Do Smooth Earnings Lower Investors` Perceptions of

Do Smooth Earnings Lower Investors’ Perceptions of Investment Risk?
ABSTRACT
Prior research provides mixed evidence concerning whether smooth earnings lower investors’
perceptions of investment risk. In this paper, we use three experiments to examine whether and
why the presence of smooth earnings lowers investors’ risk judgments. In our first experiment,
participants judged a firm with smooth earnings to have lower investment risk when the earnings
subtotal was present than when it was absent. Given this finding, our second experiment
examines the mechanism underlying investors’ perceptions of investment risk when earnings are
smooth. We find that our participants do not demonstrate a conscious preference for smoothed
earnings. Our findings suggest that investors either assume that managers use little reporting
discretion or are unaware of this discretion. Our final experiment considers how disclosures
concerning available reporting discretion affect investors’ risk judgments. We find that the risk
judgments of participants who were informed that managers have high available reporting
discretion were less influenced by the presence of smooth earnings. All three of our experiments
highlight a negative relation between smooth earnings and investment risk judgments. However,
factors that increase investors’ attention to the earnings components, such as disclosures
concerning the use and availability of reporting discretion, mitigate the impact of smooth
earnings on investors’ risk judgments.
Do Smooth Earnings Lower Investors’ Perceptions of Investment Risk?
1. Introduction
The accounting literature has long debated whether smooth earnings lower investors’
perceptions of investment risk relative to volatile earnings. Reviews of the early literature
concerning smooth earnings suggest a negative relation between smooth earnings and investors’
perceptions of a firm’s investment risk (Ronen and Sadan [1981]; Ronen and Yaari [2008]).
However, recent archival research argues that investors base their risk judgments on the
smoothness of operating cash flows that underlie earnings, rather than on the smoothness of
earnings per se (Rountree, Weston, and Allayannis [2008]; McInnis [2010]).
Hirshleifer and Teoh [2009] propose two competing explanations for a relation between
smooth earnings and investment risk judgments. First, investors may attend to smooth earnings
and fail to fully discern whether this smoothness is corroborated by the smoothness of the
earnings components (‘limited attention’). In this study, we define ‘limited attention’ as investors
not fully processing all information concerning the earnings components. For example, limited
attention exists when investors either assume that managers use little reporting discretion or are
unaware of this discretion.1 Alternatively, investors might consciously believe that firms with
smoothed earnings present lower investment risk regardless of the underlying operating cash
flows (‘conscious preference’). The first explanation is consistent with investors underestimating
the discretion that managers might use to smooth earnings, while the second explanation
suggests that investors prefer managers smoothing earnings.
1
The most extreme case of limited attention would result in investors narrowing their attention to only the earnings
subtotal. However, limited attention also includes situations where investors incorporate the earnings components as
well as the earnings subtotal in their judgments yet fail to consider the possibility that earnings have been smoothed
by managers exercising their reporting discretion.
1
Given the potentially adverse effects of investors underestimating managers’ use of
discretion to smooth earnings, our study seeks to address whether investors attend more to other
earnings information when they perceive that reporting discretion is available to managers.
When investors consider the reporting discretion available to managers, we expect that they
attend more to the volatility of operating cash flows when judging risk. Focusing on the risk
associated with the volatility of the earnings components mitigates the effect of smooth earnings
on risk judgments when investors perceive that managers have higher levels of reporting
discretion available to them.
Our study uses experimental methods to address three research questions. First, given the
mixed findings in the archival literature, we use experimental methods to examine whether the
presence of smooth earnings impacts investors’ risk judgments above and beyond the effect of
the earnings components. To the extent that smooth earnings lower investors’ risk judgments,
our second research question considers the mechanism driving investors’ perceptions of smooth
earnings. Finally, we consider how disclosures concerning the reporting discretion available to
managers affect investors’ risk judgments when earnings are smooth.
Our first experiment considers whether the presence of smooth earnings lower investors’
risk judgments relative to their risk judgments based on the earnings components alone. We
employed a 2 x (2) mixed model design manipulating the presence of the earnings subtotal
between subjects (present vs. absent) for two firms (i.e., one with smooth earnings and the other
with volatile earnings). We manipulated the presence of the earnings subtotal to investigate how
investors’ risk judgments are affected when this subtotal reveals smooth earnings. Holding each
firm’s underlying operating cash flows constant, the manipulation of the presence of the earnings
subtotal allows us to observe the effect of smooth earnings on investors’ risk judgments when
2
earnings are easy to observe compared with when they are difficult to observe. If, as recent
archival research suggests, investors do not attend to the earnings subtotal when evaluating risk,
we would not expect the presence of smooth earnings to affect their risk judgments. We find that
the presence of the subtotal reflecting smooth earnings lowered participants’ investment risk
judgments. Thus, we document an interaction between our two independent variables. Our
findings suggest a negative relation between smooth earnings and investment risk judgments.
Our second experiment investigates the mechanism driving the relation between smooth
earnings and investors’ risk judgments. We employed a 2 x 2 between-subjects design
manipulating the accruals reported (smoothing vs. non-smoothing) and whether information was
provided to participants concerning managers’ use of discretion (present vs. absent). All
participants were presented with the earnings components together with the earnings subtotal and
viewed the same operating cash flows. Consistent with our first experiment, when participants
were not presented with information concerning managers’ use of discretion, we observe a
negative relation between smooth earnings and investment risk judgments. That is, the firm
whose managers used accruals to smooth earnings had lower perceived risk when participants
were not made aware of the reporting discretion used by managers. However, when participants
were presented with information concerning managers’ use of discretion, smooth earnings did
not lower participants’ risk judgments. These results suggest that the relation between smooth
earnings and investors’ risk judgments does not stem from investors’ conscious preference for
smoothed earnings. Rather, our findings are consistent with investors assuming that managers
use little reporting discretion or being unaware of this discretion.
Our second experiment illustrates how information concerning managers’ use of
discretion eliminates the negative relation between smooth earnings and investors’ risk
3
judgments. However, managers are seldom required to disclose whether they used discretion.
Instead, standard setters often require managers to disclose the extent of available reporting
discretion. Our third experiment considers whether disclosures concerning the discretion
available to managers affect investors’ risk judgments when earnings are smooth. We employed
a 2 x 2 between-subjects design manipulating the presence of the earnings subtotal and the level
of reporting discretion available to managers (low vs. high).2 We find that disclosures revealing
high reporting discretion available to managers only partially mitigate the negative relation
between smooth earnings and investment risk. We investigate these findings further by
considering participants’ perceptions of the discretion available to managers and find that
participants’ risk judgments are only lowered by smooth earnings when they perceive low
available reporting discretion. To the extent that disclosures concerning the discretion available
to managers activate perceptions of reporting discretion, they appear to assist investors in
broadening their attention to the earnings components.
Our study makes several contributions to the literature. First, all three of our experiments
provide evidence concerning the effect of smooth earnings on investors’ risk judgments.
Experimental methods allow us to capture direct measures of investors’ risk judgments and to
control for the economic fundamentals of firms – specifically, the volatility of operating cash
flows. We demonstrate the effect of smooth earnings on investors’ risk judgments using two
approaches. In our first and third experiments, we manipulated the presence of the earnings
subtotal and held operating cash flows and accruals constant to demonstrate that smooth earnings
lower investors’ risk judgments. In our second experiment, we manipulated the accruals reported
and held operating cash flows constant to demonstrate that smooth earnings lower investors’ risk
2
This reporting discretion relates to the level of reporting discretion available to managers when measuring earnings
information and not whether managers have the discretion as to whether to report the information.
4
judgments. Accordingly, our study provides additional evidence beyond that in the archival
literature on whether smooth earnings lower investors’ risk judgments.
Second, our results do not support the suggestion that investors prefer smoothed earnings.
