at 2006 because I/B/E/S stopped providing the analyst-broker identification file after that
year. Extending the sample period to include more recent years does not qualitatively affect the results.
5
I classify institutional investors into short-term-oriented and non-short-term-oriented, using Bushee’s (1998,
2001) classification. I thank Brian Bushee for providing his classification of institutional investors.
6
Degeorge et al. (2013) examine the association between analyst coverage and earnings management in an
international setting, while Yu (2008) focuses on the U.S. market. This paper considers both earnings
management and expectations management, and investigates how managers’ trade-off between the two is related
to external monitoring by analysts and institutional investors. This paper also considers whether monitors’
conflicting incentives impair their monitoring role.
Accounting Horizons
September 2014
532 Liu
(2013) suggest that managers likely choose the method least costly to implement when they can use
several tactics to meet or beat analysts’ expectations. This study provides additional evidence that
managers consider earnings management and expectations management as substitutes when there is
strong external monitoring. The findings complement those in prior studies indicating that managers
rely more on expectations management when the rigor of the financial reporting process, financial
regulations and enforcements, and net operating assets constrain earnings management (Brown and
Pinello 2007; Koh et al. 2008; Das et al. 2011). This study also helps explain the cost of downward
expectations management, especially in the presence of short-term-oriented institutional investors.
The remainder of the paper is organized as follows. The second section summarizes the related
literature and hypothesis development. The third section outlines the key variables and
sample-selection procedures. The fourth section provides the results of the empirical analyses. The
fifth section presents the results of additional analyses. The sixth section discusses the regulatory and
managerial implications of the findings, and the seventh section concludes the paper.
LITERATURE AND HYPOTHESES

The Monitoring Role of Financial Analysts and Institutional Investors
Although it has been long advocated that financial analysts and institutional investors serve as
external monitors (Jensen and Meckling 1976), the extant literature does not provide definitive
evidence on whether their monitoring improves the quality of reported earnings. On the one hand,
there is growing evidence supporting the monitoring role of analysts and institutional investors. For
example, Dyck et al. (2010) find that sell-side analysts expose more U.S. corporate frauds than
other outsiders, including auditors, media, the Securities and Exchange Commission (SEC),
industry regulators, and other regulatory agencies/organizations using lawsuits against corporate
frauds. Two recent studies show that earnings management is negatively related to analyst coverage
in the U.S. (Yu 2008) and in an international setting (Degeorge et al. 2013), consistent with analysts
helping to curb earnings management. Regarding institutional investors, Cheng, Huang, Li, and
Lobo (2010) show that institutional investors use securities litigation as an effective tool to
discipline managers and improve corporate governance. Ramalingegowda and Yu (2012) report
that conservative financial reporting is associated with higher ownership by institutions that are
likely to monitor managers.
On the other hand, prior research casts doubt on whether analysts and institutional investors have
the necessary expertise and incentive to act as monitors. For example, prior research shows that
analysts, in general, fail to incorporate the predictable future earnings declines associated with high
accruals in a timely manner when predicting future earnings (Bradshaw et al. 2001) or to fully unwind
managers’ opportunistic non-GAAP exclusions (Doyle et al. 2013), leading to questions about
analysts’ capability to monitor financial reporting and disclosure. Other studies show that analysts, as
well as institutional investors, have other incentives (e.g., investment banking/consulting fees and
trading incentives) that may impair their monitoring role (Bushee 1998, 2001; Gu, Li, and Yang
2013; Lin and McNichols 1998; O’Brien, McNichols, and Lin 2005). To provide additional evidence
on whether and how analysts and institutional investors shape managers’ financial reporting decisions,
this study examines how managers’ use of earnings and expectations management to meet or beat
analysts’ forecasts relates to characteristics of analysts and institutional investors.
Managers’ Trade-Off between Earnings and Expectations Management

Prior studies show that managers consider analysts’ earnings forecasts as a key performance
benchmark (Degeorge, Patel, and Zeckhauser 1999; Brown and Caylor 2005; Graham, Harvey, and
Rajgopal 2005) and engage in upward earnings management and downward expectations
Accounting Horizons
September 2014
management to meet or beat analysts’ forecasts (Bartov et al. 2002; Matsumoto 2002; Burgstahler
and Eames 2006).7 Assume that managers of firm i in quarter t observe a gap between unmanaged
earnings and analysts’ initial forecasts and decide to manage earnings and/or expectations to
achieve MBE. If the managers’ objective is to eliminate this gap at the lowest cost, whether and to
what extent managers engage in earnings and expectations management is affected by the relative
costs of the two tactics.8
The evidence in prior research that managers engage in earnings and expectations management
to achieve MBE and earn an MBE premium suggests that the market is not fully efficient in
unravelling the effects of the two tactics. Given the information asymmetry between managers and
outsiders, earnings management by an individual firm is, in general, difficult to detect, especially
when the magnitude of earnings management is small. In contrast, downward expectations
management, whether through public or private channels, eventually leads to observable decreases
in analysts’ forecasts. Recent studies show that downward expectations management is used as a
substitute for upward earnings management when the latter becomes more costly. That is, managers
use downward expectations management as a substitute for upward earnings management to avoid
earnings disappointments when the rigor of the financial reporting process curbs their ability to
manage earnings (Brown and Pinello 2007), when the regulatory environment makes earnings
management more costly (Koh et al. 2008), and when there are greater balance sheet constraints on
earnings management (Das et al. 2011). The present study extends this line of research by
examining the relation between managers’ use of the two MBE tactics and characteristics of outside
monitors, in particular, analysts and institutional investors. Ceteris paribus, if monitoring by
analysts and institutional investors unequally affects the costs of earnings and expectations
management, it would shift managers’ preference toward the one with relatively lower cost.
The key premise that monitoring by financial analysts and institutional investors may
unequally affect the costs of earnings and expectations management and, thus, managers’ trade-off
is consistent with the extant literature. First, monitoring by analysts and institutional investors
increases the risk of exposure and, thus, the expected cost of upward earnings management. Dyck et
al. (2010), for example, provide direct evidence that financial analysts and equity holders play a
significant role in revealing corporate fraud. Therefore, if monitoring by analysts and institutional
investors is effective, managers may face a higher marginal cost of managing earnings upward.
Second, while downward expectations management lowers outsiders’ earnings expectations, it
helps reduce information asymmetry and leads to more accurate, although downwardly biased,
analyst forecasts. Richardson, Teoh, and Wysocki (2004) and Hutton (2005), for example, show
improvement in forecast accuracy accompanying the walk-down of annual earnings expectations.
Houston, Lev, and Tucker (2010) find significant deterioration in the information environment (e.g.,
decreased analyst coverage and forecast quality) of companies that stop providing earnings
guidance. Therefore, analysts and institutional investors may view downward expectations
management as managers’ effort to timely disclose bad news and reduce information asymmetry,
which differs from the common view that earnings management deteriorates financial reporting
quality. The following discussions relate managers’ trade-off of the two tactics to measures of
effectiveness of monitoring by analysts and institutional investors and present the hypotheses.
7
Recent research also shows that managers engage in real activity manipulation (Roychowdhury 2006; Gunny
2010) and non-GAAP earnings exclusions (Doyle et al. 2013) to meet or beat analysts’ expectations.
8
Bartov et al. (2002) show that MBE premiums are slightly lower if there is earnings or expectations management.
Earnings and expectations management can affect MBE premiums differently, given that downward expectations
management is more observable than upward earnings management. Reductions in MBE premiums can be considered
as part of the costs of earnings and expectations management.
Accounting Horizons
September 2014
534 Liu
Anecdotal evidence in the financial press and analysts’ research reports suggests that financial
analysts are concerned about distortions in reported earnings. Analysts are aware of managers’
incentives to avoid negative earnings surprises, and at least some scrutinize managers’ accounting
discretion accordingly.9 Consistent with the notion that greater analyst following increases the
likelihood of detecting and, therefore, the cost of earnings management, Yu (2008) finds that U.S.
firms followed by more financial analysts manage earnings less, and Degeorge et al. (2013) report
that increased analyst coverage results in less earnings management in more financially developed
countries.10 Their findings suggest that the intensity of analysts’ monitoring on earnings
management increases with the number of analysts following a firm. In the meanwhile, prior
literature shows that firms attract more analysts if they disclose information more timely (Lang and
Lundholm 1996), and those withholding bad news experience decreases in analyst coverage
(Tucker 2010). Since analysts likely consider downward expectations management as timely
disclosure of bad news, which helps reduce information asymmetry and firms’ cost of capital,
having more analysts following a firm increases the cost of managing earnings upward relative to
the cost managing expectations downward. The first hypothesis is:
