Description
The purpose of this study is to examine whether financial analysts mislead investors in
recognizing the differential persistence of the three cash flow components of earnings, defined by
Dechow et al., in forecasting annual earnings.
Accounting Research Journal
Do analysts mislead investors?: A comparison of analysts' and investors' weightings of
cash components in forecasting annual earnings
May H. Lo Le (Emily) Xu
Article information:
To cite this document:
May H. Lo Le (Emily) Xu, (2008),"Do analysts mislead investors?", Accounting Research J ournal, Vol. 21
Iss 1 pp. 33 - 54
Permanent link to this document:
http://dx.doi.org/10.1108/10309610810891337
Downloaded on: 24 January 2016, At: 21:06 (PT)
References: this document contains references to 16 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 486 times since 2008*
Users who downloaded this article also downloaded:
Mostafa M. Maksy, Lin Zheng, (2008),"Factors associated with student performance in advanced
accounting and auditing: An empirical study in a public university", Accounting Research J ournal, Vol. 21
Iss 1 pp. 16-32 http://dx.doi.org/10.1108/10309610810891328
Raymond da Silva Rosa, J ennifer Filippetto, Ann Tarca, (2008),"ASIC actions: canaries for
poor corporate governance?", Accounting Research J ournal, Vol. 21 Iss 1 pp. 67-86 http://
dx.doi.org/10.1108/10309610810891355
Gerry Gallery, Natalie Gallery, Matthew Supranowicz, (2008),"Cash-based related party
transactions in new economy firms", Accounting Research J ournal, Vol. 21 Iss 2 pp. 147-166 http://
dx.doi.org/10.1108/10309610810905935
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Do analysts mislead investors?
A comparison of analysts’ and investors’
weightings of cash components in forecasting
annual earnings
May H. Lo
School of Business, Western New England College,
Spring?eld, Massachusetts, USA, and
Le (Emily) Xu
Department of Accounting and Finance,
Whittemore School of Business and Economics,
University of New Hampshire, Durham, New Hampshire, USA
Abstract
Purpose – The purpose of this study is to examine whether ?nancial analysts mislead investors in
recognizing the differential persistence of the three cash ?ow components of earnings, de?ned by
Dechow et al., in forecasting annual earnings.
Design/methodology/approach – The paper uses Mishkin’s econometric approach to compare the
persistence of the cash ?ow components within and across the historical, analysts’ and investors’
weightings.
Findings – It is found that ?nancial analysts’ weightings of the cash ?ow components are more
closely aligned with the historical relations than are investors’ weightings, both in direction and in
magnitude. The degree of analysts’ mis-weighting is economically small and much lower than the
degree of investors’ mis-weighting. Moreover, the extent of both investors’ and analysts’
mis-weightings of the cash components is generally smaller for ?rms with greater levels of analyst
following, a proxy for the quality of the information environment.
Research limitations/implications – The ?ndings suggest that ?nancial analysts’ bias in
weighting the cash components of earnings is at best a partial explanation for investors’ bias.
Practical implications – This study is important to academics and the investment community that
relies upon ?nancial analysts as information intermediaries, because the ability of analysts to
incorporate value-relevant information in their published expectations may impact securities prices.
Originality/value – The study is the ?rst to document the weightings of the cash components of
earnings by ?nancial analysts. In addition, this paper provides evidence that ?nancial analysts, as
information intermediaries, are less biased than investors in processing not only the accrual but also
the cash components of earnings.
Keywords Financial analysis, Investors, Earnings, Financial forecasting
Paper type Research paper
1. Introduction
Sloan (1996) demonstrates that operating cash ?ows are more persistent than accruals, in
the sense that operating cash ?ows have a stronger linear relation with subsequent-year
earnings than do accruals. Securities prices, however, act as if investors’ earnings
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors acknowledge I/B/E/S International, Inc. for providing earnings per share forecast
data available through the Institutional Brokers Estimate System.
Forecasting
annual earnings
33
Accounting Research Journal
Vol. 21 No. 1, 2008
pp. 33-54
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309610810891337
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
expectations fail to re?ect the greater persistence of the cash ?ow component of current
earnings. Following Sloan (1996), several studies show that investors under-weight the
persistence of operating cash ?ows and over-weight the accrual component of earnings.
For example, Bradshaw et al. (2001) use ?nancial analysts’ forecasts of earnings as
empirical proxies for investors’ expectations of earnings. They ?nd that the subsequently
revealed analysts’ forecast bias is positively related to the current period’s earnings
accruals. They also document that the apparent investor bias is driven solely by the
working capital component of the earnings accruals.