Instead, our findings are consistent with the ‘limited attention’ explanation for why investors
perceive that firms with smooth earnings present lower investment risk. These findings respond
to the call from Hirshleifer and Teoh [2009] for research to examine the behavioral mechanism
underlying investors’ perceptions of firms with smooth earnings. Our study suggests that
investors either assume that managers use little reporting discretion or are unaware of this
discretion. Further, we document that smooth earnings have less impact on the risk judgments of
participants who perceive that managers had high reporting discretion. Given that perceptions of
reporting discretion likely vary with a number of firm- and time-specific factors, our results
suggest that any variation in these factors could affect the link between smooth earnings and risk
judgments. By identifying the nature of this relation, our study potentially provides an
opportunity for future research to reexamine situations under which smooth earnings are priced.
Finally, our study broadly adds to the extant behavioral accounting literature examining
investors’ attention to earnings components (e.g., Sloan [1996]; Hirshleifer and Teoh [2003];
Hewitt [2009]). Consistent with this prior research, we demonstrate that investors’ judgments are
affected by attending to aggregate earnings. However, unlike prior research, we demonstrate that
this attention can be detrimental or beneficial depending on the circumstances. In our first
experiment, the presence of an earnings subtotal revealing smooth earnings lowered participants’
risk judgments; however, in our third experiment, the presence of the earnings subtotal helped
participants to perceive how managers potentially used their reporting discretion to smooth
5
earnings. Therefore, the presence of the earnings subtotal can enhance, as well as help mitigate,
the negative relation between smooth earnings and investors’ risk judgments.
Section two of this paper discusses our first research question and experiment addressing
the negative relation between smooth earnings and investors’ risk judgments. Section three
discusses our second research question and experiment concerning the underlying mechanism
driving this negative relation. Section four discusses our third research question and experiment
considering whether disclosures concerning available reporting discretion mitigate the relation
between smooth earnings and investors’ risk judgments. Section five concludes the paper.
2. Do Smooth Earnings Affect Investors’ Risk Judgments?
2.1 HYPOTHESIS DEVELOPMENT
Numerous studies examine the causes (e.g., Lambert [1984]; Dye [1988]; Trueman and
Titman [1988]; Goel and Thakor [2003]) and effects (e.g., Francis, LaFond, Olsson, and
Schipper [2004]; Bitner and Dolan [1996]) of smooth earnings.3 Many of these studies
distinguish between smooth earnings that stem from natural causes as opposed to managerial
intervention (Ronen and Yaari [2008]). As suggested by Graham, Harvey, and Rajgopal [2005],
managers might manage earnings to ensure that earnings depict a smooth time series by making
choices that either affect the firm’s operating cash flows or the measurement of accruals.
While smoothing may simply distort the reporting of the firm’s performance (Demski
[1998]), theoretical research in this area establishes reasons why artificially smoothed
performance can benefit shareholders. Trueman and Titman [1988] suggest that investors benefit
from smooth operating performance due to the decreased cost of borrowing and favorable terms
3
Ronen and Yaari [2008] provide a comprehensive review of these studies and other empirical studies addressing
smooth earnings. Ronen and Sadan [1981] provide a thorough discussion and overview of earlier empirical studies
concerning this topic.
6
of trade received by firms with smooth performance. Goel and Thakor [2003] posit that artificial
income smoothing reduces the information losses of liquidity traders and increases the amount
that investors are willing to pay for a firm. Thus, managers could use artificial smoothing to
reduce information asymmetry about the firm’s future prospects and signal their competence in
enhancing the value of the firm for investors.
Given that managers can use the reporting process to either distort or improve the signal
of operating performance, empirical research considers how investors value smooth operating
performance (operating cash flows or earnings). A number of studies report a positive
association between the reporting of smooth performance and stock returns (e.g., Bitner and
Dolan [1996]; Hunt, Moyer, and Shevlin [1997]; Allayannis, Rountree, and Weston [2005];
Bhattacharya, Daouk, and Welker [2003]). Further, Francis et al. [2004] document a negative
association between smooth operating performance and estimates of cost of capital. Recent
survey evidence suggests that corporate managers strongly believe that investors value smooth
earnings independent of cash flow characteristics. In a survey of more than 400 financial
executives, Graham et al. [2005] find that these executives would sacrifice projects with positive
net present values to manufacture the smooth earnings they believe that investors desire. This
evidence suggests that investors face costs as a result of income smoothing.
Most recently, however, studies fail to find evidence that investors perceive smooth
earnings to be valuable. Rountree et al. [2008] find only the smoothness of operating cash flows
– and not earnings – to be associated with firm value. However, the authors highlight that
“(u)nfortunately from a research perspective such a high correlation [between operating cash
flows and earnings] makes it difficult to disentangle the marginal effects of earnings and cashflow volatility” (Rountree et al. [2008, p. 240]). Further, McInnis [2010] attributes the apparent
7
association between smooth earnings and implied cost of capital estimates to optimism in
analysts’ earnings forecasts. Nevertheless, McInnis [2010, p. 317] suggests that “tests that
examine the link between earnings smoothness and cost of equity capital are actually joint tests
of two hypotheses: (1) smooth earnings reduce some form of information risk, and (2) this
information risk is priced.” While archival methods might struggle to identify whether smooth
earnings affect investors’ risk judgments, we argue that experimental methods can help address
this link. Therefore, the primary objective of our study is to utilize experimental methods to
isolate the relation between smooth earnings and investors’ risk judgments.
Our first experiment tests whether smooth earnings lower risk judgments beyond the
effects of the earnings components. This experiment utilizes a comparative advantage of
experimental methods by manipulating only the presence of the earnings subtotal for each firm.
Thus, for each firm, all economic attributes (including operating cash flows) are held constant.
Whereas in natural settings the earnings subtotal is always present, an experiment that
manipulates the presence of the earnings subtotal allows us to consider a setting where the ease
with which investors can attend to smooth earnings is varied while each firm’s earnings
components are held constant. In our setting, a simple effect of the presence of earnings for a
firm with volatile cash flows and smooth earnings would signify that the earnings subtotal affects
investors’ risk judgments. An interaction effect involving the firm analyzed and the presence of
the earnings subtotal suggests that smooth earnings, and not the presence of the earnings subtotal
per se, lower investors’ perceptions of risk.4
When the earnings subtotal is present, investors can easily attend to this information and
are arguably more likely to process it. In contrast, when the earnings subtotal is absent, we
4
In our setting, a main effect between investment risk judgments for the two firms would represent differences in
the two experimental firms’ underlying economics.
8
expect investors to be less likely to process aggregate earnings information. If investors’ risk
judgments are affected by smooth earnings, we expect the presence of the earnings subtotal
reflecting smooth earnings to decrease investors’ risk judgments relative to when only the
earnings components are shown. This expectation is described in the following hypothesis.
H1:
The presence of smooth earnings lowers investors’ risk judgments.
2.2 EXPERIMENT ONE
2.2.1 Design and Participants
We used a 2 x (2) mixed model experimental design. Earnings presence
(NI_PRESENCE) was manipulated between subjects by the earnings subtotal being either
present or absent after the information concerning operating cash flows and accruals.
Importantly, there is no differential information content across the two NI_PRESENCE
conditions for each firm. Adding operating cash flows and accruals together produces the
earnings subtotal. All participants were informed that the information presented to them
consisted of the components of earnings. Therefore, participants in the NI_Absent condition were
aware that they could calculate earnings from the earnings components presented to them.5
Within subjects, we manipulated whether participants judged the investment risk for a
firm with smooth earnings and volatile operating cash flows or a firm with volatile earnings and
smooth operating cash flows (FIRM).6 The latter firm is used to control for the possibility that
the presence of the earnings subtotal reduces participants’ investment risk judgments due to
presentation of more data or earnings itself. Essentially, this firm acts as a benchmark to ensure
5
We believe that such calculations would work against us finding the hypothesized effect. That is, consistent with
Nelson and Tayler [2007], we expect that participants who expended effort to calculate aggregate earnings would
then have relied on this information to judge investment risk. Essentially, these participants would have behaved
similarly to participants in the NI_Present condition. Recall that the purpose of the NI_Absent condition is to
observe participants’ risk judgments when they were presented with the earnings components in isolation.
6
To control for order effects, we also randomly assigned participants to one of two possible orders to view the
firms. Order effects were not significant in our analyses, so we report the 2 x (2) x 2 mixed model design as a
collapsed 2 x (2) mixed model design.