H1: Managers’ preference for upward earnings management over downward earnings
guidance is negatively related to the number of financial analysts following the firm.
A prerequisite of effective analyst monitoring is that analysts can detect irregularities in
reported earnings. Prior research suggests that analysts’ ability to understand earnings is related to
their experience. For example, more experienced financial analysts generate more accurate forecasts
and underreact less to prior earnings information (Mikhail, Walther, and Willis 1997, 2003). While
Bradshaw et al. (2001) show that financial analysts, on average, fail to fully unravel the effects of
accruals, Drake and Myers (2011) find that such an effect is mitigated by greater analyst
experience.11 Therefore, more experienced analysts are more likely to identify questionable
accounting practices, increasing the cost of upward earnings management. On the other hand, more
experienced analysts can better learn from past experience with downward expectations
management and avoid overreacting to the revealed ‘‘bad news.’’12 Together, if more experienced
analysts increase the cost of earnings management, but not the cost of downward expectations
management, managers of firms followed by more experienced analysts would prefer downward
expectations management over upward earnings management. The second hypothesis is:
guidance is negatively related to the experience of financial analysts following the firm.
9
Kaufman Bros., LP, for example, states in a published research report that ‘‘monitoring the earnings quality of
any public company should be an ongoing process in an effort to increase the chances of avoiding the next
company embroiled in an accounting scandal’’ (Kaufman Bros. Equity Research 2004, 2). Credit Suisse, in a
special research report on earnings management, points out that ‘‘companies manage earnings when they ask,
‘How can we best report the desired results?’ as opposed to, ‘How can we best report the actual results?’’’ (Patel
et al. 2007, 1).
10
Both studies use annual accounting data and focus on the magnitude of earnings management.
11
In addition, Brochet, Miller, and Srinivasan (2014) suggest that a longer history of coverage helps build a
professional relationship between managers and analysts, which creates benefits beyond analysts’ preferential
access to managers’ information and leads to more accurate earnings forecasts.
12
There is little evidence in prior literature on how analysts’ experience affects their reactions or market reactions
to downward expectations management. However, Mikhail et al. (2003) show that experienced analysts better
learn from their past forecast errors. It is likely that experienced analysts better learn from past downward
expectations management and discount the bad news revealed by managers in such cases. This is supported by an
untabulated analysis that shows smaller market reactions to downward expectations management for MBE firms
followed by more experienced analysts.
Accounting Horizons
September 2014
In addition, analysts should have sufficient incentives to provide effective monitoring and
assume the related costs, including managers’ retaliation. Prior research classifies financial analysts
into two groups, affiliated versus independent, and suggests that affiliated analysts who work for
underwriters of public firms’ security offerings may have conflicts of interest as monitors (Lin and
McNichols 1998; O’Brien et al. 2005; Feng and McVay 2010).13 In contrast, independent analysts
are less likely to have conflicts of interest and appear to help discipline affiliated analysts’ behavior
(Gu and Xue 2008). Hence, the presence of independent analysts improves the overall effectiveness
of analysts’ monitoring on earnings management. Meanwhile, downward expectations management
reveals additional information to independent analysts and may reduce their information collection
cost to a greater extent. Rana (2008) suggests that independent analysts are attracted to firms that
provide additional information in regular communications. If independent analysts are more likely
to scrutinize firms for earnings management than for expectations management, the presence of
independent analysts increases the relative cost of upward earnings management over downward
expectations management. The third hypothesis is:
guidance is negatively related to the presence of independent analysts.
Research on corporate governance also suggests that institutional investors can play an active
role in shaping corporate policies (Shleifer and Vishny 1986; Guercio and Hawkins 1999; Cadman
and Sunder 2014). With respect to managers’ financial reporting decisions, Bushee (1998) finds that
companies with high institutional ownership are less likely to cut research and development (R&D)
expenditures to boost earnings, suggesting that institutional investors may help reduce earnings
management. Burns, Kedia, and Lipson (2010), however, report that the likelihood and severity of
financial misreporting is positively related to total institutional ownership. Ajinkya, Bhojraj, and
Sengupta (2005) show that firms with greater institutional ownership are more likely to issue
management earnings forecasts. Matsumoto (2002) reports that overall institutional ownership is
positively associated with the likelihood of both upward earnings management and downward
expectations management achieving MBE.14 Given the mixed evidence of the impacts of
institutional investors’ monitoring activities, the fourth hypothesis is non-directional:
guidance is related to total institutional holdings in the firm.
Similar to analysts’ experience, institutional investors’ investment experience with a specific
firm can also affect the effectiveness of monitoring by institutional investors. Institutional investors
that have invested in a firm in prior periods better understand the firm’s business cycle and its
financial position. Therefore, more experienced institutional investors are more likely to detect
earnings management. At the same time, institutional investors with a longer investment history in a
firm are less likely to overreact to bad news disclosed in downward expectations management.
Consequently, institutional investors that have greater experience with a firm increase the relative
cost of upward earnings management over downward expectations management in meeting or
beating analysts’ forecasts. Hence, the fifth hypothesis is:
guidance is negatively related to institutional investors’ experience with the firm.
13
Groysberg, Healy, and Maber (2011) provide direct evidence that affiliated analysts’ compensations are affected
by their contributions to investment banking from equity underwriting activities.
14
Matsumoto (2002) does not examine whether the two tactics can be substitutes or how the trade-off is related to
institutional investors.
Accounting Horizons
September 2014
536 Liu
Finally, managers’ trade-off between earnings and expectations management may be related
to whether institutional investors are short-term-oriented. Managers who are concerned about
short-term-focused investors need to maintain the momentum of earnings growth, which cannot be
achieved by downward expectations management. Instead, the observable downward revisions of
analysts’ forecasts as a result of expectations management can trigger selling by institutional
investors that focus on short-term performance and engage in momentum trading. In addition,
institutional investors that have short-term investment horizons and momentum trading strategies
are less likely to act as monitors. Bushee (1998, 2001) shows that transient institutional ownership
induces managers to cut R&D expenditures to boost earnings, likely driven by such institutions’
over-weighing of short-term earnings in pricing stocks. Therefore, short-term-oriented institutional
investors may create incentives for managers to rely more on upward earnings management
instead of downward expectations management to meet or beat analysts’ forecasts. The last
hypothesis is:
guidance is positively related to institutional investors’ short-term-oriented investment
strategy.
RESEARCH DESIGN
Data
Prior studies suggest that firms meeting or just beating earnings expectations by a narrow
margin are probably playing the numbers game (Phillips et al. 2003; Ayers et al. 2006; Burgstahler
and Eames 2006). Following these studies, I focus on firms with non-negative earnings surprises of
no more than two cents over the period from 1988 to 2006. The sample period ends at 2006 because
I/B/E/S stops providing the analyst-broker identification file, which is needed to define analyst
independence, after that year.15 I exclude firms that appear to engage in neither upward earnings
management nor downward expectations management (i.e., firms with no positive abnormal accruals
or downward forecast revisions before the earnings release), because such firms do not rely on the
two tactics to achieve MBE and, thus, managers do not face the trade-off. The analysis then focuses
on firms that meet or just beat analysts’ forecasts, likely through earnings and/or expectations
management.
Analysts’ quarterly earnings forecasts and earnings surprises are retrieved from the split-
unadjusted I/B/E/S detail files. I exclude stock split quarters to avoid complications of combining
unadjusted forecast information and possible errors in measuring downward expectations
management.16 To alleviate the effects of low forecast frequency and concurrent forecasts on the
expectations management measure, I require at least two analyst forecasts that are at least 20 trading
days apart in a sample firm-quarter, and use the latest analyst forecasts as earnings expectations
(Bartov et al. 2002; Brown and Pinello 2007).17 To control for the impact of managers’ historical
use of the two MBE tactics on analysts and institutional investors, I require earnings and
expectations management in the prior quarter, as well. Other quarterly accounting information is
from Compustat. There are 7,389 firm-quarters from 1988 to 2006 that meet or beat analysts’
forecasts by no more than two cents and have all the information required by the analysis. The final
15
Extending the sample period to include more recent years does not qualitatively affect the results.
16
See ‘‘A Note on I/B/E/S Unadjusted Data’’ by Wharton Research Data Services for details (Robinson and
Glushkov 2006).
17
To reduce coding errors in the I/B/E/S data, I exclude firm-quarters with earnings announcement dates more than
120 days after the fiscal quarter ends, both reported by I/B/E/S.
Accounting Horizons
September 2014
sample includes 4,578 firm-quarters that likely meet or just beat analysts’ forecasts using at least
one of the two tactics, upward earnings management or downward expectations management.
Proxy for Managers’ Use of Earnings and Expectations Management