In order to examine whether ?nancial analysts’ forecasts of annual earnings re?ect
an over-weighting of working capital accruals that is comparable to the mis-weighting
documented in Sloan (1996) and Bradshaw et al. (2001), Elgers et al. (2003) ?nd that the
over-weighting of working capital accruals in analysts’ earnings forecasts is less than
one-third of the over-weighting by investors that is implicit in stock prices. Moreover,
they are able to attribute less than 40 percent of the delayed securities returns
associated with working capital accruals to subsequent errors in analysts’ annual
earnings forecasts, for ?rms with less than median levels of analyst coverage. Their
?ndings suggest that securities market inef?ciencies that are unrelated to ?nancial
analysts’ earnings forecasts underlie at least part of the accruals-related anomaly.
Recently, Dechow et al. (2006) investigate the persistence and pricing of the cash
component of earnings, as distinct from the accrual-related anomaly. They decompose
the cash component of earnings into three sub-components:
(1) cash that is retained by the ?rm;
(2) cash that is distributed to equity holders as a result of equity ?nancing; and
(3) cash that is distributed to debt holders as a result of debt ?nancing.
They demonstrate that the higher persistence of the cash component (relative to
accruals) is entirely attributable to net cash distributions to equity holders. Moreover,
investors appear to correctly anticipate the lower persistence of cash distributed to
debt holders relative to cash distributed to equity holders. However, investors also
appear to overestimate the persistence of cash retained by the ?rm relative to cash
distributed to equity holders and debt holders.
To extend this line of inquiry, our study examines whether or not ?nancial analysts
contribute to investors’ mis-weighting of cash ?ow components. We ?rst determine
whether ?nancial analysts recognize the differential persistence of the three cash ?ow
components in forming their forecasts of annual earnings. We next compare investors’
and analysts’ weightings of cash components using the historical persistence as a
benchmark. Our aim is to assess whether ?nancial analysts’ bias, if any, in weighting
cash components may contribute to investors’ bias. Moreover, Dechow et al. (2006) do
not partition their sample according to the quality of the information environment,
although recent evidence suggests that richer information environments are associated
with more ef?cient securities pricing. This study examines the impact of the quality of
the information environment on both investors’ and analysts’ weightings of cash
components. We achieve this purpose by grouping the cases based on the level of
?nancial analyst coverage. In summary, the purpose of this paper is to evaluate:
(1) whether analysts’ weightings of the three cash components are similar in
direction and magnitude to their historical weightings;
ARJ
21,1
34
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(2) whether investors’ and analysts’ weightings (and bias, if any) are similar; and
(3) whether the level of sophistication of the information environment affects
investors’ and ?nancial analysts’ weightings of differential cash components.
We examine differential persistence from two aspects:
(1) relative persistence among the three cash ?ow components compared within
each of the historical, investors’ and analysts’ weightings; and
(2) absolute persistence of the three cash ?ow components compared across the
historical, investors’ and analysts’ weightings.
The benchmark for assessing persistence is based upon the historical linear relations
between realized earnings and prior-year earnings components (detailed model
speci?cation is indicated in the next section). The coef?cients obtained from the
historical relations are historical weightings, and the rankings of the historical
weightings based on their magnitudes are historical rankings. With respect to relative
persistence, we examine the relative weightings (or rankings) of cash components that
are implicit in analysts’ earnings forecasts as well as in securities’ prices. With respect
to absolute persistence, we focus on the magnitudes of ?nancial analysts’ weightings
and compare these weightings to both the historical and investors’ weightings.
The focus of the paper is on ?nancial analysts’ (in contrast to investors’) utilization
of the information contained in the three cash components. This inquiry is important to
academics as well as to the investment community that relies upon ?nancial analysts
as information intermediaries. As information intermediaries, ?nancial analysts play a
prominent role in the ?nancial market. Consequently, the ability of analysts to
incorporate value-relevant information in their published expectations may impact
securities prices. If we provide evidence that ?nancial analysts mis-weight the cash
?ow components in the same manner, as do investors, this would be consistent with
the interpretation that analysts mislead investors in weighting the cash components.
Such evidence would suggest that ?nancial analysts do not fully impound the
earnings-relevant information in historical accounting information, and that this
aspect of their behavior may contribute at least in part to market inef?ciency. On the
other hand, if ?nancial analysts’ mis-weightings of cash components are statistically
less than those of investors’, this evidence would suggest that analysts are relatively
more ef?cient in processing ?nancial information, and that either investors fail to
impound earnings-relevant information contained in the analysts’ forecasts, or other
factors contribute to investors’ mis-weightings of the cash ?ow components.