9
that we correctly attribute any findings to smooth earnings. The NI_PRESENCE manipulation
inherently enables us to hold the underlying economics constant for each firm.
All participants received information concerning the firm’s earnings components
(operating cash flows and accruals) for five years. We asked participants to separately judge the
investment risk for each firm and to indicate which firm presented the higher investment risk.
Consistent with prior research (e.g., Pinello [2008], Hirst, Hopkins, and Wahlen [2004], Lipe
[1998]), we asked participants to provide investment risk judgments based on their perceptions
of the risk associated with investing in each firm. After completing these tasks, participants were
asked to state whether operating cash flows or earnings provide the better measure of operating
volatility. As manipulation checks, participants identified the presentation format of the earnings
information provided to them and evaluated the volatility of two number series.
Fifty-eight first-year M.B.A. students from a large state university served as surrogates
for investors. We recruited these students from a required managerial accounting class in the
second semester of the M.B.A. program. These students had recently completed the required
financial accounting class and were familiar with earnings and its components (i.e., operating
cash flows and accruals). Over 93% of participants responded that they had previously evaluated
firm performance using financial information. Participants completed the experiment by
accessing the study’s web address that was sent to each of them via email.
Twenty-six percent of the participants were female, and most participants’ major area of
study was either marketing (43%) or finance (33%). Participants had an average of 5.12 years of
work experience, and 16% of the participants had worked previously as financial analysts. On
average, participants had taken 2.66 and 3.03 undergraduate and graduate accounting/finance
classes, respectively. Finally, 74% of the participants had bought/sold securities, and 91% of all
10
the participants planned to buy/sell securities in the future. Consistent with the findings of Elliott,
Hodge, Kennedy, and Pronk [2007], we believe that our M.B.A.-student participants were
appropriate surrogates for investors due to our experiment’s low level of integrative complexity
and the lack of an investment decision in our experiment. Our results are insensitive to the
demographic information that participants provided.
2.2.2 Earnings Information
Appendix A displays the earnings information presented to participants in the NI_Present
condition. This information was presented such that the earnings components add together to
arrive at earnings. Hodder, Hopkins, and Wood [2008] show that presenting operating cash
flows, accruals, and earnings in this order is more intuitive to investors and leads to lower
forecasting errors relative to commencing with earnings and adding back accruals to arrive at
operating cash flows. To ensure equivalency across conditions, we do not use the presentation
format to communicate that earnings is a subtotal to participants in the NI_Present condition
(i.e., we do not include any summation lines in the presentation format of this condition).
Both firms report similar accrual information and differ only in whether they report
operating cash flows or earnings in a smooth time series. Firm A’s accruals smooth out the
volatility in operating cash flows to achieve smooth earnings, while Firm B’s accruals add
volatility to smooth operating cash flows. As FIRM is manipulated within subjects, we multiply
the accruals reported by Firm A by a constant factor (i.e., 0.97) to produce Firm B’s accruals so
as not to sensitize participants to our underlying manipulation.
The NI_PRESENCE manipulation alters the information in Appendix A by removing the
“Net income from operations” subtotal from each firm’s earnings information for participants in
the NI_Absent condition. If the presence of the earnings subtotal leads to lower investment risk
11
judgments simply due to the disclosure of more data, we would expect to observe this decrease
for both firms (i.e., a main effect of NI_PRESENCE). If, however, only the disclosure of smooth
earnings lowers investors’ risk judgments, we would expect NI_PRESENCE to interact with our
FIRM manipulation to affect risk judgments. Therefore, our manipulation of NI_PRESENCE
enabled us to observe whether smooth earnings affect participants’ investment risk judgments.
2.2.3 Dependent Variables
We asked participants three questions concerning investment risk. First, we asked “...
(w)hich of the following two companies [Firm A or Firm B] do you believe presents greater risk
to investors?” Participants were next asked to provide a separate risk judgment for each firm.
Specifically, we asked, “... please rate your perception of the risk associated with investing in
[Firm A or Firm B].” Participants provided these risk judgments for Firm A and B on separate
13-point Likert scales ranging from “Low Risk” to “High Risk.”
2.3 EXPERIMENT ONE – RESULTS
2.3.1 Manipulation and Comprehension Checks
We asked participants several questions concerning key aspects of our experimental
design. Related to our NI_PRESENCE manipulation, we asked participants at the end of the
experiment, “... which of the following presentations of the components of net income did you
receive?” Fifty-five of the 58 (95%) participants correctly identified the presentation format they
had received. We also asked participants to evaluate the volatility of two number series
equivalent to those underlying operating cash flows and earnings in the experiment to ensure that
their beliefs concerning volatility were consistent with our expectations. Only one participant
incorrectly evaluated the relative volatility of these number series. Finally, we asked participants
to recall which firm had the more volatile operating cash flows to gauge the extent to which they
12
attended to the materials. Only eleven participants did not correctly answer this question. All of
our hypothesized results are robust to either treating these participants’ responses as covariates or
excluding them. Accordingly, we report all of our analyses for the full sample of participants.
2.3.2 Tests of H1 – Do Smooth Earnings Affect Investors’ Risk Judgments?
Our first hypothesis predicts that the presence of smooth earnings lowers investors’ risk
judgments. Table 1, Panel A indicates that participants’ average investment risk judgment for
Firm A (i.e., the firm with smooth earnings) was lower when earnings were present than when
they were absent (NI_Present: 6.700 vs. NI_Absent: 8.143; t56 = 2.27; p < 0.02, untabulated).7
Importantly, while participants’ risk judgments for Firm A decreased in the presence of the
earnings subtotal, we do not observe the same effect for participants’ risk judgments for the
control firm. Collectively, these findings suggest that the decrease in investors’ risk judgments
for Firm A in the presence of the earnings subtotal is not due to the presentation of more data.
Consistent with participants’ risk judgments, Table 1, Panel A also documents that significantly
fewer participants viewed Firm A as the riskier investment in the NI_Present condition (53%)
compared to participants in the NI_Absent condition (86%) (z = 2.38; p < 0.01, untabulated).
[INSERT TABLE 1 HERE]
Figure 1 illustrates the relative effect of the presence of the earnings subtotal on
participants’ investment risk judgments for both firms. Our analysis of variance test controls for
the alternative explanation that our results are driven by the presence of more information and
not smooth earnings per se. Given our first hypothesis, we expect to observe an interaction effect
7
NI_PRESENCE has an insignificant effect in the control conditions (i.e., Firm B). In these conditions, we find an
insignificant increase in participants’ investment risk judgments when earnings were present (NI_Absent: 5.679 vs.
NI_Present: 6.400; t56 = 1.26; p = 0.21, untabulated). While this increase is not statistically significant, it is in the
opposite direction of that expected if more data (i.e., the presence of the earnings subtotal) reduces risk judgments.
The (insignificant) increase in risk judgments is more consistent with participants in the NI_Present condition being
influenced by the volatility of Firm B’s earnings.
13
between NI_PRESENCE and FIRM. Consistent with H1, Table 1, Panel B documents a
significant interaction effect between these two variables on participants’ risk judgments (F =
8.10; p < 0.01). Controlling for the simple presence of more information, our results suggest that
the presence of smooth earnings lowers investors’ risk judgments independent from the effects of
the earnings components. However, participants still attended to the volatility of operating cash
flows as indicated by our participants judging the firms as roughly equally risky in the presence
of the earnings subtotal (Firm A: 6.700 vs. Firm B: 6.400; t29 = 0.51; p = 0.62, untabulated).
[INSERT FIGURE 1 HERE]
2.4 SUMMARY
Our findings with respect to our first research question suggest a negative relation
between smooth earnings and investment risk judgments. When earnings are smooth, we find
that participants judged lower investment risk when the earnings subtotal was present than when
this subtotal was absent. Our first experiment controls for the volatility of operating cash flows
and the amount of information disclosed. In summary, our findings support the hypothesis that
investors attend to smooth earnings when judging investment risk.