I use two binary variables as proxies for managers’ decisions to manage earnings upward or
expectations downward. The upward earnings management measure (UEM) is based on abnormal
accruals estimated from the modified Jones model. Following Hribar and Collins (2002), I compute
total accruals as earnings from continuing operations minus cash flows from operating activities
(Compustat quarterly item #76 #108 þ #78), scaled by lagged total assets (Compustat quarterly
item #44). I define abnormal accruals (DAC) for firm i in quarter t as the residuals from the
following regression:

Total Accrualsit 1 DSalesit DReceivablesit PPEit
¼ a1 þ a2 þ a3
Assetsi;t1 Assetsi;t1 Assetsi;t1 Assetsi;t1
X4
þ Qk þ ROAi;t ;
k¼1
where the quarterly dummies control for seasonality and other effects on accruals related to fiscal
quarters, including the impact of independent audit and annual expense recognition in the fourth
quarter (Barton and Simko 2002; Brown and Pinello 2007). ROA is included as an additional
explanatory variable to control for the effect of performance. I partition all firm-quarters in
Compustat into groups by two-digit SIC code and fiscal year and estimate the model for each group
with at least 12 firm-quarters.18 UEM equals 1 if DAC . 0, and 0 otherwise.
Following prior studies (Bartov et al. 2002; Richardson et al. 2004; Brown and Pinello 2007), I
use a binary variable, DEX, to measure downward expectations management by comparing the
latest analyst forecast error with the initial analyst forecast error.19 The underlying assumption of
this approach is that the outcome of managers’ downward expectations management is reflected in
the downward forecast revisions toward a beatable level. The latest forecast is the last individual
forecast of firm i’s earnings in quarter t at least three days before the earnings announcement date.
The initial forecast is the earliest individual forecast of firm i’s earnings in quarter t at least one day
after the announcement date of firm i’s earnings in quarter t1. DEX equals 1 if the initial forecast is
higher than the actual earnings, while the latest forecast is lower than or equal to the actual earnings.
Therefore, how managers use earnings and expectations management to achieve MBE in
quarter t can be summarized into three categories: Choicet ¼ 1: earnings management only (UEMt ¼
1 and DEXt ¼ 0); Choicet ¼ 2: expectations management only (UEMt ¼ 0 and DEXt ¼ 1); and
Choicet ¼ 3: both earnings and expectations management (UEMt ¼ 1 and DEXt ¼ 1). I then use a
discrete choice model to analyze the association between managers’ choice and monitoring by
analysts and institutional investors.
Monitoring Measures
Following the discussion in the previous section, the effectiveness of monitoring may be
related to monitoring intensity, monitors’ experience, and potentially conflicting incentives. I use
three variables to measure the effectiveness of monitoring by analysts. AFt is the number of
18
To mitigate the influence of outliers, I winsorize all variables used in the estimation at the top and bottom 1
percentile, following Brown and Pinello (2007).
19
Alternatively, I use managers’ public earnings guidance to measure expectations management and find
qualitatively similar results. See the ‘‘Sensitivity Analysis’’ section for details.
Accounting Horizons
September 2014
538 Liu
financial analysts issuing earnings forecasts for a sample firm in quarter t. AEXPt is the average
experience of all analysts following a sample firm in quarter t. An individual analyst’s experience is
measured by the number of quarters the analyst has issued quarterly earnings forecasts for a sample
firm (Mikhail et al. 1997). AIDPt is a binary variable indicating the presence of independent
analysts, equal to 1 if there is at least one independent analyst following a sample firm in quarter t,
and 0 otherwise. Independent analysts are defined as those who work for brokers that are not
engaged in equity issuances over the 1985–2006 period.
I use total institutional holdings (IO) to measure institutional investors’ monitoring intensity. IOt
is the percentage of a sample firm’s outstanding shares held by institutional investors in quarter t
reported in CDA 13(f ). I use the number of quarters in which an institutional investor has invested in
a firm to measure the institution’s experience.20 IEXPt is the average experience of all institutional
investors invested in a sample firm in quarter t. I refer to Bushee’s (1998) classification to measure to
what extent institutional investors are short-term-oriented. ISTt is the percentage of transient
institutional ownership divided by total institutional ownership for a sample firm in quarter t. Table 1
provides a complete list of variable definitions.
Assessing the effect of analysts’ and institutional investors’ monitoring on managers’ use of
MBE tactics can be challenging because of potential endogeneity caused by analysts’ or
institutional investors’ decision to follow or invest in firms with certain characteristics, which may
include the quality of financial reporting and disclosure. I implement the following procedures to
mitigate such concerns. First, I use analyst and institutional investor monitoring measures
constructed in the prior quarter, following Matsumoto (2002) and Degeorge et al. (2013). The
values of these measures are predetermined by the time managers need to decide on MBE strategies
in the current quarter. Second, similar to the residual analyst coverage measure in Yu (2008), I
regress the characteristic-based analyst and institutional investor monitoring variables on managers’
use of earnings and expectations management in the prior quarter, along with other control
variables. I use the residuals as monitoring variables in all tests. The rationale is that prior use of
earnings and expectations management may reveal managers’ style in MBE strategies, and analysts
and institutional investors may prefer firms with a certain style. The residuals by construction are
independent of managers’ use of earnings and expectations management in the prior quarter. Third,
I include managers’ use of earnings and expectations management in the prior quarter in the
explanatory variables to control for managers’ historical MBE strategies. By including managers’
decisions in both current and prior quarters, the monitoring variables are considered to ‘‘Granger-
cause’’ managers’ choices in the current quarter.
MAIN RESULTS
Descriptive Statistics
Table 2 summarizes managers’ use of earnings and expectations management in achieving
MBE. In Panel A, out of 4,578 firm-quarters that likely achieved MBE using at least one of the two
tactics, 1,590 (34.73 percent) are classified as only managing earnings upward, 1,825 (39.87
20
This measure captures the institutional investment experience with an investee company and is different from a
portfolio manager’s personal investment experience. It is possible that there are turnovers of portfolio managers
in institutions. Future research may further examine the effect of portfolio managers’ personal experience in this
setting, which is not addressed in the current study because portfolio managers’ personal identification
information is not available in the institutional holdings data (Form 13F).
Accounting Horizons
September 2014
TABLE 1
Definitions of Variables
Variable Definition
UEt Actual earnings minus the latest analyst forecast, both from the unadjusted I/B/E/S files,
for a firm in quarter t.
DACt Abnormal accruals estimated using the modified Jones model by year and industry for a
firm in quarter t.
UEMt A dummy variable equal to 1 if DAC is positive in quarter t, and 0 otherwise.
REVt The latest analyst forecast minus the first analyst forecast after the previous earnings
announcement date in quarter t, as defined in Bartov et al. (2002) and Brown and
Pinello (2007).
DEXt A dummy variable equal to 1 if REV is negative in quarter t, and 0 otherwise.
AFt1 The number of analysts forecasting firm i’s earnings in quarter t1. rAFt1 is the
residual of regressing AFt1 on UEMt1 and DEXt1 and other monitoring variables
and firm characteristics in quarter t1.
AEXPt1 The average experience of all analysts predicting firm i’s earnings in quarter t1.
Analyst experience is measured by the number of quarters an analyst has released a
quarterly earnings forecast for firm i as of quarter t1. rAEXPt1 is the residual of
regressing AEXPt1 on UEMt1 and DEXt1 and other monitoring variables and firm
characteristics in quarter t1.
AIDPt1 A dummy variable measuring analyst independence, equal to 1 if there are independent
analysts following firm i in quarter t1. Independent analysts are defined as those
working for brokers who are not engaged in equity issuances in the 1985–2006
period, as reported in the SDC database. rAIDPt1 is the residual of regressing AIDPt
1 on UEMt1 and DEXt1 and other monitoring variables and firm characteristics in
quarter t1.
IOt1 The percentage of firm i’s outstanding shares held by institutional investors, computed
at the ending quarter t1 using CDA 13(f ). rIOt1 is the residual of regressing IOt1
on UEMt1 and DEXt1 and other monitoring variables and firm characteristics in
quarter t1.
IEXPt1 The average experience of all institutional investors invested in firm i as of quarter t1.
Institutional investors’ experience is measured by the number of quarters an
institution has an investment in firm i. rIEXPt1 is the residual of regressing IEXPt1
on UEMt1 and DEXt1 and other monitoring variables and firm characteristics in
quarter t1.
ISTt1 An indicator measuring the extent to which institutional investors investing in firm i are
short-term-focused. Institutional investors are defined as dedicated (IODED), transient
(IOTRA), and quasi-index (IOQIX), using Bushee’s (1998, 2001) classification. ISTt
1 is the percentage of transient institutional ownership divided by total institutional
ownership for firm i in quarter t1. rISTt1 is the residual of regressing ISTt1 on
UEMt1 and DEXt1 and other monitoring variables and firm characteristics in
quarter t1.
LOGMVt1 The logarithm of the market value (Compustat data item 14 3 data item 61) of firm i’s
common shares at the end of quarter t1.
LOGMTBt1 The logarithm of the market-to-book ratio of firm i at the end of quarter t1, computed
as the market value of common equity divided by the book value of total assets.
NOAt1 Net operating assets scaled by total assets, following Barton and Simko (2002),
measured in quarter t1.
(continued on next page)
Accounting Horizons
September 2014
540 Liu
TABLE 1 (continued)
Variable Definition
HLITt1 An indicator measuring industries with high litigation risk based on SIC code in quarter
t1.
MBEHISt1 The number of firm-quarters out of the most recent four quarters in which the reported
earnings meet or beat analysts’ forecasts, scaled by 4, as of quarter t1.
LOSSHISt1 The number of firm-quarters out of the most recent four quarters in which a firm
reported losses, scaled by 4, as of quarter t1.
DISPt1 The standard deviation of analyst forecasts in quarter t1, scaled by the absolute value
of actual earnings in quarter t1.
percent) are classified as only managing expectations downward, and 1,163 (25.40 percent) are
classified as engaging in both activities.21
Panel B of Table 2 shows the association between the MBE premium, as defined in Bartov et
al. (2002), and managers’ use of the two tactics in achieving MBE. I regress both short-window
abnormal returns upon the release of quarterly earnings and long-window cumulative abnormal
returns over the quarter on earnings surprises and managers’ choices. Similar to the finding in
Bartov et al. (2002), firms that meet or just beat analysts’ forecasts through earnings and
expectations management have a reduction in the MBE premium, and the reduction varies with
managers’ use of MBE tactics. In both regressions, downward expectations management has a
larger negative impact on returns than upward earnings management. Therefore, downward
expectations management appears to be more costly than upward earnings management, perhaps
because forecast revisions are publicly observable.
Table 3 presents the characteristics of firms that meet or just beat analyst forecasts by a narrow
margin and likely use at least one of the two MBE tactics. On average, the actual quarterly earnings
surprise is 0.006 (because of the sample construction). The next four variables proxy for managers’
use of upward earnings management (DACt and UEMt) and downward expectations management
(REVt and DEXt). The average DACt is significantly positive (0.004, with a p-value of 0.000), and
60.1 percent of the sample firm-quarters have positive abnormal accruals (UEMt ¼ 1). The average
REVt is significantly negative (0.043, with a p-value of 0.000), and 65.3 percent of the sample
firm-quarters are classified as managing expectations downward (DEXt ¼ 1).
The next six variables depict the hypothesized factors related to analysts’ and institutional
investors’ monitoring. On average, there are 7.303 analysts following a sample firm in the previous
quarter, and these analysts, on average, have issued 8.192 earnings forecasts in all prior quarters.
Independent analysts issue earnings forecasts for 60.5 percent of the sample firm-quarters.
Institutional investors hold 56.5 percent of the outstanding common shares of the sample firms and,
on average, they have invested in the sample firms for about 12.906 quarters (not necessarily
consecutive investment). Of the total institutional holdings, 27.0 percent are held by short-term-
oriented institutional investors.
The rest are control variables that represent other important characteristics of the sample firms.
LOGMVt1 is the log of the market value of firm i’s common equity in the prior quarter.
LOGMTBt1 is the log of firm i’s market-to-book ratio in the prior quarter. NOAt1 is firm i’s net
21
I exclude firm-quarters that are classified as neither managing earnings upward nor managing expectations
downward. These are firms that likely do not need to manage earnings or expectations to achieve MBE and, thus,
I exclude them from the analysis of the relation between managers’ trade-off and monitoring.
Accounting Horizons
September 2014
TABLE 2
Managers’ Choice between Earnings Management and Expectations Management
Panel A: Frequency of Managers’ Choices

Downward
Observed Upward Earnings Earnings Number of
Choice Management Management Observations Percentage
Choicet ¼ 1 UEMt ¼ 1 DEXt ¼ 0 1,590 34.73%
Panel B: MBE Premiums and Managers’ Choices