Our initial empirical results show that cash distributed to equity holders is the most
persistent among all three cash components of earnings, and that investors correctly
anticipate the relatively higher (lower) persistence of cash distributed to equity holders
(debt holders). These results are consistent with Dechow et al. (2006). We then
document that ?nancial analysts’ forecasts incorporate the differential persistence of
the cash components better than investors do. Speci?cally, ?nancial analysts’ rankings
of the cash components of earnings are more consistent with the historical rankings
than the corresponding rankings by investors. Furthermore, the degree of analysts’
mis-weighting is economically small and much lower than the degree of investors’
mis-weighting that is implied in security returns. Our ?ndings suggest that ?nancial
analysts do not mislead investors in recognizing the differential persistence of cash
Forecasting
annual earnings
35
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components in forecasting annual earnings. Analysts’ bias in weighting the cash
components is at best a partial explanation for investors’ bias. We also ?nd that
?nancial analysts’ coverage has an impact on the weightings of the cash components
by investors. Consistent with our expectations, investors’ overweighting of accruals,
cash retained by the ?rm, and cash distributed to equity holders are greater for the
lower analyst coverage group than for the higher analyst coverage group.
The evidence in this study contributes to the literature in several ways. First, we
document that cash distributed to equity holders is the most persistent among the three
cash components across different information environments. Using ?nancial analysts’
following as proxy for the quality of the information environment, we ?nd that the
“cash distributed to equity-holders” component is more persistent than the “cash
retained by the ?rm” or “cash distributed to debt-holders” components in both lower
and higher analyst coverage groups. This ?nding suggests that the quality of the
information environment is not associated with factors that affect the higher
persistence of cash distributed to equity holders. This result is consistent with the
?nding in Elgers et al. (2003) on the relative persistence of the operating cash ?ows,
working capital accrual and other accrual components. Second, we show that investors’
overweighting of the cash components of earnings, as documented by Dechow et al.
(2006), is conditioned by the quality of the information environment, for which the level
of analyst following is a proxy. Our results indicate that investors’ overweighting of
cash components is in general greater for the lower analyst coverage group.
Finally, this study is the ?rst to document the weightings of the cash components of
earnings by ?nancial analysts. Prior research has examined both investors’ and
analysts’ weightings of the accrual component of annual earnings. For example,
Bradshaw et al. (2001) ?nd that investors overweight working capital accruals, and
Elgers et al. (2003) document that ?nancial analysts’ overweighting of working capital
accruals is much less than the overweighting by investors. In terms of the weightings
of the cash components of annual earnings, Dechow et al. (2006) examine investors’
weightings as compared to historical relations. Our study ?lls a gap in this line of
research by investigating analysts’ weightings of the cash components as compared to
both historical relations and investors’ weightings. Such an examination allows us to
assess whether analysts mislead investors in estimating the persistence of the three
cash components. Our empirical results indicate that ?nancial analysts’ weightings of
the cash ?ow components are more closely aligned with the historical relations than
are the corresponding weightings by investors, both in direction and in magnitude.
Our ?ndings complement prior literature and suggest that ?nancial analysts, as
information intermediaries, are less biased than are investors in processing not only
the accrual but also the cash components of earnings.
The rest of the paper is organized as follows. Section 2 develops our main
hypotheses and describes the research design. Section 3 explains our sample selection
and the measurement of the analysis variables. Section 4 presents the empirical results
and Section 5 provides a summary and conclusions.
2. Hypotheses development
This paper addresses the research question: “do analysts mislead investors?” in the
context of recognizing the differential persistence of the cash components of annual
earnings. To pursue this question, we ?rst examine whether the weightings of cash
ARJ
21,1
36
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components implied in analysts’ earnings forecasts are consistent with the historical
relations. We then compare analysts’ and investors’ weightings using historical
persistence estimate as a benchmark. Following the approach in Dechow et al. (2006),
we decompose annual earnings (E) into total accruals (ACCR), changes in cash
(DCash), cash distributions to equity holders (DIST
EQ
), and cash distributions to debt
holders (DIST
D
)[1]:
E ¼ ACCR þDCash þ DIST
EQ
þ DIST
D
ð1Þ
As noted earlier, Dechow et al. (2006) document that the higher persistence of the cash
component of earnings is attributable to cash distributions to equity holders, and that
investors tend to overestimate the persistence of cash retained by the ?rms, although
they correctly anticipate the relatively higher (lower) persistence of cash distributed to
equity (debt) holders. That evidence suggests that investors fail to fully recognize the
differential persistence of the cash components. As information intermediaries,
?nancial analysts presumably in?uence investors’ expectations. A primary aim of this
paper is to assess to what extent analysts’ bias contributes to investors’ bias in
weighting these cash components.