3. What Mechanism Drives Investors’ Risk Judgments when Earnings are Smooth?
3.1 HYPOTHESIS DEVELOPMENT
Hirshleifer and Teoh [2009] provide two possible explanations as to why investors might
perceive firms with smooth earnings as relatively less risky holding operating cash flows
constant. First, investors might limit their attention and inadvertently place greater weight on
earnings than on its components in arriving at their judgments (Hirshleifer and Teoh [2003]). In
effect, investors attend to the earnings subtotal in addition to the components of earnings when
14
judging investment risk. Empirical accounting research presents results consistent with investors
weighting earnings in addition to its underlying components (e.g., Sloan [1996]; Hewitt [2009]).
A second explanation provided by Hirshleifer and Teoh [2009] suggests that investors
believe that smoothed earnings present lower investment risk (‘conscious preference’). This
explanation embeds the idea that investors view smoothed earnings as having benefits beyond
smooth operating cash flows alone. In effect, investors perceive lower risk when managers use
their discretion to smooth earnings. This decrease in investment risk could stem from those
mechanisms discussed in the theoretical literature on earnings smoothing (Trueman and Titman
[1988]; Goel and Thakor [2003]) or reflect managers’ signaling of future firm performance.8
Ceteris paribus, the ‘limited attention’ and ‘conscious preference’ explanations both
suggest that smooth earnings lower investors’ risk judgments. However, the two explanations
provide different predictions when investors are aware that managers used reporting discretion to
smooth earnings. The ‘conscious preference’ explanation suggests that smooth earnings will
lower investors’ risk judgments even when investors are aware of the reporting discretion used
by managers. In contrast, the ‘limited attention’ explanation suggests that investors will no
longer lower their risk judgments when they are aware of the discretion used by managers. In
this instance, managers’ use of discretion leads investors to question the trustworthiness of
smooth earnings (Healy and Wahlen [1999]; Fields, Lys, and Vincent [2001]).
We argue that investors either assume that managers use little reporting discretion or are
unaware of this discretion, consistent with prior research documenting the effects of limited
8
There is a third explanation for an association between smooth earnings and investors’ risk judgments not
considered by Hirshleifer and Teoh [2009]. Investors may realize that the smoothness of operating cash flows, not
earnings, is associated with investment risk. However, they may consciously choose a heuristic strategy of using
earnings smoothness as a surrogate for cash flow smoothness if the reduced cognitive cost of using the more-readily
available earnings number is greater than the cost of judgment errors using this strategy. Given that acquiring
information concerning operating cash flows was at least as easy as acquiring aggregate earnings in our experiments,
we do not believe that this explanation explains our findings.
15
attention on investors’ forecasts (Hewitt [2009]) and trading decisions (Sloan [1996]). When
investors are aware of the discretion used by managers to smooth earnings, we expect them to
place less trust in this information and to shift their attention from the earnings subtotal to the
earnings components. Thus, our second hypothesis predicts that smooth earnings will have less
effect on risk judgments when investors are aware of the discretion used by managers.
H2:
Smooth earnings will have relatively less effect on investors’ risk judgments
when they are aware of the reporting discretion used by managers.
3.2 EXPERIMENT TWO
3.2.1 Design and Participants
We used a 2 x 2 between-subjects experimental design. Earnings smoothness
(NI_SMOOTHNESS) was manipulated between subjects by the reported accruals either (1)
smoothing earnings, or (2) ensuring that earnings reflected the same volatility as operating cash
flows. All participants were presented with the same operating cash flows for five years. We also
manipulated whether participants were presented with information concerning managers’ use of
reporting discretion. Half of the participants received information concerning managers’ use of
reporting discretion, while the remaining participants did not receive this information.
All participants were asked to judge investment risk after receiving information
concerning the earnings components as well as the earnings subtotal for five years.9 Similar to
our first experiment, participants were asked to provide risk judgments for the firm on a 13-point
Likert scale ranging from “Low Risk” to “High Risk.” We also asked participants to judge the
volatility of the firm’s operations as well as several questions about the firm’s managers. After
providing their judgments, participants were asked several questions including their perceptions
of managers’ use of reporting discretion, and whether operating cash flows or earnings provide
9
Appendix B illustrates the earnings information presented to participants in our second experiment.
16
the better measure of operating volatility. To participate in our second experiment, we recruited
73 M.B.A. students from the same student population as the first experiment.10
3.2.2 Information Concerning Discretion Used by Managers
We manipulated whether participants received information concerning the discretion
used by managers to report earnings (DISC_USED). Participants who received information about
the discretion used by managers to report earnings were all told that managers had discretion
available to them concerning the amount of accruals reported each year over the five-year period.
However, the total accruals over this period needed to sum to a particular amount. This
information reflects two key features of real-world settings: namely, that (1) managers have
discretion available to them when reporting earnings, and (2) managers’ use of reporting
discretion is constrained over the long-run due to the reversal of accruals. To ensure that the
effect of accruals was clear to participants, we reported accruals as a single line item in this
experiment. Participants were then informed whether managers had used their discretion to
smooth earnings (NI_Smooth) or to report their perception of the firm’s economic reality
(NI_Volatile). Participants who were not informed about the discretion used by managers when
reporting earnings were presented with only the earnings information.
3.3 EXPERIMENT TWO – RESULTS
3.3.1 Manipulation and Comprehension Checks
We asked participants several manipulation checks. All participants correctly evaluated
the relative volatility of two number series equivalent to those underlying earnings and its
10
The procedures used to recruit these participants as well as their characteristics were similar to those of our
participants in our first experiment. Approximately 25% of the participants were female, and most participants’
major area of study was either marketing (49%) or finance (33%). They had an average of 4.84 years of work
experience, and 12% of the participants had worked previously as financial analysts. On average, participants had
taken 3.07 and 4.56 undergraduate and graduate accounting/finance classes, respectively. Finally, 63% of the
participants had bought/sold securities, and 90% of all the participants planned to buy/sell securities in the future.
Our results are insensitive to this demographic information.
17
components in the experiment. Participants who were informed that managers used a ‘low’ level
of discretion provided significantly lower responses to the question soliciting their perception of
the discretion used by managers than participants that were informed that managers used a ‘high’
level of discretion (‘low’: 6.222 vs. ‘high’: 9.158; t35 = 2.76; p < 0.01, untabulated). Equally,
participants who were informed that managers used a ‘low’ level of discretion provided higher
responses to the question soliciting their perception of whether the firm’s earnings information
reflected economic reality than participants that were informed that managers used a ‘high’ level
of discretion (‘low’: 7.500 vs. ‘high’: 5.211; t35 = 2.61; p < 0.01, untabulated).
3.3.2 Tests of H2 – What Mechanism Drives Investors’ Risk Judgments?
H2 predicts that smooth earnings will have less effect on investors’ risk judgments when
they are informed of the discretion used by managers than when this information is absent.
Consistent with our first experiment, Table 2, Panel A indicates that participants’ risk judgments
were lower when they were presented with smooth earnings relative to when they were presented
with volatile earnings in the absence of information concerning the amount of discretion that
managers used (NI_Smooth: 5.944 vs. NI_Volatile: 7.833; t34 = 1.76; p = 0.04, untabulated).
However, when information concerning the discretion that managers used was present, we
observe no difference between participants’ risk judgments when they were presented with
smooth earnings as opposed to volatile earnings (NI_Smooth: 7.000 vs. NI_Volatile: 7.222; t35 =
0.24; p = 0.81, untabulated). Consistent with H2, participants did not weight smooth earnings in
their risk judgments when managers used a ‘high’ level of discretion to report earnings.11
[INSERT TABLE 2 HERE]
Figure 2, Panel A illustrates the predicted effects of earnings smoothness and disclosure
of the discretion used by managers on investors’ risk judgments when operating cash flows are
11
We observe the identical pattern of results for participants’ perceptions of operating volatility.
18
volatile. We hypothesize that investors will shift their attention from smooth earnings to the
volatility of the underlying operating cash flows when they are aware that managers used
discretion to smooth earnings. Therefore, we expect investors to perceive higher investment risk
when either (1) accruals are not used to smooth earnings (and hence earnings possess the same
volatility as operating cash flows), or (2) managers used reporting discretion to smooth earnings.