Dependent Variable
Two-Day Excess Returns Quarterly Cumulative
on Day [0, 1] Abnormal Returns
Independent Variables Coefficient p-value Coefficient p-value
Intercept 0.002 0.163 0.004 0.308
UEt 0.719 0.000 1.025 0.003
Dummy (Choicet ¼ 1) 0.001 0.794 0.011 0.061
Dummy (Choicet ¼ 2) 0.004 0.049 0.066 0.000
Dummy (Choicet ¼ 3) 0.007 0.004 0.100 0.000
R2 0.01 0.03
F-stat 14.90 0.000 63.72 0.000
This table provides summary statistics of sample firms’ MBE strategies and related stock price reactions. Panel A
includes 4,578 firm-quarters with non-negative earnings surprises not greater than two cents during the 1988–2006
period. It reports the number and the proportion of firm-quarters classified by managers’ use of earnings and expectations
management. UEMt and DEXt are indicators of upward earnings management and downward expectations management,
respectively. Panel B reports the impact of managers’ choice on the MBE premium. The test includes all 7,389 firm-
quarters with non-negative earnings surprises not greater than two cents during the 1988–2006 period. The quarterly
cumulative abnormal returns are computed based on the beta-adjusted return over the period beginning two days
following the date of the first analyst forecast and one day after the release of quarterly earnings, following Bartov et al.
(2002).
I also use CAPM-based cumulative abnormal returns and abnormal buy-and-hold returns as alternative measures of
cumulative abnormal returns and find qualitatively similar results. UE is actual earnings surprise, as defined in Table 1.
Variable Definitions:
Choicet ¼ 1 if UEMt ¼ 1 and DEXt ¼ 0 (upward earnings management only), and 0 otherwise;
Choicet ¼ 2 if UEMt ¼ 0 and DEXt ¼ 1 (downward expectations management only), and 0 otherwise; and
Choicet ¼ 3 if UEMt ¼ 1 and DEXt ¼ 1 (both upward earnings and downward expectations management), and 0
otherwise.
operating assets in the prior quarter, as defined in Barton and Simko (2002). HLITt1 is a dummy
variable to describe a firm’s litigation risk based on industry classification. MBEHISt1 is the
number of MBE quarters in the previous four quarters. LOSSHISt1 is the number of LOSS quarters
in the previous four quarters. DISPt1 is the dispersion of analyst forecasts scaled by the absolute
value of actual earnings in the prior quarter. These variables are included to control for factors that
may be related to both managers’ choice and external monitoring.
Accounting Horizons
September 2014
542 Liu
TABLE 3
Descriptive Statistics of the Sample
Standard
Variables Mean Deviation 1st Quartile Median 3rd Quartile
UEt 0.006 0.007 0.000 0.005 0.010
Proxies for Managers’ Choices
DACt 0.004 0.039 0.016 0.005 0.022
UEMt 0.601 0.490 0.000 1.000 1.000
REVt 0.043 0.097 0.060 0.020 0.000
DEXt 0.653 0.476 0.000 1.000 1.000
Proxies for Analysts and Institutional Investors Monitoring
AFt1 7.303 5.046 4.000 6.000 9.000
AEXPt1 8.192 4.830 4.500 7.000 11.000
AIDPt1 0.605 0.489 0.000 1.000 1.000
IOt1 0.565 0.203 0.432 0.585 0.715
IEXPt1 12.906 7.130 7.076 11.547 17.430
ISTt1 0.270 0.160 0.143 0.243 0.375
Other Control Variables
LOGMVt1 7.022 1.624 5.817 6.882 8.075
LOGMTBt1 0.254 0.786 0.273 0.239 0.791
NOAt1 0.563 0.188 0.452 0.587 0.709
HLITt1 0.359 0.480 0.000 0.000 1.000
MBEHISt1 0.473 0.292 0.250 0.500 0.750
LOSSHISt1 0.090 0.232 0.000 0.000 0.000
DISPt1 0.212 0.433 0.033 0.070 0.184
The sample includes 4,578 firm-quarters with non-negative earnings surprises not greater than two cents and during
which the firms engaged in at least one of the two MBE tactics, upward earnings management, and downward
expectations management, in the 1988–2006 period.
See Table 1 for variable definitions.
Univariate Analysis
Since managers sometimes adopt both tactics to achieve MBE, there are three possible MBE
strategies for managers: upward earnings management only, downward expectations management
only, and the two combined. To examine whether monitoring by analysts and institutional investors
affects managers’ trade-off between earnings and expectations management in achieving MBE, I
partition the sample into three groups based on managers’ use of MBE tactics: Choicet ¼ 1 for firms
that manage only earnings upward (UEMt ¼ 1 and DEXt ¼ 0); Choicet ¼ 2 for firms that manage
only expectations downward (UEMt ¼ 0 and DEXt ¼ 1); and Choicet ¼ 3 for firms that engage in
both tactics (UEMt ¼ 1 and DEXt ¼ 1). I compare monitoring measures and firm characteristics
across groups and report the results in Table 4.
The univariate comparison shows significant differences in monitoring measures between
groups, especially between Group 1 and Group 2. Group 1 firms are followed by the fewest analysts
(AFt1 ¼ 6.89) and the least experienced analysts (AEXPt1 ¼ 7.78). In contrast, Group 2 firms are
followed by the highest number of analysts (AFt1 ¼ 7.72) and the most experienced analysts
(AEXPt1 ¼ 8.47). There is also significantly greater independent analyst presence for Group 2
firms (AIDPt1 ¼ 0.64) than for Group 1 firms (AIDPt1 ¼ 0.55). Regarding institutional investors,
there is no significant difference in total institutional ownership between Group 1 and Group 2 firms
Accounting Horizons
September 2014
TABLE 4
Firm Characteristics by Managers’ Choice
Group 1 Group 2 Group 3 Difference
UEM ¼ 1 UEM ¼ 0 UEM ¼ 1
Choice DEX ¼ 0 DEX ¼ 1 DEX ¼ 1 1 vs. 2 1 vs. 3 2 vs. 3
AFt1 Mean 6.89 7.72 7.21 0.83*** 0.32* 0.51***
Median 5.00 6.00 6.00 1.00*** 1.00* 0.00**
AEXPt1 Mean 7.78 8.47 8.31 0.69*** 0.53*** 0.16
Median 6.71 7.38 7.00 0.67*** 0.29** 0.38*
AIDPt1 Mean 0.55 0.64 0.62 0.09*** 0.07*** 0.02
Median 1.00 1.00 1.00 0.00*** 0.00*** 0.00
IOt1 Mean 0.58 0.57 0.55 0.01 0.03*** 0.02***
Median 0.60 0.59 0.56 0.01* 0.04*** 0.03**
IEXPt1 Mean 12.32 13.56 12.69 1.24*** 0.37 0.87***
Median 10.96 12.15 11.30 1.19*** 0.34 0.85***
ISTt1 Mean 0.29 0.26 0.26 0.03*** 0.03*** 0.00
Median 0.27 0.23 0.23 0.04*** 0.04*** 0.00
LOGMVt1 Mean 7.03 7.16 6.79 0.13** 0.03*** 0.37***
Median 6.88 7.06 6.64 0.18** 0.04*** 0.42***
LOGMTBt1 Mean 0.31 0.30 0.11 0.01 0.20*** 0.19***
Median 0.29 0.29 0.09 0.00 0.20*** 0.20***
NOAt1 Mean 0.55 0.57 0.57 0.02*** 0.02*** 0.00
Median 0.57 0.60 0.59 0.03*** 0.02*** 0.01
HLITt1 Mean 0.37 0.37 0.33 0.00 0.04* 0.04**
Median 0.00 0.00 0.00 0.00 0.00* 0.00**
MBEHISt1 Mean 0.43 0.50 0.49 0.07*** 0.06*** 0.01
Median 0.50 0.50 0.50 0.00*** 0.00*** 0.00
LOSSHISt1 Mean 0.11 0.06 0.11 0.05*** 0.00 0.05***
Median 0.00 0.00 0.00 0.00*** 0.00 0.00***
DISPt1 Mean 0.19 0.20 0.26 0.01 0.07*** 0.06***
Median 0.06 0.07 0.09 0.01** 0.03*** 0.02***
*, **, *** Indicate significance for the two-tailed tests at the 0.10, 0.05, and 0.01 levels, respectively.
This table reports analysts’ and institutional investors’ characteristics by managers’ MBE actions. Group k is defined
based on managers’ MBE strategies: k ¼ 1 if managers engage only in upward earnings management; k ¼ 2 if managers
engage only in downward expectations management; and k ¼ 3 if managers engage in both upward earnings management
and downward expectations management.
(IORt1 ¼ 0.58 and 0.57). The institutional investors’ experience and investment strategies,
however, are significantly different. Group 1 firms are held by institutional investors that are less
experienced (IEXPt1 ¼ 12.32) and more short-term-oriented (ISTt1 ¼ 0.29). Overall, the results are
consistent with the hypotheses (except H4) that managers’ trade-off is related to monitoring by
analysts and institutional investors.
Table 4 also shows that firms taking different actions have systematic differences in other firm
characteristics, such as size (LOGMVt1), balance sheet constraints on earnings management
(NOAt1), MBE records (MBEHISt1), and loss history (LOSSHISt1). In terms of size, Group 1
firms are significantly smaller than Group 2 firms, probably because larger firms are more
extensively covered by the press and are associated with less information asymmetry. In terms of
Accounting Horizons
September 2014
544 Liu
earnings management constraints, Group 1 firms have significantly lower net operating assets than
Group 2 firms, which is consistent with the notion that bloated net operating assets constrain
managers’ capability to manage earnings upward (Barton and Simko 2002). In terms of historical
performance, Group 1 firms have a longer history of losses, while Group 2 firms have a longer
history of MBE. This is consistent with the distinct nature of earnings and expectations
management, because the former focuses on managing reported earnings, while the latter focuses on
managing the market’s expectations.
Multivariate Analysis
Correlations
Next, I conduct a multivariate analysis of the relation between managers’ trade-off and external
monitoring, controlling for other firm characteristics. Table 5 shows pair-wise Pearson correlations
among measures of managers’ choices, monitoring variables, and several firm characteristics for
sample firm-quarters. Conditioning on managers engaging in at least one tactic, the correlation
between UEMt and DEXt is significantly negative, consistent with the two being substitutes in
certain circumstances (Brown and Pinello 2007). Consistent with H1 to H3, the correlations
between UEMt and measures of analyst monitoring (AFt1, AEXPt1, and AIDPt1) are significantly
negative, while those between DEXt and analyst monitoring measures are significantly positive.
Regarding institutional ownership, IOt1 is negatively correlated with both UEMt and DEXt, but
only the correlation with DEXt is significant. The correlations between two other measures of
monitoring by institutional investors (IEXPt1 and ISTt1) and managers’ choices (UEMt and DEXt)
are significant and consistent with H5 and H6. Together, the correlations between monitoring
variables and indicators of earnings and expectations management are consistent with the
hypotheses that managers’ trade-off between the two is related to monitoring by analysts and
institutional investors, with the exception of H4.22
Multivariate Regression
Prior literature suggests that managers possibly make earnings and expectations management
decisions sequentially. A review of analysts’ forecast revisions over time suggests that expectations
management is usually done long before the earnings releases. Figure 1 shows that, on average,
analysts’ forecasts are revised toward actual earnings early on during the quarter. Earnings
management decisions, particularly accruals management, can occur very late in the period, or even
after the end of the period (Dhaliwal, Gleason, and Mills 2004). For example, Kasznik (1999)
demonstrates a scenario in which management earnings forecasts precede earnings management
decisions in the same reporting period. Therefore, I posit that managers decide whether to manage
expectations before they make earnings management decisions.
To examine the structure of managers’ decisions, I conduct my main test using a sequential
logistic estimation (Van Ophem and Schram 1997; Buis 2007). Once analysts release their initial
earnings forecasts, managers decide whether the initial forecasts are too high and downward
expectations management is necessary. If managers choose not to manage expectations downward,
they rely solely on upward earnings management to achieve MBE (Choicet ¼ 1). If managers
choose to manage expectations downward, they then decide whether they need to manage earnings
upward (Choicet ¼ 2 or 3). Consistent with such a sequential setting, a Hausman test confirms that
22
Note that many monitoring variables are significantly correlated. To avoid multicollinearity, when computing a
residual monitoring variable, I also include other monitoring variables in the regression. The residual monitoring
variables are, thus, independent of each other. The residual monitoring variables are also uncorrelated with other
firm characteristics.
Accounting Horizons
September 2014
TABLE 5
Correlations between Managers’ Actions and Proxies for External Monitoring
and Firm Characteristics
Panel A: Correlation Variables UEMt to IEXPt1

UEMt DEXt AFt1 AEXPt1 AIDPt1 IORt1 IEXPt1
DEXt 0.594
(0.00)
AFt1 0.068 0.060
(0.00) (0.00)
AEXPt1 0.048 0.062 0.211
(0.00) (0.00) (0.00)
AIDPt1 0.065 0.079 0.253 0.104
(0.00) (0.00) (0.00) (0.00)
IORt1 0.008 0.039 0.201 0.208 0.126
(0.59) (0.01) (0.00) (0.00) (0.00)
IEXPt1 0.074 0.060 0.320 0.677 0.193 0.247
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
ISTt1 0.069 0.095 0.018 0.211 0.022 0.181 0.248
(0.00) (0.00) (0.22) (0.00) (0.14) (0.00) (0.00)
LOGMVt1 0.071 0.004 0.523 0.395 0.215 0.313 0.611
(0.00) (0.79) (0.00) (0.00) (0.00) (0.00) (0.00)
LOGMTBt1 0.046 0.053 0.065 0.179 0.052 0.027 0.195
(0.00) (0.00) (0.00) (0.00) (0.00) (0.06) (0.00)
NOAt1 0.042 0.063 0.027 0.020 0.025 0.045 0.022
(0.00) (0.00) (0.06) (0.17) (0.09) (0.00) (0.13)
HLITt1 0.018 0.010 0.116 0.063 0.041 0.014 0.122
(0.22) (0.49) (0.00) (0.00) (0.01) (0.36) (0.00)
MBEHISt1 0.067 0.109 0.023 0.009 0.021 0.137 0.030
(0.00) (0.00) (0.12) (0.56) (0.15) (0.00) (0.04)
LOSSHISt1 0.104 0.054 0.075 0.088 0.057 0.160 0.126
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
DISPt1 0.020 0.042 0.055 0.031 0.025 0.090 0.056
(0.18) (0.00) (0.00) (0.03) (0.09) (0.00) (0.00)
Panel B: Correlation Variables ISTt1 to LOSSHISt1, continued from Panel A

ISTt1 LOGMVt1 LOGMTBt1 NOAt1 HLITt1 MBEHISt1 LOSSHISt1
LOGMVt1 0.053
(0.00)
LOGMTBt1 0.149 0.223
(0.00) (0.00)
NOAt1 0.084 0.134 0.294
(0.00) (0.00) (0.00)
HLITt1 0.104 0.037 0.278 0.257
(0.00) (0.01) (0.00) (0.00)
MBEHISt1 0.185 0.087 0.071 0.103 0.016
(0.00) (0.00) (0.00) (0.00) (0.29)
Accounting Horizons
September 2014
546 Liu
TABLE 5 (continued)
ISTt1 LOGMVt1 LOGMTBt1 NOAt1 HLITt1 MBEHISt1 LOSSHISt1
LOSSHISt1 0.100 0.220 0.089 0.094 0.090 0.065
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
DISPt1 0.003 0.171 0.169 0.017 0.046 0.059 0.224
(0.85) (0.00) (0.00) (0.24) (0.00) (0.00) (0.00)
p-values are reported in italic below the correlation coefficients.