Prior literature has shown that in some contexts ?nancial analysts utilize ?nancial
information more effectively than investors do, and that market mispricing appears to
re?ect bias in investors’ earnings expectations that is more pronounced than any
related bias in ?nancial analysts’ forecasts. For example, Bernard and Thomas (1989,
1990) report a failure of stock prices to re?ect fully the implications of current quarterly
earnings for future earnings. Subsequently, Abarbanell and Bernard (1992) examine
whether ?nancial analysts’ under-reaction to prior earnings information could explain
these anomalous securities returns, and ?nd that under-reactions in analysts’
forecasts are only about half as large as necessary to explain the lagged securities price
adjustments. Elgers et al. (2003) ?nd that the over-weighting of working capital
accruals in analysts’ earnings forecasts is less than one-third of the over-weighting by
investors that is implicit in stock prices. Frankel and Lee (1998) develop pro?table
trading rules based on the ratio of “intrinsic value” (measured using a variant of
Ohlson’s (1995) residual income valuation model) to market value. The intrinsic value
measures are based upon ?nancial analysts’ earnings forecasts and growth
predictions, implying that investors do not ef?ciently incorporate these analysts’
expectations in securities prices. Similarly, Elgers et al. (2001) report delayed securities
price adjustments to share-price scaled analysts’ earnings forecasts. Collectively, these
studies suggest that in some contexts ?nancial analysts suffer less from information
inef?ciencies than do investors.
The above evidence leads us to hypothesize that ?nancial analysts are better able
than investors to recognize the differential persistence of various cash ?ow components
in forming their forecasts of annual earnings. We examine differential persistence from
two aspects: relative persistence among the three cash ?ow components compared
withineachof the historical investors’ and analysts’ weightings andabsolute persistence
of the three cash ?ow components compared across the historical, investors’ and
analysts’ weightings. With respect to relative persistence, we examine the relative
weightings (or rankings) of these cash components that are implicit in analysts’ earnings
forecasts as well as in securities’ prices. We expect that ?nancial analysts rank the
persistence of the cash components in a manner that is more consistent with their
Forecasting
annual earnings
37
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
historical rankings than do investors. With respect to absolute persistence, we focus on
the magnitudes of ?nancial analysts’ weightings and compare these weightings to both
the historical weightings and investors’ weightings. Speci?cally, we expect that
?nancial analysts’ weightings of the cash components are more consistent with their
historical weightings than the weightings by investors. Formally, we state our ?rst two
hypotheses as follows:
H1. The rankings of the persistence of the three cash ?ow components by
?nancial analysts are more consistent with the historical rankings than are
the corresponding rankings by investors.
H2. The magnitudes of the weightings of the three cash ?ow components by
?nancial analysts are more consistent with the historical weightings than are
the corresponding weightings by investors.
To evaluate whether ?nancial analysts mislead investors in weighting the cash
components in forecasting annual earnings, this study compares historical relations,
analysts’ weightings, and investors’ weightings. A ?nding that analysts perform better
than investors in weighting the cash ?ow components would complement the prior
literature and suggest that:
(1) analysts are more ef?cient than investors in processing not only accruals but
also cash ?ow information; and
(2) investors’ mis-weightings of the cash ?ow components are not caused entirely
by analysts’ bias.