The only instance where we would expect investors’ risk judgments to be lower is when they are
not aware of the discretion used by managers to smooth earnings.
[INSERT FIGURE 2 HERE]
We formally test H2 using planned contrast analysis. Consistent with prior research, we
use planned contrast analysis to increase the power of the statistical test of the hypothesized
interaction (Bobko [1986]; Buckless and Ravenscroft [1990]; Rosnow and Rosenthal [1995]).
Given the hypothesized effects of earnings smoothness and disclosure of the discretion managers
used on investors’ risk judgments, we use the following contrast weights for each condition:
NI_Smooth / DISC_Absent = -3; NI_Volatile / DISC_Absent = +1; NI_Smooth / DISC_Present =
+1; and, NI_Volatile / DISC_Present = +1. Figure 2, Panel B documents the observed pattern of
the condition means. Table 2, Panel B provides evidence supporting H2 (F = 2.99; p = 0.04).12
3.3.3 Additional Analyses
We asked participants to rate their perception of the trustworthiness of the firm’s
managers. We find that participants’ perceptions of the managers’ trustworthiness is negatively
correlated with their perceptions of the discretion used by managers (r = -0.34; p < 0.01,
untabulated). In of itself, this finding refutes the ‘conscious preference’ explanation that
investors view managers favorably who use discretion.
12
Our findings are robust to alternative sets of contrast weights (e.g., -4,+1,+1,+2; -9,+2,+3,+4; -7,+1,+2,+4) that
reflect the predicted pattern, i.e., smooth earnings will have less effect on investors’ risk judgments when they are
informed of the discretion used by managers when reporting earnings than when this information is absent.
19
Our first experiment also provides evidence as to whether investors consciously prefer
firms with smoothed earnings. We asked participants whether they view operating cash flows or
earnings as a better measure of operating volatility (i.e., each participant’s conscious proxy for
investment risk).13 When the earnings subtotal was absent, 89% of the participants who chose
operating cash flows as the better measure of operating volatility selected the firm with the more
volatile operating cash flows (i.e., Firm A) as having higher investment risk. Essentially, these
participants’ choices were consistent with their beliefs. However, when the earnings subtotal was
present, only 68% of the participants who stated that operating cash flows is the better measure
of operating volatility selected Firm A as having higher investment risk. Therefore, the presence
of the earnings subtotal significantly reduced the likelihood of participants making risk choices
consistent with their stated beliefs (z = 1.64; p = 0.05, untabulated).
3.4 SUMMARY
Our second experiment’s findings suggest that investors do not have a conscious
preference for smoothed earnings. When made aware of managers’ use of discretion, our
participants no longer lowered their risk judgments when earnings were smooth. These results
are consistent with investors either assuming that managers use little reporting discretion or
being unaware of this discretion. Additional analyses suggest that investors are (1) less likely to
trust managers who use discretion to smooth earnings, and (2) unaware of the extent to which
they attend to smooth earnings when making risk judgments. Overall, the evidence presented in
this section fails to support the ‘conscious preference’ explanation provided by Hirshleifer and
Teoh [2009] and provides evidence consistent with the ‘limited attention’ explanation.
13
A vast majority of participants identified operating cash flows as the better measure of operating volatility
regardless of whether the earnings subtotal was absent (19 out of 28) or present (22 out of 30). These proportions do
not differ statistically (z = 0.17; p > 0.85), suggesting that the condition that participants were assigned to did not
influence their beliefs regarding which measure represents a better measure of operating volatility.
20
4. Do Disclosures about Available Discretion Reduce Attention to Smooth Earnings?
4.1 HYPOTHESIS DEVELOPMENT
Hirshleifer and Teoh [2009] question whether the impact of smooth earnings on
investors’ risk judgments is influenced by the reporting discretion available to managers. Our
second experiment suggests that investors do not consciously prefer firms with smoothed
earnings. That is, investors naïvely view smooth earnings as a favorable signal with respect to
investment risk and either assume that managers use little reporting discretion or are unaware of
this discretion. We find that directly informing investors of the discretion used by managers to
report earnings broadens investors’ attention to the earnings components. Such disclosures may
not always exist in real-world settings; however, disclosures revealing the extent of discretion
available to managers may help investors infer the likelihood of managers smoothing earnings.
Standard setters have increased the requirement for firms to issue disclosures concerning
the discretion available to managers when reporting earnings. For example, the Financial
Accounting Standards Board released SFAS 157 (Accounting Standards Codification Topic 820)
that requires disclosure of the amount of discretion available to managers when measuring fair
values (FASB [2006]). Such cues are likely to activate investors’ beliefs concerning the level of
discretion available to managers. However, they may not provide investors with a definitive
signal as to whether managers have used discretion to smooth earnings.
Our third experiment tests whether informing investors that the reporting discretion
available to managers is high broadens investors’ attention to the earnings components. We
argue that such disclosures shift investors’ attention to the volatility of operating cash flows. We
expect that investors will discern the likelihood that managers used discretion to smooth earnings
21
from both the earnings information and disclosures concerning available reporting discretion. We
argue that the effect of smooth earnings on investors’ risk judgments will decrease when
investors are aware of high available reporting discretion. Essentially, investors observe that
managers had the opportunity to smooth earnings (i.e., high available reporting discretion) as
well as the effects of this reporting behavior (i.e., smooth earnings).
H3:
Smooth earnings will have relatively less effect on investors’ risk judgments
when they are aware that the reporting discretion available to managers is ‘high’.
4.2 EXPERIMENT THREE
4.2.1 Design and Participants
Our third experiment used a 2 x 2 between-subjects design that manipulated the presence
of the earnings subtotal (NI_PRESENCE) and available reporting discretion (DISC_AVAIL). As
in our first experiment, NI_PRESENCE was manipulated by the earnings subtotal being either
present or absent. To manipulate DISC_AVAIL, we told participants that the managers had either
a lot of discretion (‘high’) or very little (‘low’) discretion available to them when measuring
changes in the firm’s operating assets and operating liabilities. Similar to the other experiments,
participants provided investment risk judgments and responded to a variety of other questions
(e.g., perceptions of the firm’s operating volatility). For this experiment we recruited 122 M.B.A
students from the same population of students who participated in the other experiments.14
4.2.2 Earnings Information
All participants were provided with Firm A’s earnings information (see Appendix A,
Panel A). As Firm A is characterized by volatile (smooth) operating cash flows (earnings), we
14
The procedures used to recruit this group of participants were similar to those in our first two experiments. Thirtyseven percent of the participants were female, and most participants’ major area of study was marketing (46%) or
finance (30%). The average work experience was 5.00 years, and 14% of the participants had worked as financial
analysts. On average, participants had taken 2.62 and 2.98 undergraduate and graduate accounting/finance classes,
respectively. Finally, 66% of the participants had bought/sold securities, and 94% of all the participants planned to
buy/sell securities in the future. Our results are insensitive to participants’ demographic information.
22
would expect participants’ risk judgments to be higher if they attended less to smooth earnings.
If both operating cash flows and earnings possess the same volatility (i.e., communicate the same
message about information risk), it is arguably (1) less important for investors to question
managers’ use of discretion, and (2) more difficult to discern empirically the information to
which investors attend when judging investment risk.
4.3 RESULTS
4.3.1 Manipulation and Comprehension Checks
At the end of the experiment, we asked participants several questions concerning key
aspects of our experimental design. Related to our NI_PRESENCE manipulation, 106 of the 122
(87%) participants correctly identified the format that they received. To observe the effect of the
DISC_AVAIL manipulation, we asked participants at the end of the experiment, “... how much
discretion did the company’s managers have available to them in measuring and recognizing
changes in the company’s operating assets and operating liabilities?” Participants responded on a
13-point Likert scale ranging from “Low Discretion” to “High Discretion.” Participants who
were informed that managers had a ‘low’ level of reporting discretion available to them provided
significantly lower responses to this question than participants who were informed that managers
had a ‘high’ level of discretion available to them (‘low’: 6.383 vs. ‘high’: 9.193; t120 = 5.33; p <
0.01, untabulated). Only four participants incorrectly evaluated the volatility of two number
series equivalent to those underlying operating cash flows and earnings in the experiment.15
4.3.2 Tests of H3 – Do Disclosures about Available Discretion Broaden Attention?
H3 predicts that smooth earnings will have less effect on investors’ risk judgments when
they are informed that the discretion available to managers is ‘high’ than when they are informed
15
Our results are insensitive to all of these manipulation and comprehension checks. We perform all analyses with
the full sample of participants.