This table reports the sample Pearson correlations between managers’ MBE actions and characteristics of analysts and
institutional investors, and firm characteristics.
Choice 1, earnings management only, is significantly different from Choices 2 and 3, both with
expectations management. The regression is, thus, estimated using a sequential logistic model (a
variation of the nested logistic regression), and the results are reported in Table 6.23
Panel A of Table 6 presents the relation between monitoring and managers’ trade-off between
earnings management only (Choicet ¼ 1) and expectations management (Choicet ¼ 2 or 3). The
coefficients are logs of odds ratios. A positive (negative) coefficient on a monitoring variable
suggests that managers are more (less) likely to manage earnings instead of expectations when the
value of that variable is high. Model (1) is the main regression based on the residual values of
lagged monitoring variables, controlling for other firm characteristics and prior quarter’s earnings
and expectations management. Model (2) is based on the raw values of lagged monitoring variables,
reported for comparison purposes. In support of H1 to H3, the coefficients on rAFt1, rAEXPt1,
and rAIDPt1 are all significantly negative, suggesting that managers’ preference shifts from
earnings management toward expectations management when their firms are followed by more
analysts, by more experienced analysts, and by independent analysts. Regarding institutional
investors, the coefficient on rIOt1 is significantly positive, suggesting that managers are less likely
to manage expectations downward if their firms have high institutional ownership. The coefficient
on rIEXPt1 is significantly negative and that on rISTt1 is significantly positive, suggesting that
managers’ preference shifts toward expectations management when their firms are held by more
experienced and less short-term-focused institutional investors.24
Panel B of Table 6 presents the managers’ trade-off between expectations management only
(Choicet ¼ 2) and engaging in both earnings and expectations management (Choicet ¼ 3). The
coefficients on monitoring variables are not significantly different from zero after controlling for the
prior quarter’s earnings and expectations management. The combined results in Panel A and Panel
B suggest that the main effect of analysts’ and institutional investors’ monitoring is to induce
managers to disclose bad news in a timely manner rather than to distort reported earnings.
23
In all regressions, I allow the error terms to cluster by firm and include year and quarter dummies in the
regressions (Petersen 2009). Koh et al. (2008) report that regulatory changes in the post-scandal period increase
the use of expectations management and decrease the use of earnings management. Consistent with their finding,
a comparison of the coefficients on the year dummies shows that managers are more likely to switch to
expectations management in recent years (0.204 pre-scandal versus 0.468 during/post-scandal, on average).
24
The coefficient on IEXPt1 is negative, but insignificant, in the original model, because IEXPt1 and AEXPt1 are
highly positively correlated (0.677 with p-value of 0.00 in Table 5), which leads to multicollinearity. The
coefficient becomes significant if AEXPt1 is dropped. The residual values in Model (2) by construction are not
subject to this problem.
Accounting Horizons
September 2014
FIGURE 1
Median Analyst Forecast Errors before Earnings Announcements
Figure 1 shows the revisions of analysts’ forecasts before earnings announcements for firm-quarters with non-
negative earnings surprises less than two cents. The average number of days between the first forecast date and
the earnings announcement date is 87, and the minimum is 34. I include only firm-quarters with at least 70 days
(five two-week periods) between the first forecast date and the earnings announcement date, which includes
approximately 90 percent of the sample firm-quarters. The horizontal axis shows the two-week window periods
before earnings announcement. The vertical axis shows the sign and the magnitude of the median analyst
forecast errors in the corresponding two-week window period. Window 6 includes all forecasts released at
least 71 calendar days before the earnings announcement. Window 5 includes all forecasts released in the two
weeks from 70 days to 57 days before the earnings announcement. Window 4 includes all forecasts released
in the two weeks from 56 days to 43 days before the earnings announcement. Window 3 includes all forecasts
released in the two weeks from 42 days to 29 days before the earnings announcement. Window 2 includes all
forecasts released in the two weeks from 28 days to 14 days before the earnings announcement. Window 1
includes all forecasts released in the most recent two weeks before the earnings announcement.
Sensitivity Analysis
I conduct a series of additional tests to assess the sensitivity of the results in Table 6. First, I test
whether the findings are sensitive to the sample selection criterion based on earnings surprises and
report the results in Panel A of Table 7. Column (1) in Panel A reports regression results using firms on
both sides of the discontinuity of earnings surprise (with absolute earnings surprises less than two
cents). It is possible that managers engage in earnings and/or expectations management, but still fail to
meet or beat analysts’ forecasts. Column (2) in Panel A reports results for firms that beat analysts’
forecasts with large positive earnings surprises. The results are not sensitive to these alternative samples.
Panel B of Table 7 reports results of simplified logistic regressions. The dependent variables of
Models (1) and (2) are DEXt and UEMt, respectively. Model (3) excludes the MBE strategy of using
Accounting Horizons
September 2014
548 Liu
TABLE 6
Main Results: Analysts’ and Institutional Investors’ Monitoring and Managers’ Use of
Earnings and Expectations Management
Panel A: Choicet ¼ 1 versus Choicet ¼ 2 or 3
Model (1): Residual Values of Monitoring Model (2): Raw Values of Monitoring
Indep. Var. Coeff. p-value Indep. Var. Coeff. p-value
rAFt1 0.028 0.001 AFt1 0.025 0.003
rAEXPt1 0.043 0.000 AEXPt1 0.020 0.043
rAIDPt1 0.151 0.000 AIDPt1 0.327 0.000
rIOt1 0.543 0.006 IOt1 0.382 0.040
rIEXPt1 0.034 0.000 IEXPt1 0.005 0.538
rISTt1 1.767 0.000 ISTt1 1.400 0.000
UEMt1 0.072 0.305 UEMt1 0.051 0.462
DEXt1 0.471 0.000 DEXt1 0.393 0.000
LOGMVt1 0.014 0.567 LOGMVt1 0.114 0.001
LOGMTBt1 0.080 0.100 LOGMTBt1 0.017 0.752
NOAt1 0.416 0.036 NOAt1 0.365 0.067
DISPt1 0.160 0.069 DISPt1 0.124 0.162
HLITt1 0.072 0.370 HLITt1 0.032 0.701
MBEHISt1 0.879 0.000 MBEHISt1 0.687 0.000
LOSSHISt1 0.688 0.000 LOSSHISt1 0.617 0.000
Intercept 0.008 0.624 Intercept 0.191 0.598
Panel B: Choicet ¼ 3 versus Choicet ¼ 2

Model (1): Residual Values of Monitoring Model (2): Raw Values of Monitoring
Indep. Var. Coeff. p-value Indep. Var. Coeff. p-value
rAFt1 0.002 0.868 AFt1 0.000 0.983
rAEXPt1 0.004 0.799 AEXPt1 0.008 0.505
rAIDPt1 0.024 0.515 AIDPt1 0.062 0.468
rIOt1 0.050 0.835 IOt1 0.081 0.717
rIEXPt1 0.016 0.166 IEXPt1 0.013 0.175
rISTt1 0.285 0.434 ISTt1 0.211 0.528
UEMt1 0.300 0.000 UEMt1 0.295 0.000
DEXt1 0.105 0.244 DEXt1 0.112 0.216
LOGMVt1 0.085 0.003 LOGMVt1 0.054 0.177
LOGMTBt1 0.236 0.000 LOGMTBt1 0.272 0.000
NOAt1 0.535 0.021 NOAt1 0.550 0.019
DISPt1 0.102 0.230 DISPt1 0.099 0.247
HLITt1 0.143 0.113 HLITt1 0.152 0.098
MBEHISt1 0.152 0.279 MBEHISt1 0.118 0.400
LOSSHISt1 0.773 0.000 LOSSHISt1 0.752 0.000
Intercept 1.206 0.001 Intercept 1.057 0.008
v2 291.68 0.000 v2 291.98 0.000
Accounting Horizons
September 2014
TABLE 6 (continued)
p-values are computed using firm-clustered error terms.