Recent evidence indicates that ?nancial analyst coverage is positively associated with
the ef?ciency of investors’ information usage. For example, Walther (1997) reports that
investors in the securities of lightly followed ?rms over-rely on time-series predictions
of earnings, relative to ?nancial analysts’ earnings forecasts. Similarly, Bhattacharya
(2001) ?nds that, for ?rms with little to moderate analyst following, trading around
earnings announcements is more closely related to seasonal random-walk forecast
errors than to analysts’ forecast errors. Hong et al. (2000) ?nd that momentum-based
investment strategies are substantially more pro?table for ?rms with low analyst
coverage, because the initial under-reaction to value-relevant information, and
subsequent returns momentum, are stronger for these ?rms. Elgers et al. (2001)
document delayed security price adjustments to the information in ?nancial analysts’
earnings forecasts, and that the magnitudes of the delayed abnormal returns are larger
for ?rms with lower analyst coverage. These studies provide persuasive evidence that
analyst coverage is likely to be an effective partitioning variable. Given the premise
that there is more competition in information acquisition for ?rms with higher analyst
coverage, the quality of the information environment for highly followed ?rms is
expected to be better. Hence investors’ and analysts’ earnings forecasts for such ?rms
are expected to be more informed, or less biased. Moreover, to the extent that there is a
greater prevalence of less-sophisticated investors for ?rms with lower analyst
following, there is likely to be additional security pricing inef?ciency by investors,
beyond that induced by analysts’ forecasts. Accordingly, we adopt analyst coverage as
a partitioning device and formulate the following hypothesis:
ARJ
21,1
38
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H3. The level of ?nancial analyst coverage affects ?nancial analysts’ and
investors’ rankings of the persistence of the cash ?ow components.
H4. The level of ?nancial analyst coverage affects the magnitudes of ?nancial
analysts’ and investors’ weightings of the cash ?ow components.
Based on the evidence in prior research, we expect that the mis-weightings of the cash
components of earnings by investors or analysts will vary inversely with the level of
analyst coverage.
3. Research design
3.1 Sample selection and measurement of variables
The empirical analysis uses all December ?scal-year ?rm-years available from the
intersection of the I/B/E/S US Summary data base (consensus forecasts), the CRSP
monthly stock return data base and the Compustat annual data base (active and
research companies) with suf?cient information to measure the following required
variables. The speci?cations of the earnings, accruals, the cash ?ows components,
securities returns and analysts’ forecasts measures are indicated below (Compustat
item numbers are in parentheses and ?rm speci?c subscripts are omitted):
E
tþ1
¼ Annual income before extraordinary items and discontinued operations
available for common stockholders (A18) for year t þ 1.
ACCR
t
¼ Total annual accruals for year t.
¼ DNon-Cash Assets 2 DNon-Debt Liabilities
Where Non-Cash Assets ¼ Total assets (A6) 2 Cash & equivalents (A1)
Non-Debt Liabilities ¼ Total liabilities (A181) 2 Debt (A9 þ A34).
DCASH
t
¼ Change in cash for year t. Cash is de?ned as cash and equivalents (A1).
DIST
EQ
t
¼ Cash distribution to equity holders for year t.
¼ E
t
2 Change in equity for year t.
¼ E
t
(DTotal Assets 2 DTotal Liabilities) for year t.
DIST
D
t
¼ Cash distributions to debt holders for year t.
¼ Reduction in debts for year t.
¼ Reduction in long-term debt (A9) þ Reduction in short-term debt (A34),
for year t.
SAR
tþ1
¼ Size-adjusted security return, measured as the realized market return in
year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the
corresponding median return for all Compustat ?rms in the same
market capitalization decile at the start of year t þ 1.
FAF
tþ1
¼ I/B/E/S median (consensus) analyst forecast of annual earnings,
reported in May of the following year, multiplied by shares outstanding
and scaled by average total assets.
Because the start of the cumulating period for size-adjusted securities returns, and
hence the date at which the implicit securities market earnings expectations are
measured, is May 1 of the earnings year, we use the May consensus forecasts to ensure
comparability between analysts’ and market’s implicit earnings expectations.
All earnings and components variables are scaled by average total assets, as in
Sloan (1996), Bradshaw et al. (2001) and Elgers et al. (2003).
Forecasting
annual earnings
39
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
To enter our sample, we require ?rm-years to have complete data during our
analysis period for each of the variables described above. Speci?cally, we ?rst
eliminate ?rm-years with missing required data from COMPUSTAT and CRSP. This
screen yields 94,760 observations over the time span 1952-2004. We then eliminate
?rm-years with missing analysts’ forecasts of annual earnings from I/B/E/S, yielding
34,731 ?rm-year observations over the time span 1976-2004. Because I/B/E/S analyst
data was sparse prior to year 1985, we eliminate cases between 1976 and 1984 in order
to minimize self selection bias. Our ?nal sample contains 34,205 ?rm-years over the
sample period 1985-2004. Lastly, to mitigate the potential effects of outliers on our
analyses, we winsorize the extreme cases for the asset-scaled earnings components
variables at þ1 and 21 in each year, as used in Dechow et al. (2006).