23
that it is ‘low’. Consistent with our first two experiments, when participants were informed that
the discretion available to managers was ‘low’, Table 3, Panel A indicates that participants’ risk
judgments were lower when the earnings subtotal was present (NI_Present: 6.448 vs. NI_Absent:
8.129; t58 = 2.57; p < 0.01, untabulated). When participants were informed that the discretion
available to managers was ‘high’, we also observe a negative relation between smooth earnings
and investment risk judgments. However, we note that the relationship becomes weaker.
(NI_Present: 6.281 vs. NI_Absent: 7.387; t60 = 1.84; p = 0.07, untabulated).
[INSERT TABLE 3 HERE]
We hypothesize that investors will shift their attention from smooth earnings to the
volatility of the underlying operating cash flows when they are aware that managers had a lot of
discretion available to them. Therefore, we expect investors to perceive higher levels of
investment risk when either (1) the earnings subtotal was absent, or (2) managers had a lot of
reporting discretion available to them. The only instance where we would expect investors’ risk
judgments to be lower is when the discretion available to managers is low and the earnings
subtotal is present. Thus, similar to our second experiment, we test H3 using planned contrast
analysis. We use the following contrast weights for each condition: NI_Present / DISC_Low = 3; NI_Absent / DISC_Low = +1; NI_Present / DISC_High = +1; and, NI_Absent / DISC_High =
+1. Table 3, Panel B provides evidence supporting H3 (F = 2.46; p = 0.06) for participants’
investment risk judgments. Therefore, our results suggest that smooth earnings will have
relatively less effect on investors’ risk judgments when they are informed that managers had
‘high’ reporting discretion available to them. Nevertheless, we also find that smooth earnings
24
still lower risk judgments even when investors are made aware that managers had ‘high’
reporting discretion available to them.16
4.3.3 Additional Analysis – Perceptions of Available Reporting Discretion
To gain further insights into our third experiment, we conduct additional analyses
involving participants’ actual perceptions of the discretion available to managers, as opposed to
our manipulated disclosure relating to available discretion. Using a continuous measure of
participants’ perceptions of available reporting discretion, we regress participants’ investment
risk judgments on the presence of the earnings subtotal, perceptions of available reporting
discretion, and the interaction of these two variables. We observe a significantly negative
coefficient on both the presence of the earnings subtotal (coefficient = -4.39; t = 3.73; p < 0.01,
untabulated) and the amount of discretion available to managers (coefficient = -0.20; t = 1.95; p
= 0.05) suggesting that both of these variables decreased participants’ risk judgments in
isolation. As predicted, we also find a positive coefficient on the interaction term suggesting that
investors’ risk judgments were increased by higher perceptions of available reporting discretion
when the earnings subtotal was present (coefficient = 0.38; t = 2.71; p < 0.01, untabulated).
Interestingly, combining the main and interaction effects of participants’ perceptions of
available reporting discretion, we observe that the overall effect of the amount of discretion
available to managers in the presence of smooth earnings is significantly positive (coefficient =
0.18 [-0.20 + 0.38]; t = 1.88; p = 0.03, untabulated). Therefore, when the earnings subtotal was
present, participants increased their investment risk judgments as they perceived higher levels of
available reporting discretion. Overall, we find that participants’ risk judgments increased with
their perception of the reporting discretion available to managers when a separate subtotal was
16
Our findings with respect to participants’ perceptions of operating volatility are stronger. In this instance,
participants’ perceptions only differed when the reporting discretion available to managers was ‘low’.
25
provided for earnings, but decreased with their discretion perceptions in the absence of the
subtotal. These findings illustrate the combined role of investors’ perceptions of reporting
discretion and attention to smooth earnings in identifying potentially increased investment risk.
4.3.4 Additional Analysis – Perceptions of the Effect of Smooth Earnings on Firm Value
McInnis [2010] suggests that examining whether smooth earnings are priced involves
testing both the relation between smooth earnings and risk judgments, as well as the relation
between risk judgments and firm value. Our study primarily concerns the first relation; however,
our third experiment extends our analyses by explicitly asking participants to indicate the
implication of smooth earnings for firm value holding cash flows constant. Essentially, this
analysis considers whether investors’ perceptions of available reporting discretion moderate the
relation between smooth earnings and perceptions of the effect of smooth earnings on firm value.
Using participants’ perceptions of available reporting discretion, we identify a significant
interaction effect between the presence of the earnings subtotal and perceptions of available
reporting discretion on investors’ beliefs concerning the effect of smooth earnings on firm value
(coefficient = -0.48; t = 2.64; p = 0.01, untabulated). These results provide initial evidence that
investors’ perceptions of firm value diminish when disclosures suggest that managers used the
reporting discretion available to them to smooth earnings.
4.4 SUMMARY
This section documents several key findings. We provide evidence concerning the
interaction effect of the presence of smooth earnings and available reporting discretion on
investment risk judgments. Consistent with our first two experiments, the presence of smooth
earnings lowered participants’ risk judgments, particularly when participants were informed that
managers had a low level of reporting discretion available to them. These findings suggest that
26
participants in our first experiment either assume that managers have little discretion available to
them when reporting accruals or are unaware of this discretion. We find that disclosures
concerning the discretion available to managers reduce the extent to which investors attend to
smooth earnings when judging risk.
5. Conclusions
Prior archival research documents mixed evidence as to whether investors perceive that
firms with smooth earnings present lower risk (McInnis [2010]; Rountree et al. [2008]; Francis et
al. [2004]). Our study provides experimental evidence related to this question. Using
experimental methods that hold constant each firm’s operating cash flows, we present the
findings from three experiments. In each experiment, we find that participants attended to
smooth earnings when judging investment risk. Therefore, we document a negative relation
between smooth earnings and investment risk judgments.
Our first experiment examines whether the presence of the earnings subtotal lowers
investors’ risk judgments when earnings are smooth. We manipulate the presence of the earnings
subtotal and consider its effects when varying earnings smoothness. While the presence of the
earnings subtotal lowered participants’ risk judgments when earnings were smooth, we did not
observe the same effect when earnings were volatile. Our findings suggest that investors lower
their risk judgments when they attend to smooth earnings.
Given the existence of a relation between smooth earnings and investment risk
judgments, our second experiment examines the underlying mechanism driving this relation.
Hirshleifer and Teoh [2009] propose two explanations for the negative relation between smooth
earnings and investment risk judgments. By manipulating the discretion used by managers as
27
well as the accruals reported, we examine whether this relation stems from investors consciously
preferring smooth earnings or failing to fully attend to the likelihood that managers have used
discretion to smooth earnings. Our findings suggest that investors do not consciously prefer
smoothed earnings. Instead, we find that participants do not lower their risk judgments when
managers use their reporting discretion to smooth earnings. In the absence of information
concerning the reporting discretion used by managers, we observe a negative relation between
smooth earnings and investment risk judgments, supporting our conclusions from our first
experiment. Together these findings highlight that investors do not fully attend to the possibility
that managers might use their discretion to smooth earnings in the absence of information
concerning the discretion used by managers.
Building from our second experiment, our third experiment considers the role of
disclosures concerning available reporting discretion in mitigating the potentially adverse effects
of limited attention. We predict and find an interaction effect between disclosures concerning
available reporting discretion and the presence of smooth earnings on investment risk judgments.
However, we find that the presence of smooth earnings lowers investors’ risk judgments across
both levels of available reporting discretion. Therefore, our findings suggest that disclosures
concerning available reporting discretion do not completely eliminate the negative relation
between smooth earnings and investment risk judgments. We also provide evidence that the
presence of the earnings subtotal helps investors understand the implications of available
reporting discretion for smooth earnings and adjust their judgments accordingly.