This table reports the main results of sequential logistic regressions of managers’ choice among three earnings/
expectations-management-based MBE strategies on analysts’ and institutional investors’ characteristics that are
hypothesized to be related to the effectiveness of their monitoring. Managers first decide whether to manage expectations
downward (Choices 2 and 3 versus Choice 1) and later decide whether to manage earnings upwards (Choice 2 versus
Choice 3), conditioning on their first-step decisions. k depicts managers’ choices: k ¼ 1 if managers engage only in
upward earnings management; k ¼ 2 if managers engage only in downward expectations management; and k ¼ 3 if
managers engage in both upward earnings management and downward expectations management.
ProbðChoicet ¼ kÞ
¼ f ða þ b1 rAFt1 þ b2 rAEXPt1 þ b3 rAIDPt1 þ b4 rIOt1 þ b5 rIEXPt1 þ b6 rISTt1
þb7 UEMt1 þ b8 DEXt1 þ b9 LOGMVt1 þ b10 LOGMTBt1 þ b11 NOAt1
þb12 DISPt1 þ b13 HLITt1 þ b14 MBEHISt1 þ b15 LOSSHISt1 Þ þ e: ð1Þ
rAF (and likewise, rAEXP, rAIDP, rIO, rIEXP, and rIST) in Model (1) are residual values from the following regression:
AFt1 ¼ a þ b1 AEXPt1 þ b2 AIDPt1 þ b3 IOt1 þ b4 IEXPt1 þ b5 ISTt1 þ b6 UEMt1 þ b7 DEXt1
þ b8 LOGMVt1 þ b9 LOGMTBt1 þ b10 NOAt1 þ b11 DISPt1 þ b12 HLITt1 þ b13 MBEHISt1
þ b14 LOSSHISt1 þ e; where e ¼ rAFt1 :
¼ f ða þ b1 AFt1 þ b2 AEXPt1 þ b3 AIDPt1 þ b4 IOt1 þ b5 IEXPt1 þ b6 ISTt1
þb12 DISPt1 þ b13 HLITt1 þ b14 MBEHISt1 þ b15 LOSSHISt1 Þ þ e: ð2Þ
AF, AEXP, AIDP, IO, IEXP, and IST in Model (2) are original values.
The coefficients are log odds ratios and represent the effects of the explanatory variables on managers’ choices
sequentially. Year and quarter dummies are not reported, but are included in all regressions.
both earnings and expectations management and focuses on comparing earnings management only
and expectations management only (the dependent variable is DEXt). The effects of analysts’ and
institutional investors’ monitoring are not qualitatively affected by these simplifications. Overall,
Table 7 reports results consistent with the findings in Table 6, that monitoring by analysts and
institutional investors shifts managers’ MBE strategy from earnings management toward
expectations management.
Second, I examine whether the findings are sensitive to the proxy used for downward
expectations management.25 DEX is a result-oriented expectations management measure; that is, it
measures whether managers guide expectations downward based on the overall revision of
25
Prior research measures expectations management using one of two approaches: an outcome-based approach or
an action-based approach. The outcome-based approach argues that while managers’ efforts to guide analysts’
earnings forecasts may not be observable, the outcome of managers’ downward expectations management can be
observed in the downward analyst forecast revisions toward a beatable level. The action-based approach argues
that publicly announced management earnings forecasts effectively move analysts’ forecasts to a beatable level
and can be used to proxy for expectations management. Both measures have their own limitations. The outcome-
based measure may capture the impact of events other than guidance, and the public guidance-based measure
may not capture expectations management through private communications with analysts, especially before
Regulation FD. Using both measures, thus, better captures managers’ expectations management.
Accounting Horizons
September 2014
550 Liu
TABLE 7
Sensitivity Checks: Replication of Main Result Using Alternative Samples
Panel A: Alternative Samples
(1) (2)
(0.02 UEt 0.02) (0.02 , UEt 0.20)
Choicet: 1 vs. 2, 3 Choicet: 3 vs. 2 Choicet: 1 vs. 2, 3 Choicet: 3 vs. 2
Indep. Var. Coeff. p-value Coeff. p-value Coeff. p-value Coeff. p-value
rAFt1 0.016 0.024 0.000 0.994 0.023 0.000 0.006 0.454
rAEXPt1 0.021 0.020 0.004 0.747 0.019 0.009 0.002 0.871
rAIDPt1 0.138 0.000 0.021 0.565 0.120 0.000 0.025 0.369
rIOt1 0.337 0.039 0.093 0.693 0.258 0.057 0.252 0.168
rIEXPt1 0.021 0.010 0.017 0.137 0.022 0.001 0.006 0.545
rISTt1 1.231 0.000 0.286 0.419 1.338 0.000 0.408 0.163
UEMt1 0.054 0.363 0.300 0.000 0.041 0.396 0.269 0.000
DEXt1 0.323 0.000 0.119 0.182 0.500 0.000 0.002 0.981
LOGMVt1 0.011 0.591 0.083 0.003 0.000 0.980 0.035 0.127
LOGMTBt1 0.029 0.498 0.240 0.000 0.035 0.292 0.251 0.000
NOAt1 0.096 0.571 0.516 0.025 0.422 0.002 0.567 0.002
DISPt1 0.044 0.232 0.002 0.946 0.031 0.220 0.026 0.387
HLITt1 0.123 0.079 0.131 0.143 0.035 0.563 0.156 0.040
MBEHISt1 0.661 0.000 0.144 0.303 1.113 0.000 0.124 0.245
LOSSHISt1 0.721 0.000 0.804 0.000 0.684 0.000 0.823 0.000
Intercept 0.941 0.001 1.195 0.002 1.108 0.000 0.637 0.637
Fit statistics
v2 239.2 0.000 631.9 0.000
Panel B: Alternative Models

(1) (2) (3)
DV ¼ DEXt DV ¼ UEMt DV ¼ DEXt
Indep. Var. Coeff. p-value Coeff. p-value Coeff. p-value
rAFt1 0.028 0.001 0.017 0.034 0.028 0.002
rAEXPt1 0.042 0.000 0.025 0.029 0.045 0.001
rAIDPt1 0.152 0.000 0.081 0.006 0.148 0.000
rIOt1 0.496 0.009 0.232 0.236 0.407 0.071
rIEXPt1 0.035 0.000 0.033 0.001 0.046 0.000
rISTt1 1.727 0.000 1.268 0.000 1.970 0.000
UEMt1 0.072 0.302 0.247 0.000 0.180 0.029
DEXt1 0.473 0.000 0.168 0.026 0.416 0.000
LOGMVt1 0.015 0.542 0.050 0.038 0.018 0.526
LOGMTBt1 0.081 0.096 0.113 0.022 0.002 0.973
NOAt1 0.417 0.036 0.605 0.002 0.666 0.004
DISPt1 0.147 0.094 0.009 0.902 0.118 0.237
HLITt1 0.073 0.364 0.127 0.106 0.132 0.149
MBEHISt1 0.878 0.000 0.603 0.000 0.904 0.000
LOSSHISt1 0.683 0.000 0.943 0.000 1.143 0.000
Accounting Horizons
September 2014
TABLE 7 (continued)
(1) (2) (3)
DV ¼ DEXt DV ¼ UEMt DV ¼ DEXt
Intercept 0.615 0.073 2.298 0.000 1.836 0.000
Fit statistics
v2 294.2 0.000 211.6 0.000 276.1 0.000

This table reports the results of sensitivity checks for the main results using alternative samples and model specifications.
Panel A reports the results of the sequential logistic regressions using two alternative samples. The sequential logistic
regressions regress managers’ choice among three earnings/expectations-management-based MBE strategies on analysts’
and institutional investors’ characteristics that are hypothesized to be related to the effectiveness of their monitoring.
Managers first decide whether to manage expectations downward (Choices 2 and 3 versus Choice 1) and later decide
whether to manage earnings upwards (Choice 2 versus Choice 3), conditioning on their first-step decisions. k depicts
managers’ choices: k ¼ 1 if managers engage only in upward earnings management; k ¼ 2 if managers engage only in
downward expectations management; and k ¼ 3 if managers engage in both upward earnings management and downward
expectations management.
¼ f ða þ b1 rAFt1 þ b2 rAEXPt1 þ b3 rAIDPt1 þ b4 rIOt1 þ b5 rIEXPt1 þ b6 rISTt1
þb12 FDEVt1 þ b13 HLITt1 þ b14 MBEHISt1 þ b15 LOSSHISt1 Þ þ e:
sequentially. Panel B reports the results of simple logistic regressions using the same set of explanatory variables as in
the above equation. The dependent variable of Model (1) is DEXt, that of Model (2) is UEMt, and that of Model (3) is
DEXt (and the comparison is only between Choicet ¼ 1 and Choicet ¼ 2). Year and quarter dummies are not reported, but
are included in all regressions.
See Table 1 for variable definitions and Table 6 for the calculation of rAF, rAEXP, rAIDP, rIO, rIEXP, and rIST.
analysts’ forecasts. A limitation of this measure is that it also captures the impact of other
information available to analysts. I use managers’ public earnings guidance as an alternative
measure to replicate the tests in Table 6 and report the results in Column (A) of Table 8. The
coefficients on the monitoring variables overall remain qualitatively unchanged, except that rAFt1
and rIOt1 become insignificant.
Third, recent research shows that managers may also use real earnings management to achieve
MBE (Shih 2014). To test whether the findings in Table 6 are sensitive to managers’ use of real
earnings management as an alternative MBE tactic, I partition the sample into two subsamples, with
and without income-increasing real earnings management, and replicate the main test in Table 6.26
Column (B) in Table 8 shows that, overall, the effects of analysts and institutional investors on
managers’ use of earnings and expectations management are not affected, except that the coefficient
on rIOt1 becomes insignificant for firms with real earnings management.
26
Following Roychowdhury (2006), a firm accelerating sales through increased price discounts or more lenient
credit terms will have abnormally low cash flow from operations. Abnormal cash flow from operations is
estimated using the cross-sectional regression in Roychowdhury (2006) for every industry and year, including
quarter dummies.
Accounting Horizons
September 2014
552 Liu
TABLE 8
Sensitivity Checks: Alternative Expectations Management Measure and Real Earnings
Management
Panel A: Choicet ¼ 1 versus Choicet ¼ 2 or 3

(B)
(A)
Real Earnings Management
Guidance-Based
Expectations Without Abnormally Low With Abnormally Low
Management Operating Cash Flows Operating Cash Flows
rAFt1 0.004 0.724 0.041 0.000 0.034 0.013
rAEXPt1 0.050 0.001 0.054 0.001 0.056 0.000
rAIDPt1 0.252 0.000 0.068 0.094 0.240 0.000
rIOt1 0.418 0.142 0.727 0.004 0.297 0.285
rIEXPt1 0.051 0.000 0.047 0.000 0.023 0.095
rISTt1 0.778 0.053 2.636 0.000 1.440 0.000
UEMt1 0.084 0.395 0.016 0.872 0.018 0.858
DEXt1 1.459 0.000 0.153 0.347 0.558 0.002
LOGMVt1 0.089 0.005 0.082 0.006 0.073 0.051
LOGMTBt1 0.038 0.538 0.087 0.172 0.193 0.007
NOAt1 0.154 0.551 0.049 0.871 0.008 0.980
DISPt1 0.078 0.189 0.097 0.284 0.290 0.012
HLITt1 0.338 0.002 0.050 0.654 0.262 0.029
MBEHISt1 0.442 0.005 0.914 0.000 0.745 0.000
LOSSHISt1 1.856 0.000 0.999 0.000 0.080 0.701
Intercept 6.372 0.000 0.555 0.222 0.548 0.276
Panel B: Choicet ¼ 3 versus Choicet ¼ 2