3.2 Empirical models
To examine the historical persistence of earnings components, we relate realized
earnings to the earnings components from the previous year and estimate the following
model:
E
tþ1
¼ a
0
þa
1
ACCR
t
þa
2
DCASH
t
þa
3
DIST
EQ
t
þa
4
DIST
D
t
þm
tþ1
ð2Þ
where E
tþ1
is year t þ 1 earnings, ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
are the
accruals, change in cash, cash distributions to equity holders, and cash distributions to
debt holders, respectively, of prior-year’s earnings. The parameter estimates for the three
cash components of earnings (i.e.
_
a
i
, i ¼ 2; 3 and 4) indicate the historical persistence or
weightings of these cash components. These historical weightings are the benchmarks
against which the investors’ and ?nancial analysts’ weightings are compared.
We next examine investors’ weighting of these prior-year earnings components in
their earnings expectations. Because the earnings expectations impounded in securities
prices cannot be directly observed, our analysis relies on relations of cash components
and accruals to investor earnings expectations that are inferred from empirical
relations between abnormal security returns and “unexpected” earnings. We follow the
approach developed in Mishkin (1983) and subsequently adapted by Sloan (1996)
to assess the market’s weighting of earnings components. Following this general
approach, we estimate the historical linear relations of realized earnings, E
tþ1
, to
prior-year earnings components, as speci?ed in equation (2). To measure the implicit
linear weighting of these four earnings components in investor’s earnings expectations,
we specify the relation between abnormal returns and unexpected earnings as follows:
SAR
tþ1
¼ d
0
þd
1
UE
tþ1
þm
tþ1
ð3Þ
where SAR
tþ1
represents abnormal (size-adjusted) security returns, a commonly used
measure of unexpected returns (e.g. Abarbanell and Bushee, 1998; Elgers et al., 2003),
and UE
tþ1
represents unexpected earnings in year t þ 1. The unexpected earnings
variable, UE
tþ1
is then decomposed into realized earnings, E
tþ1
and expected earnings,
E(E
tþ1
):
UE
tþ1
¼ E
tþ1
2EðE
tþ1
Þ
we substitute the linear prediction of E
tþ1
in equation (2) for expected earnings E(E
tþ1
)
in equation (3), and rearrange equation (3) as follows:
ARJ
21,1
40
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
SAR
tþ1
¼ d
0
þd
1
½E
tþ1
2EðE
tþ1
Þ? þm
tþ1
ð4aÞ
¼d
0
þd
1
E
tþ1
2d
1
_
g
0
þ
_
g
1
ACCR
t
þ
_
g
2
DCASH
t
þ
_
g
3
DIST
EQ
1
þ
_
g
4
DIST
D
t
þm
tþ1
ð4bÞ
¼f
0
þd
1
E
tþ1
þf
1
ACCR
t
þf
2
DCASHþf
3
DIST
EQ
t
þf
4
DIST
D
t
þm
tþ1
ð4cÞ
where the parameter estimates
_
g
i
(i ¼ 1, 2, 3 and 4) represent the slope parameters of
the linear relation of the earnings components to the security market’s expectations,
E(E
tþ1
). In other words, they indicate investor’s implicit weightings of prior periods’
earnings components in forecasting earnings. Empirical estimation of equation (4c)
yields estimates of d
1
and f
i
. We then de?ne investors’ weightings as
_
g
i
¼2
^
f
i
=
_
d
1
. If
investors utilize the information contained in these earnings components
effectively, then investors’ weightings of earnings components in equation (4b)
would be consistent with the historical weightings in equation (2), i.e.
_
g
i
¼
_
a
i
. We use
the following equation to indicate investors’ weighting model hereafter to simplify the
expression:
SAR
tþ1
¼d
0
þd
1
E
tþ1
2g
0
2g
1
ACCR
t
2g
2
DCASH
t
2g
3
DIST
EQ
t
2g
4
DIST
D
t
þm
tþ1
ð5Þ
To examine the weightings of the prior-year earnings components by ?nancial
analysts, we regress ?nancial analysts’ earnings forecasts on the accrual and cash
components of earnings from the previous year and estimate the following model:
FAF
tþ1
¼b
0
þb
1
ACCR
t
þb
2
DCASH
t
þb
3
DIST
EQ
t
þb
4
DIST
D
t
þm
tþ1
ð6Þ
where FAF
tþ1
is analyst forecasts of annual earnings in year t þ 1. This model is
parallel to the historical relation in equation (2). The parameter estimates
_
b
i
(i ¼ 1, 2, 3
and 4) indicate analysts’ weightings of prior-year earnings components implied in their
earnings forecasts. To test our ?rst two hypotheses, we compare three sets of weightings
of the cash components, i.e. the historical persistence (the coef?cient estimates of a
2
, a
3
,
and a
4
in expression (2)), the investors’ weighting (the coef?cient estimates of g
2
, g
3
, and
g
4
in expression (5)), and the analysts’ weighting (the coef?cient estimates of b
2
, b
3
, and
b
4
in expression (6)) both in direction and in magnitude.