Our study makes several contributions to the extant literature. First, all three of our
study’s experiments demonstrate that smooth earnings affect investors’ risk judgments holding
operating cash flows constant. Thus, we provide evidence regarding the debate in the archival
28
literature as to whether smooth earnings lower investors’ perceptions of investment risk. We also
provide evidence suggesting that investors’ limited attention, and not their conscious preference
for smoothed earnings, drives this relation. Responding to the call from Hirshleifer and Teoh
[2009] for research to consider the role of available reporting discretion on investors’ perceptions
of smooth earnings, we also document that investors’ perceptions of available reporting
discretion help mitigate the effects of limited attention that occur when investors attend to the
earnings subtotal. Our results suggest that factors which highlight managers’ reporting discretion
(e.g., disclosures concerning available and used discretion) likely limit the effectiveness of
earnings management in influencing investors’ risk assessments. Finally, our study adds to the
growing literature documenting how investors’ attention to earnings affects their judgments.
29
REFERENCES
ALLAYANNIS, G.; B. ROUNTREE, and J.P. WESTON. “Earnings Volatility, Cash Flow
Volatility, and Firm Value.” Working Paper (2005), University of Virginia.
BHATTACHARYA, U.; H. DAOUK; and M. WELKER. “The World Price of Earnings
Opacity.” The Accounting Review 78 (2003): 641-678.
BITNER, L.N., and R.C. DOLAN. “Assessing the Relationship between Income Smoothing and
the Value of the Firm.” Quarterly Journal of Business and Economics 35 (1996): 16-35.
BOBKO, P. “A Solution to Some Dilemmas when Testing Hypotheses about Ordinal
Interactions.” Journal of Applied Psychology 71 (1986): 323-326.
BUCKLESS, F.A., and S.P. RAVENSCROFT. “Contrast Coding: A Refinement of ANOVA in
Behavioral Analysis.” The Accounting Review 65 (1990): 933-945.
DEMSKI, J.S. “Performance Measure Manipulation.” Contemporary Accounting Research 15
(1998): 261-285.
DYE, R.A. “Earnings Management in an Overlapping Generations Model.” Journal of
Accounting Research 26 (1988): 195-235.
ELLIOTT, W.B.; F.D. HODGE; J.J. KENNEDY; and M. PRONK. “Are M.B.A. Students a
Good Proxy for Nonprofessional Investors?” The Accounting Review 82 (2007): 139-168.
FIELDS, T.D.; T.Z. LYS; and L. VINCENT. “Empirical Research on Accounting Choice.”
Journal of Accounting and Economics 31 (2001): 255-307.
FINANCIAL ACCOUNTING STANDARDS BOARD (FASB). Statement of Financial
Accounting Standards No. 157: Fair Value Measurements. Norwalk, CT: FASB, 2006.
FRANCIS, J.; R. LAFOND; P.M. OLSSON; and K. SCHIPPER. “Cost of Equity and Earnings
Attributes.” The Accounting Review 79 (2004): 967-1010.
GOEL, A.M., and A.V. THAKOR. “Why Do Firms Smooth Earnings?” Journal of Business 76
(2003): 151-192.
GRAHAM, J.R.; C.R. HARVEY; and S. RAJGOPAL. “The Economic Implications of
Corporate Financial Reporting.” Journal of Accounting and Economics 40 (2005): 3-73.
HEALY, P.M., and J.M. WAHLEN. “A Review of the Earnings Management Literature and its
Implications for Standard Setting.” Accounting Horizons 13 (1999): 365-383.
30
HEWITT, M. “Improving Investors’ Forecast Accuracy when Operating Cash Flows and
Accruals are Differentially Persistent.” The Accounting Review 84 (2009): 1913-1931.
HIRSHLEIFER, D., and S.H. TEOH. “Limited Attention, Information Disclosure, and Financial
Reporting.” Journal of Accounting and Economics 36 (2003): 337-386.
HIRSHLEIFER, D., and S.H. TEOH. “The Psychological Attraction Approach to Accounting
and Disclosure Policy.” Contemporary Accounting Research 26 (2009): 1067-1090.
HIRST, D.E.; P.E. HOPKINS; and J.M. WAHLEN. “Fair Values, Income Measurement,
and Bank Analysts’ Risk and Valuation Judgments.” The Accounting Review 79 (2004):
453-472.
HODDER, L.; P.E. HOPKINS; and D.A. WOOD. “The Effects of Financial Statement and
Informational Complexity on Analysts’ Cash Flow Forecasts.” The Accounting Review
83 (2008): 915-956.
HUNT, A.; S.E. MOYER; and T. SHEVLIN. “Earnings Volatility, Earnings Management, and
Equity Value.” Working Paper (1997), University of Washington.
LAMBERT, R.A. “Income Smoothing as Rational Equilibrium Behavior.” The Accounting
Review 59 (1984): 604-618.
LIPE, M.G. “Individual Investors’ Risk Judgments and Investment Decisions: The Impact of
Accounting and Market Data.” Accounting, Organizations and Society 23 (1998): 625640.
MCINNIS, J. “Earnings Smoothness, Average Returns, and Implied Cost of Equity Capital.” The
Accounting Review 85 (2010): 315-341.
NELSON, M.W., and W.B. TAYLER. “Information Pursuit in Financial Statement Analysis:
Effects of Choice, Effort, and Reconciliation.” The Accounting Review 82 (2007): 731758.
PINELLO, A.S. “Investors’ Differential Reaction to Positive versus Negative Earnings
Surprises.” Contemporary Accounting Research 25 (2008): 891-920.
RONEN, J., and S. SADAN. Smoothing Income Numbers: Objectives, Means, and Implications.
Reading, MA: Addison-Wesley, 1981.
RONEN, J., and V. YAARI. Earnings Management: Emerging Insights in Theory, Practice, and
Research. New York, NY: Springer, 2008.
ROSNOW, R.L., and R. ROSENTHAL. “Some Things You Learn Aren’t So: Cohen’s Paradox,
Asch’s Paradigm, and the Interpretation of Interaction.” Psychological Science 6 (1995):
3-9.
31
ROUNTREE, B.; J.P. WESTON; and G. ALLAYANNIS. “Do Investors Value Smooth
Performance?” Journal of Financial Economics 90 (2008): 237-251.
SLOAN, R.G. “Do Stock Prices Fully Reflect Information in Accruals and Cash Flows about
Future Earnings?” The Accounting Review 71 (1996): 289-315.
TRUEMAN, B., and S. TITMAN. “An Explanation for Accounting Income Smoothing.”
Journal of Accounting Research 26 (1988): 127-139.
32
APPENDIX A
Experiment One – Earnings Information (NI_Present Condition)
Panel A: Earnings Information for Firm A
Cash flows from operations
Accruals
Depreciation
Change in Operating Assets
Change in Operating Liabilities
Net income from operations
Year 1
Year 2
Year 3
Year 4
Year 5
3,914
6,732
4,646
7,326
5,454
(564)
952
758
5,060
(636)
(865)
136
(612)
841
824
(576)
243
(933)
(624)
729
922
5,367
5,699
6,060
6,481
Year 1
Year 2
Year 3
Year 4
Year 5
4,811
5,109
5,430
5,781
6,163
Panel B: Earnings Information for Firm B
Cash flows from operations
Accruals
Depreciation
Change in Operating Assets
Change in Operating Liabilities
Net income from operations
(547)
923
735
5,923
(617)
(839)
132
3,785
(594)
816
799
6,451
(559)
236
(905)
4,553
(605)
707
894
7,160
Appendix A presents the earnings information provided to participants in our first experiment. All participants were
provided with information for Firms A and B before being asked to separately judge the investment risk for each
firm in addition to providing a choice concerning which firm presents the higher investment risk. Firm A (Firm B) is
characterized by volatile (smooth) operating cash flows and smooth (volatile) earnings. Accruals are materially the
same across the two firms, i.e., Firm B’s accruals are 0.97 of Firm A’s reported accruals. Participants in the
NI_Absent condition received the same earnings information with the “Net income from operations” subtotal
removed from each firm’s information.