(B)
(A) Real Earnings Management
Guidance-Based
rAFt1 0.012 0.429 0.002 0.899 0.008 0.715
rAEXPt1 0.017 0.455 0.024 0.183 0.020 0.538
rAIDPt1 0.007 0.915 0.026 0.609 0.059 0.494
rIOt1 0.639 0.132 0.122 0.708 0.479 0.369
rIEXPt1 0.030 0.168 0.003 0.826 0.025 0.363
rISTt1 0.066 0.913 0.701 0.162 0.578 0.505
UEMt1 0.179 0.213 0.280 0.011 0.046 0.807
DEXt1 0.141 0.395 0.202 0.253 0.103 0.700
LOGMVt1 0.010 0.864 0.065 0.110 0.153 0.030
LOGMTBt1 0.326 0.004 0.152 0.080 0.018 0.906
NOAt1 0.634 0.155 0.128 0.704 0.143 0.787
DISPt1 0.025 0.632 0.061 0.323 0.005 0.944
HLITt1 0.074 0.664 0.039 0.740 0.014 0.953
MBEHISt1 0.033 0.896 0.322 0.090 0.115 0.723
Accounting Horizons
September 2014
TABLE 8 (continued)
(B)
(A) Real Earnings Management
Guidance-Based
LOSSHISt1 0.662 0.127 0.910 0.004 0.232 0.469
Intercept 15.191 0.000 1.075 0.047 4.482 0.000
Fit statistics for the entire model (A and B)
v2 391.2 0.000 631.9 0.000 655.9 0.000
Year and quarter dummies are not reported, but are included in all regressions.
This table reports the results of sensitivity checks for the main results using an alternative expectations management
measure based on managers’ public earnings guidance and incorporating real earnings management. The sequential
logistic regressions regress managers’ choice among three earnings/expectations-management-based MBE strategies on
analysts’ and institutional investors’ characteristics that are hypothesized to be related to the effectiveness of their
monitoring. Managers first decide whether to manage expectations downward (Choices 2 and 3 versus Choice 1) and
later decide whether to manage earnings upwards (Choice 2 versus Choice 3), conditioning on their first-step decisions. k
depicts managers’ choices: k ¼ 1 if managers engage only in upward earnings management; k ¼ 2 if managers engage
only in downward expectations management; and k ¼ 3 if managers engage in both upward earnings management and
downward expectations management.
sequentially.
See Table 1 for variable definitions and Table 6 for the calculation of rAF, rAEXP, rAIDP, rIO, rIEXP, and rIST.
ADDITIONAL ANALYSES
Do Financial Analysts Understand Earnings Management?
The finding in the previous section that managers prefer expectations management to earnings
management under strong analyst monitoring is consistent with the findings of Degeorge et al.
(2013) and Yu (2008) that analysts help curb earnings management. One explanation is that higher
intensity and quality of analyst coverage result in more effective monitoring of earnings
management, because these factors lead to analysts having a better understanding of accruals. To
examine if this is the case, I adopt Bradshaw et al.’s (2001) methodology and examine whether the
association between analysts’ optimistic initial forecasts and positive abnormal accruals is
moderated by monitoring characteristics.
Table 9 presents how analysts’ initial over-optimism is related to abnormal accruals and
characteristics of analyst monitoring. I partition DAC into deciles by quarter and examine the initial
forecast errors for the top decile. The initial forecast errors are negative in all three panels of Table
9, regardless of the characteristics of analyst monitoring, suggesting that analysts are over-
optimistic and, in general, do not fully undo the effect of earnings management.
Analysts’ initial over-optimism, however, is mitigated by monitoring measures. In Panel A, I
partition the sample into quartiles by AF and find that for firms with the most positive DAC (Decile
10), the initial forecast errors are smaller in magnitude for firms followed by more financial analysts
(Quartile 4). The difference is relatively large (0.0013 versus 0.0043) and statistically
significant, which suggests that analysts are more conservative in predicting future earnings for
firms with large positive abnormal accruals when analyst following is high. In Panel B, I partition
the sample into quartiles by AEXP and find that for firms with the most positive DAC, the initial
Accounting Horizons
September 2014
554 Liu
TABLE 9
Analyst Monitoring and Initial Forecast Errors
Panel A: Initial Forecast Errors and Analyst Following
Initial Forecast Error
Deciles of DAC Analyst Following Mean Median
Decile 10 Quartile 1 0.0043 0.0000
Quartile 4 0.0013 0.0000
Difference 0.0030 0.0000
p-value 0.000 0.019
Panel B: Initial Forecast Errors and Analyst Experience

Deciles of DAC Analyst Experience Mean Median
Decile 10 Quartile 1 0.0033 0.0000
Quartile 4 0.0017 0.0000
p-value 0.010 0.021
Panel C: Initial Forecast Errors and Analyst Independence

Deciles of DAC Analyst Independence Mean Median
Decile 10 AIDP ¼ 0 0.0038 0.0000
AIDP ¼ 1 0.0023 0.0000
p-value 0.0011 0.004
This table compares the over-optimism at the beginning of the forecast period documented by Bradshaw et al. (2001)
across firms with different levels of analyst monitoring measures. Initial forecast error is computed using the actual
earnings and the average (median) analyst forecasts in the first three weeks following the previous quarter’s earnings
announcement (90 percent of firms have the first individual forecast within 21 days after the previous quarterly earnings
announcement date), scaled by the beginning-of-quarter stock price.
Deciles of DAC are created for every fiscal quarter. Decile 10 includes firms with the highest (most positive) DAC.
Analyst following quartiles are created using AF. Quartile 1 includes firms with the lowest analyst following, and
Quartile 4 includes firms with the highest analyst following. Analyst experience quartiles are created using AEXP.
Quartile 1 includes firms followed by the least experienced analysts, on average, and Quartile 4 includes firms followed
by the most experienced analysts, on average. Analyst independence partitions are created using the binary variable
AIDP. The group with AIDP ¼ 1 includes firms that are followed by at least one independent analyst.
forecast errors are significantly less optimistic for firms followed by more experienced analysts
(0.0017 versus 0.0033). In Panel C, I find that for firms with the most positive DAC, the initial
forecast errors are significantly less optimistic for firms followed by at least one independent analyst
(0.0023 versus 0.0038). Overall, the results suggest that financial analysts can better incorporate
the impact of upward earnings management into earnings forecasts when firms are under stronger
analysts’ monitoring.
Accounting Horizons
September 2014
Downward Expectations Management and Short-Term Institutional Investors

The results of the multivariate analysis suggest that managers prefer earnings management to
expectations management when institutional investors are short-term-oriented. One explanation is
that downward expectations management causes short-term-oriented institutional investors to sell
their holdings. Since the extant literature does not provide direct evidence regarding the impact of
expectations management on institutional holdings, I conduct an additional analysis to examine
whether institutional ownership changes along with expectations management. I partition all firms
that meet or just beat analysts’ forecasts into two groups, based on whether they engage in
downward expectations management or not. I then compare the changes of institutional ownership
between the two groups. The results are reported in Table 10.
In Panel A of Table 10, for firms that achieve MBE without downward expectations
management, the increase of total institutional ownership is significantly larger (0.0121 versus
0.0015). I classify institutional investors into long-term and short-term-oriented based on
Bushee’s (1998, 2001) classification and report the changes in the partitioned institutional
ownerships separately in Panels B and C. In Panel B, long-term-oriented institutional ownership
increases for all firms that achieve MBE, although firms managing expectations downward have
significantly lower increases in long-term-oriented institutional ownership (0.0062 versus
0.0034). Panel C shows, however, that firms managing expectations downward experience
statistically significant decreases in short-term-oriented institutional ownership (0.0018, p-value
of 0.000), although these firms eventually achieve MBE. In contrast, firms without downward
expectations management have statistically significant increases in short-term-oriented institu-
tional ownership (0.0061, p-value of 0.000). Overall, the changes of institutional ownership
suggest that downward expectations management has a negative impact on contemporaneous
institutional holdings, especially on short-term-oriented institutional ownership. In anticipation of
short-term-oriented institutional investors’ reactions to downward expectations management,
managers who observe high short-term-oriented institutional ownership in the prior quarter would
consider downward expectations management to be more costly relative to upward earnings
management.
REGULATORY AND MANAGERIAL IMPLICATIONS

This study has implications for both regulators and corporate executives. In the late 1990s,
regulators who were concerned about the deteriorating quality of financial reporting and disclosure
called on the professional investment community to act as watchdogs for investors (Levitt 1998).
This study shows that regulators’ emphasis on enhancing the ethics and qualification of investment
professionals was not misdirected.27 This study suggests that analysts’ and institutional investors’
monitoring becomes more effective if they are more experienced and have the right motivation or
fewer conflicts of interest. Firms under more effective monitoring by analysts and institutional
investors are more likely to engage in expectations management than earnings management, which
leads to early disclosure of bad news and less distortion in reported earnings.
For managers, this study shows that sophisticated users of financial information play a
monitoring role on financial reporting and disclosure. Monitoring by analysts and institutional
investors increases the cost of managing earnings upward relative to that of managing expectations
downward. Alleged earnings management or withholding bad news can trigger costly litigations.
Managers who are concerned about litigation risk might want to avoid earnings management-based
27
For example, regulations targeting analysts’ conflicts of interests and independence help enhance analysts’
monitoring incentives, and regulations limiting managers’ selective disclosure help lower the cost of analysts’
monitoring by maintaining fair access to information.
Accounting Horizons
September 2014
556 Liu
TABLE 10
Downward Expectations Management and Change of Institutional Ownership
Panel A: Total Institutional Ownership
Change of Total No Downward Downward Difference in
Institutional Expectations Expectations Institutional
Ownership Management Management Ownership Change
Mean 0.0121 0.0015 0.0106
(p-value) (0.000) (0.002) (0.000)
Median 0.0073 0.0014 0.0059
(p-value) (0.000) (0.000) (0.000)
Panel B: Long-Term-Oriented Institutional Ownership

Change of Dedicated No Downward Downward Difference in
Mean 0.0062 0.0034 0.0028
(p-value) (0.000) (0.000) (0.000)
Median 0.0039 0.0022 0.0017
(p-value) (0.000) (0.000) (0.000)
Panel C: Short-Term-Oriented Institutional Ownership