Speci?cally, to test H1, we examine the rankings of the cash ?ow components by
both investors and analysts, and compare their rankings to the historical rankings
observed from expression (2). To test H2, we formally contrast investors’ and analysts’
weightings to historical weightings separately for the three cash ?ow components.
equations (2), (5) and (6) are also estimated separately for both the lightly and closely
followed ?rms, in order to examine how the quality of the information environment
(for which analyst coverage is a proxy) affects investors’ and analysts’ weightings of
the cash components (H3 and H4).
Following Sloan (1996) and Dechow et al. (2006), we use Mishkin’s (1983)
econometric approach, i.e. simultaneous non-linear least squares regression, that
allows us to simultaneously estimate the historical persistence of the accrual and cash
Forecasting
annual earnings
41
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components of earnings (in expression (2)), the investors’ weightings of the
corresponding components (in expression (5)) and the ?nancial analysts’ weightings
of the corresponding components (in expression (6)). For more detailed discussion
about this framework, see Mishkin (1983) and Sloan (1996).
4. Empirical results
We organize our empirical results as follows. Section 4.1 presents descriptive statistics
for the analysis variables. Section 4.2 discusses historical, investors’ and ?nancial
analysts’ weightings of prior-year accruals and cash ?ow components. Section 4.3
presents the results from the comparisons of historical, investors’ and analysts’
weightings of the cash ?ow components. Section 4.4 describes the weighting results for
?rm-years in different ?nancial analyst coverage groups.
4.1 Descriptive statistics
Table I presents the univariate statistics and Pearson correlations between the key
variables for the ?rm-years included in the subsequent analysis. These descriptive
statistics re?ect various regularities reported in recent research. The univariate
statistics for the earnings components variables display the same characteristics shown
by Dechow et al. (2006): positive means for both Accruals, ACCR
t
, and Change in cash,
DCASH
t
, indicating that our sample ?rms have been growing during our sample years;
negative means for DIST
EQ
t
and DIST
D
t
, indicating that the amount of capital raised by
those ?rms from their capital holders is more than the amounts distributed to capital
holders. The standard deviations of individual components of earnings show that each
Mean SD
Pearson correlations
ACCR
t
DCASH
t
DIST
EQ
t
DIST
D
t
SAR
tþ1
FAF
tþ1
E
t
0.017 0.163 0.231 0.217 0.226 0.128 20.043 0.581
ACCR
t
0.081 0.198 1 0.028 20.358 20.521 20.082 0.071
DCASH
t
0.040 0.195 1 20.659 20.008 20.060 20.041
DIST
EQ
t
20.075 0.249 1 20.085 0.062 0.330
DIST
D
t
20.030 0.147 1 0.037 0.046
SAR
tþ1
0.024 0.652 1 20.001
FAF
tþ1
0.028 0.170 1
Notes: (All earnings and earnings components variables are scaled by average total assets. Compustat
item numbers are in parentheses and ?rm speci?c subscripts are omitted): E
t
, annual income before
extraordinary items and discontinued operations available for common stockholders (A18) for year t;
ACCR
t
, total annual accruals for year t ¼ DNon-Cash Assets 2 DNon-Debt Liabilities, where
Non-Cash Assets ¼ Total assets (A6) 2 Cash and equivalents (A1), and Non-Debt Liabilities ¼ Total
liabilities (A181) 2 Debt (A9 þ A34); DCASH
t
, change in cash for year t. Cash is de?ned as cash and
equivalents (A1); DIST
EQ
t
, cash distributions to equity holders for year t ¼ E
t
2 Change in equity for
year t ¼ E
t
2 (DTotal Assets 2 DTotal Liabilities) for year t; DIST
D
t
, cash distributions to debt
holders for year t ¼ Reduction in debts for year t ¼ Reduction in long-term debt (A9) þ Reduction in
short-term debt (A34), for year t; SAR
tþ1
, size-adjusted security return, measured as the realized
market return in year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the corresponding
median return for all Compustat ?rms in the same market capitalization decile at the start of year t þ 1;
FAF
tþ1
, I/B/E/S median (consensus) analyst forecast of annual earnings for year t þ 1, reported in
May of year t þ 1, multiplied by shares outstanding and scaled by average total assets
Table I.