33
APPENDIX B
Experiment Two – Earnings Information
Panel A: NI_Volatile Conditions
Year 1
Year 2
Year 3
Year 4
Year 5
3,914
6,732
4,646
7,326
5,454
133
106
123
131
102
4,047
6,838
4,769
7,457
5,556
Year 1
Year 2
Year 3
Year 4
Year 5
Cash flows from operations
3,914
6,732
4,646
7,326
5,454
Accruals
1,146
(1,365)
1,053
(1,266)
1,027
Net income from operations
5,060
5,367
5,699
6,060
6,481
Cash flows from operations
Accruals
Net income from operations
Panel B: NI_Smooth Conditions
Appendix B presents the earnings information provided to participants in our second experiment. All participants
were provided with earnings information before being asked to judge the investment risk for the firm. Both sets of
earnings information are characterized by the same operating cash flows as well as the same average accruals and
earnings per year. However, the earnings information for the NI_Smooth conditions is characterized by accruals that
offset the volatility in operating cash flows. Accordingly, earnings are relatively smoother in Panel B relative to
Panel A.
34
FIGURE 1
Experiment One – Do Smooth Earnings Affect Investors’ Risk Judgments?
9
8
7
Firm A
Firm B
6
5
NI_PRESENCE NI_Absent
NI_Present
Figure 1 illustrates the observed effect of the presence of the earnings subtotal on participants’ risk judgments in our
first experiment.
35
FIGURE 2
Experiment Two – What Mechanism Drives Investors’ Risk Judgments?
Panel A: Predicted Effect of NI_SMOOTHNESS and DISC_USED on Risk Judgments
7.5
+1 +1 DISC_USED Absent
Present
6.5
‐3 5.5
NI_SMOOTHNESS NI_Volatile
NI_Smooth
Panel B: Observed Effect of NI_ SMOOTHNESS and DISC_USED on Risk Judgments
7.5
DISC_USED Absent
Present
6.5
5.5
NI_SMOOTHNESS NI_Volatile
NI_Smooth
Figure 2 illustrates the effect of the presence of the earnings subtotal on participants’ risk judgments in our second
experiment. Panel A shows the predicted effects and the corresponding contrast weights for the planned contrast
analysis used to test H2. Panel B shows the observed condition means.
36
TABLE 1
Experiment One – Do Smooth Earnings Affect Investors’ Risk Judgments?
Panel A: Descriptive Statistics
Investment Risk Judgments
n
% (Proportion)
Judging Firm A
with the Highest
Investment Risk
Firm A
Mean
(SD)
Firm B
Mean
(SD)
Difference
[Firm A-Firm B]
Mean (SD)
NI_PRESENCE
NI_Absent
28
86%
(24/28)
8.143
(2.534)
5.679
(2.161)
2.464
(2.472)
NI_Present
30
53%
(16/30)
6.700
(2.307)
6.400
(2.207)
0.300
(3.239)
-33%
-1.443
+0.721
-2.164
Difference
[NI_Present –
NI_Absent]
Panel B: Hypothesis Testing (Investment Risk Judgments)
Source of variance
Main effects:
NI_PRESENCE
FIRM
Two-way interaction effect:
NI_PRESENCE x FIRM
Hypothesis
1
df
F-statistic
p-value
1
1
0.59
13.21
0.45
< 0.01
1
8.10
< 0.01
Table 1 reports the descriptive statistics related to our first experiment. Experiment one manipulated NI_PRESENCE
between subjects. Participants in the NI_Absent condition were presented with earnings information but not the
earnings subtotal. Participants in the NI_Present condition were presented with earnings information including the
earnings subtotal. Participants were required to (1) provide investment risk judgments for each of the two firms, and
(2) choose which firm presented the higher risk to investors. Participants’ investment risk judgments were provided
on 13-point Likert scales ranging from “Low Risk” to “High Risk.” Firm A (Firm B) was characterized by volatile
(smooth) operating cash flows and smooth (volatile) earnings. The p-value for the hypothesized effect corresponds
to a one-tailed test. All other p-values correspond to two-tailed tests.
37
TABLE 2
Experiment Two – What Mechanism Drives Investors’ Risk Judgments?
Panel A: Descriptive Statistics (Investment Risk Judgments)
Discretion Used by Managers (DISC_USED)
‘Absent’
Mean (SD)
‘Present’
Mean (SD)
Difference
[Absent – Present]
7.833
(3.185)
n = 18
7.222
(2.625)
n = 18
+0.611
NI_Volatile
5.944
(3.244)
n = 18
7.000
(2.906)
n = 19
-1.056
NI_Smooth
-1.889
-0.222
-1.667
NI_SMOOTHNESS
Difference
[NI_Smooth – NI_Volatile]
Panel B: Hypothesis Testing (Investment Risk Judgments)
Source of variance
Main effects:
NI_SMOOTHNESS
DISC_USED
Two-way interaction effect:
NI_SMOOTHNESS x DISC_USED
Planned contrast (H2)
df
F-statistic
p-value
1
1
2.26
0.10
0.14
0.75
1
1.41
0.12
69
2.99
0.04
Table 2 reports the descriptive statistics related to our second experiment. Experiment two manipulated
NI_SMOOTHNESS and DISC_USED between subjects. Participants in the NI_Volatile condition were presented
with earnings information characterized by accruals that ensured that earnings reflected the same volatility as
operating cash flows. Participants in the NI_Smooth condition were presented with earnings information
characterized by accruals that smoothed the volatility in operating cash flows. We also manipulated whether
participants received information concerning managers’ use of discretion when reporting earnings (DISC_USED).
Participants were required to provide investment risk judgments. Participants’ judgments were provided on 13-point
Likert scales. With respect to our planned contrast analyses, we use the following contrast weights for each
condition: NI_Smooth / DISC_Absent = -3; NI_Volatile / DISC_Absent = +1; NI_Smooth / DISC_Present = +1; and,
NI_Volatile / DISC_Present = +1. p-values for the hypothesized effect correspond to one-tailed tests. All other pvalues correspond to two-tailed tests.
38
TABLE 3
Experiment Three – Do Disclosures about Available Discretion Reduce Investors’ Attention to
Smooth Earnings?
Panel A: Descriptive Statistics (Investment Risk Judgments)
Discretion Available to Managers (DISC_AVAIL)
‘Low’
Mean (SD)
‘High’
Mean (SD)
Difference
[Low – High]
7.387
(2.512)
n = 30
+0.742
NI_Absent
8.129
(2.363)
n = 31
6.281
(2.261)
n = 32
+0.167
NI_Present
6.448
(2.707)
n = 29
-1.681
-1.106
-0.575
NI_PRESENCE
Difference
[NI_Present – NI_Absent]
Panel B: Hypothesis Testing (Investment Risk Judgments)
Source of variance
Main effects:
NI_PRESENCE
DISC_AVAIL
Two-way interaction effect:
NI_ PRESENCE x DISC_AVAIL
Planned contrast (H3)
df
F-statistic
p-value
1
1
9.85
1.05
< 0.01
0.31
1
0.42
0.26
118
2.46
0.06
Table 3 reports the descriptive statistics related to our third experiment. Experiment three manipulated
NI_PRESENCE and DISC_AVAIL between subjects. Participants in the NI_Absent condition were presented with
earnings information but not the earnings subtotal. Participants in the NI_Present condition were presented with
earnings information including the earnings subtotal. We also manipulated whether participants were informed that
managers had a lot (‘high’) or a little (‘low’) reporting discretion available to them (DISC_AVAIL). Participants
were required to provide investment risk judgments. Participants’ judgments were provided on 13-point Likert
scales. With respect to our planned contrast analyses, we use the following contrast weights for each condition:
NI_Present / DISC_Low = -3; NI_Absent / DISC_Low = +1; NI_Present / DISC_High = +1; and, NI_Absent /
DISC_High = +1. p-values for the hypothesized effect correspond to one-tailed tests. All other p-values correspond
to two-tailed tests.
39