Change of Transient No Downward Downward Difference in
Mean 0.0061 0.0018 0.0079
(p-value) (0.000) (0.000) (0.000)
Median 0.0026 0.0005 0.0031
(p-value) (0.000) (0.000) (0.000)
p-values are based on two-tailed tests (t-tests for mean comparisons and Wilcoxon tests for median comparisons).
This table compares the changes of institutional ownership for all institutional investors, as well as for categorized
institutional investors, between firms that do not manage expectations downward and firms engaged in downward
expectations management. All firms have non-negative earnings surprises that are no more than two cents. Change of
institutional ownership is computed using institutional ownership at the beginning of the quarter and at the end of the
quarter, retrieved from Thomson’s Financial 13(F) database. Long-term-oriented institutional ownership includes the
dedicated and the quasi-index institutional ownerships defined and classified by Bushee (1998, 2001), and short-term-
oriented institutional ownership includes the transient institutional ownerships defined and classified by Bushee (1998,
2001).
strategies to meet or beat analysts’ forecasts in the presence of strong external monitoring. This
study also suggests that managers need to be mindful about who are interested in or investing in
their firms. For instance, long-term-oriented investors are less likely to trade on short-term revisions
of earnings expectations, suggesting another benefit of increasing long-term-oriented institutional
ownership.
Accounting Horizons
September 2014
CONCLUDING REMARKS
This study examines managers’ trade-off between earnings and expectations management for
firms that meet or just beat analysts’ forecasts by a narrow margin, focusing on the effects of
monitoring by financial analysts and institutional investors. The results show significant
associations between managers’ trade-off and a set of measures for monitoring effectiveness,
including analyst following, analyst experience, analyst independence, institutional ownership,
institutional investors’ experience, and institutional investors’ investment strategy. Additional
analysis shows that financial analysts can, albeit inefficiently, incorporate information on abnormal
accruals into their forecasts of future earnings. Moreover, the association between the change of
institutional ownership and downward expectations management provides further evidence to
explain why managers prefer earnings management to expectations management when investors are
short-term-oriented.
Together, the findings of this study suggest that sophisticated users of accounting information,
although outsiders, can play a role monitoring financial reporting and disclosure. Overall,
monitoring by analysts and institutional investors shifts managers’ preference away from earnings
management toward expectations management in meeting or beating analysts’ earnings forecasts,
which helps reduce distortions in reported earnings and increase timely disclosure of bad news. In
addition, the findings suggest that it is necessary to consider the experience and conflicting
incentives of these two external parties when examining their monitoring of financial reporting.
REFERENCES
Ajinkya, B., S. Bhojraj, and P. Sengupta. 2005. The association between outside directors, institutional
investors and the properties of management earnings forecasts. Journal of Accounting Research 43
(3): 343–376.
Ayers, B., J. Jiang, and P. E. Yeung. 2006. Discretionary accruals and earnings management: An analysis of
pseudo earnings targets. The Accounting Review 81 (3): 617–652.
Barton, J., and P. J. Simko. 2002. The balance sheet as an earnings management constraint. The Accounting
Review 77 (Supplement): 1–27.
Bartov, E., D. Givoly, and C. Hayn. 2002. The rewards to meeting or beating earnings expectations. Journal
of Accounting and Economics 33 (2): 173–204.
Bradshaw, M., S. Richardson, and R. Sloan. 2001. Do analysts and auditors use information in accruals?
Journal of Accounting Research 39 (1): 45–74.
Brochet, F., G. Miller, and S. Srinivasan. 2014. Do analysts follow managers who switch companies? An
analysis of relationships in the capital markets. The Accounting Review 89 (2): 451–482.
Brown, L. D., and A. S. Pinello. 2007. To what extent does the financial reporting process curb earnings
surprise games? Journal of Accounting Research 45 (5): 947–981.
Brown, L. D., and M. L. Caylor. 2005. A temporal analysis of quarterly earnings thresholds: Propensities
and valuation consequences. The Accounting Review 80 (2): 423–440.
Buis, M. L. 2007. SEQLOGIT: Stata module to fit a sequential logit model. Available at: http://ideas.repec.
org/c/boc/bocode/s456843.html
Burgstahler, D., and M. Eames. 2006. Management of earnings and analysts’ forecasts to achieve zero and
small positive earnings surprises. Journal of Business Finance and Accounting 33 (5/6): 633–652.
Burns, N., S. Kedia, and M. Lipson. 2010. Institutional ownership and monitoring: Evidence from financial
misreporting. Journal of Corporate Finance 16 (4): 443–455.
Bushee, B. J. 1998. The influence of institutional investors on myopic R&D investment behavior. The
Accounting Review 73 (3): 305–333.
Bushee, B. J. 2001. Do institutional investors prefer near-term earnings over long-term value?
Contemporary Accounting Research 18 (2): 207–246.
Accounting Horizons
September 2014
558 Liu
Cadman, B., and J. Sunder. 2014. Investor horizon and CEO horizon incentives. The Accounting Review
(forthcoming).
Cheng, C. S., H. Huang, Y. Li, and G. Lobo. 2010. Institutional monitoring through shareholder litigation.
Journal of Financial Economics 95 (3): 356–383.
Das, S., K. Kim, and S. Patro. 2011. An analysis of managerial use and market consequences of earnings
management and expectation management. The Accounting Review 86 (6): 1935–1967.
Dechow, P. M., and C. M. Schrand. 2004. Earnings quality. The Research Foundation of CFA Institution.
Available at: http://www.cfapubs.org/doi/pdf/10.2470/rf.v2004.n3.3927
Degeorge, F., J. Patel, and R. Zeckhauser. 1999. Earnings management to exceed thresholds. Journal of
Business 72 (1): 1–33.
Degeorge, F., Y. Ding, T. Jeanjean, and H. Stolowy. 2013. Analyst coverage, earnings management and
financial development: An international study. Journal of Accounting and Public Policy 32: 1–25.
Dhaliwal, D. S., C. A. Gleason, and L. F. Mills. 2004. Last-chance earnings management: Using the tax
expense to meet analysts’ forecasts. Contemporary Accounting Research 21 (2): 431–459.
Donnelly, B. 1990. Profits’ ‘‘quality’’ erodes, making them less reliable. Wall Street Journal (September
18): C1.
Doyle, J., J. Jennings, and M. Soliman. 2013. Do managers define non-GAAP earnings to meet or beat
analyst forecasts? Journal of Accounting and Economics 56 (1): 40–56.
Drake, M., and L. Myers. 2011. Analysts’ accrual-related over-optimism: Do analyst characteristics play a
role? Review of Accounting Studies 16 (1): 59–88.
Dyck, A., A. Morse, and L. Zingales. 2010. Who blows the whistle on corporate fraud? Journal of Finance
65 (6): 2213–2253.
Feng, M., and S. McVay. 2010. Analysts’ incentives to overweight management guidance when revising
their short-term earnings forecasts. The Accounting Review 85 (5): 1617–1646.
First Boston Corporation. 1994. Solid Waste Earnings Quality Analysis. New York, NY: First Boston Corp.
Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial
reporting. Journal of Accounting and Economics 40 (1/3): 1–63.
Groysberg, B., P. Healy, and D. Maber. 2011. What drives sell-side analyst compensation at high-status
investment banks? Journal of Accounting Research 49 (4): 969–1000.
Gu, Z., and J. Xue. 2008. The superiority and disciplining role of independent analysts. Journal of
Accounting and Economics 45 (2/3): 289–316.
Gu, Z., Z. Li, and Y. Yang. 2013. Monitors or predators: The influence of institutional investors on sell-side
analysts. The Accounting Review 88 (1): 137–169.
Guercio, D. D., and J. Hawkins. 1999. The motivation and impact of pension fund activism. Journal of
Financial Economics 52 (3): 293–340.
Gunny, K. 2010. The relation between earnings management using real activities manipulation and future
performance: Evidence from meeting earnings benchmarks. Contemporary Accounting Research 27
(3): 855–888.
Holmstrom, B. 1979. Moral hazard and observability. The Bell Journal of Economics 10 (1): 74–91.
Houston, J., B. Lev, and J. Tucker. 2010. To guide or not to guide? Causes and consequences of stopping
quarterly earnings guidance. Contemporary Accounting Research 27 (1): 143–185.
Hribar, P., and D. Collins. 2002. Errors in estimating accruals: Implications for empirical research. Journal
of Accounting Research 40 (1): 105–134.
Hutton, A. 2005. Determinants of managerial earnings guidance prior to regulation fair disclosure and bias
in analysts’ earnings forecasts. Contemporary Accounting Research 22 (4): 867–914.
Jensen, M., and W. Meckling. 1976. Theory of the firm: Managerial behavior, agency costs, and ownership
structure. Journal of Financial Economics 3 (4): 305–360.
Kasznik, R. 1999. On the association between voluntary disclosure and earnings management. Journal of
Accounting Research 37 (1): 57–81.
Kaufman Bros. Equity Research. 2001. Analysts’ report pushes Qwest to consecutive 52-week lows. Wall
Street Journal (August 24): B4.
Accounting Horizons
September 2014
Koh, K., D. A. Matsumoto, and S. Rajgopal. 2008. Meeting or beating analyst expectations in the post-
scandals world: Changes in stock market rewards and managerial actions. Contemporary Accounting
Research 25 (4): 1067–1098.
Levitt, A. 1998. The numbers game. The CPA Journal 68: 14–19.
Lin, H. W., and M. McNichols. 1998. Underwriting relationships, analysts’ earnings forecasts and
investment recommendations. Journal of Accounting and Economics 25 (1): 101–127.
Matsumoto, D. A. 2002. Management’s incentives to avoid negative earnings surprises. The Accounting
Review 77 (3): 483–514.
Mikhail, M. B., B. R. Walther, and R. H. Willis. 1997. Do security analysts improve their performance with
experience? Journal of Accounting Research 35 (Supplement): 131–157.
Mikhail, M. B., B. R. Walther, and R. H. Willis. 2003. The effect of experience on security analyst under-
reaction. Journal of Accounting and Economics 35 (1): 101–166.
O’Brien, P. C., M. F. McNichols, and H. W. Lin. 2005. Analyst impartiality and investment banking
relationships. Journal of Accounting Research 43 (4): 623–650.
Patel, P., H. Barefoot, S. Yao, and R. Carlson. 2007. Earnings Quality: Practical Applications. Credit
Suisse (Americas/United States) Equity Research. Zürich, Switzerland: Credit Suisse.
Petersen, M. A. 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Review
of Financial Studies 22 (1): 435–480.
Phillips, J., M. Pincus, and S. Rego. 2003. Earnings management: New evidence based on deferred tax
expense. The Accounting Review 78 (2): 491–521.
Pulliam, S. 1999. Earnings management spurs selloffs now. Wall Street Journal (October 29): C1.
Ramalingegowda, S., and Y. Yu. 2012. Institutional ownership and conservatism. Journal of Accounting
and Economics 53 (1/2): 98–114.
Rana, E. 2008. Cover me: How can CFOs keep their companies in an analyst’s spotlight? CFO.com.
Available at: http://ww2.cfo.com/risk-compliance/2008/04/cover-me/
Richardson, S., S. H. Teoh, and P. Wysocki. 2004. The walk-down to beatable analyst forecasts: The role of
equity issuance and insider trading incentives. Contemporary Accounting Research 21 (4): 885–924.
Robinson, D., and D. Glushkov. 2006. A Note on I/B/E/S Unadjusted Data. Wharton Research Data
Services. Available at: http://people.stfx.ca/x2009/x2009hyu/stocksplitproblem.pdf
Roychowdhury, S. 2006. Earnings management through real activities manipulation. Journal of Accounting
and Economics 42 (3): 335–370.
Shih, A. 2014. Analyst Attributes and Trade-Offs Between Benchmark Beating Strategies. Working paper,
National University of Singapore.
Shleifer, A., and R. Vishny. 1986. Large shareholders and corporate control. Journal of Political Economy
94 (3): 461–488.
Tucker, J. W. 2010. Is silence golden? Earnings warnings and subsequent changes in analyst following.
Journal of Accounting, Auditing & Finance 25 (3): 431–456.
Van Ophem, H., and A. Schram. 1997. Sequential and multinomial logit: A nested model. Empirical
Economics 22: 131–152.
White, B. J. 2004. EMS Earnings Quality Analysis. New York, NY: Kaufman Bros. Equity Research.
Yu, F. 2008. Analyst coverage and earnings management. Journal of Financial Economics 88 (2): 245–271.
Accounting Horizons
September 2014

Can External Monitoring Affect Corporate

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Can External Monitoring Affect Corporate

Uploaded by

Copyright:

Available Formats

Accounting Horizons American Accounting Association

Vol. 28, No. 3 DOI: 10.2308/acch-50771

Can External Monitoring Affect Corporate

Keywords: external monitoring; earnings management; expectations management;

Alfred Zhu Liu is an Assistant Professor at the University at Albany, SUNY.

LITERATURE AND HYPOTHESES

Managers’ Trade-Off between Earnings and Expectations Management

Proxy for Managers’ Use of Earnings and Expectations Management

Panel A: Frequency of Managers’ Choices

Panel B: MBE Premiums and Managers’ Choices

Panel A: Correlation Variables UEMt to IEXPt1

Panel B: Correlation Variables ISTt1 to LOSSHISt1, continued from Panel A

p-values are reported in italic below the correlation coefficients.

Panel B: Choicet ¼ 3 versus Choicet ¼ 2

p-values are computed using firm-clustered error terms.

Panel B: Alternative Models

p-values are computed using firm-clustered error terms.

Panel A: Choicet ¼ 1 versus Choicet ¼ 2 or 3

Panel B: Choicet ¼ 3 versus Choicet ¼ 2

Panel B: Initial Forecast Errors and Analyst Experience

Panel C: Initial Forecast Errors and Analyst Independence

Downward Expectations Management and Short-Term Institutional Investors

REGULATORY AND MANAGERIAL IMPLICATIONS

Panel B: Long-Term-Oriented Institutional Ownership

Panel C: Short-Term-Oriented Institutional Ownership

You might also like