Descriptive statistics
(34,205 ?rm-years,
1985-2004)
ARJ
21,1
42
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
component, i.e. ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
, represents an important source of
the variation in earnings.
Pearson correlations reported in Table I are also consistent with prior studies. For
example, size-adjusted returns, SAR
tþ1
, and prior-year accruals, ACCR
t
, are negatively
correlated (with a correlation coef?cient of 20.082), consistent with the lagged security
price adjustments to accruals reported in Sloan (1996), Bradshaw et al. (2001) and Elgers
et al. (2003). The correlation between accruals, ACCR
t,
and cash distributed to equity
holders, DIST
EQ
t
, as well as to debt holders, DIST
D
t
, are negative (with a correlation
coef?cient of 20.358 and 20.521, respectively), consistent with the role of accruals in
mitigating timing problems in cash ?ow measures of earnings (Dechow, 1994). Not
surprisingly, DCASH
t
and DIST
EQ
t
are also negatively correlated (with a correlation
coef?cient of 20.659), since cashdistributions toequityholders consume a ?rm’s free cash.
4.2 Historical, investors’ and analysts’ weightings of prior-year cash components
This section presents comparisons of the estimated linear relations of accruals and
cash ?ow components to subsequent-year realized earnings, investor expectations of
earnings inferred from returns/earnings regressions, and ?nancial analysts’ forecasts
of earnings. We ?rst examine the historical persistence of earnings components. This is
followed by deriving inferred investors’ weightings of the earnings components.
Market ef?ciency requires that the weightings of earnings components by investors
re?ect the historical weightings without bias. Finally we investigate the weightings of
the earnings components by ?nancial analysts and compare investors’ and analysts’
weightings, in order to test our ?rst hypothesis (H1).
Panel A of Table II presents the estimated historical persistence of accruals and
cash ?ow components from equation (2). The historical results correspond closely to
those reported by Dechow et al. (2006). The results show that DIST
EQ
t
has the highest
persistence among the three cash components. The persistence coef?cient for DIST
EQ
t
(0.709) is signi?cantly higher than those for DCASH
t
(0.526) and DIST
D
t
(0.517). There
is, however, no signi?cant difference in the historical persistence between DCASH
t
and
DIST
D
t
. Notice that the persistence coef?cients for DCASH
t
and DIST
D
t
are both
similar to the persistence coef?cient for ACCR
t
(0.503), which suggest that the higher
persistence of cash ?ows vis-a`-vis accruals documented in prior literature is entirely
driven by cash distributed to equity holders. These historical persistence measures will
be used as benchmarks to examine the corresponding weightings by ?nancial analysts
and investors.
Panel B of Table II reports the results from estimating equation (5) to determine
investors’ weightings of the accruals and the cash components in their earnings
expectations. Descriptively, the results in Panel B indicate that investors appear to
weight ACCR
t
(DIST
D
t
) highest (lowest) among the four earnings components,
consistent with Dechow et al. (2006). The test of equality of the coef?cients of DIST
EQ
t
and DIST
D
t
shows that investors’ weighting of DIST
EQ
t
(0.891) is signi?cantly higher
than that of DIST
D
t
(0.677), suggesting that investors correctly anticipate the higher
persistence of cash distributed to equity holders relative to cash distributed to debt
holders. This result is consistent with the historical relations shown in Panel A of
Table II. Contrary to the historical relations, however, investors’ weighting of DCASH
t
(0.966) is signi?cantly higher than their weighting of DIST
EQ
t
(0.891), suggesting that
investors overestimate the persistence of DCASH
t
[2]. Investors perceive DCASH
t
as
Forecasting
annual earnings
43
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the most persistent cash component, while the historical relations indicate that DIST
EQ
t
is the most persistent. In this respect, investors’ rankings of the cash components fail to
fully re?ect the historical rankings[3].
Panel C of Table II presents ?nancial analysts’ weightings of accruals and cash
components obtained from estimating equation (6). The estimation results show that
the coef?cient for DIST
EQ
t
(0.694) is signi?cantly greater than those for DIST
D
t
(0.564)
and DCASH
t
(0.534). This contrast indicates that ?nancial analysts weight the cash
component of earnings that is distributed to equity holders more heavily than either of
the other cash components. These directional results are consistent with the historical
relations reported in Table II, Panel A. Judging from the rankings of the persistence of
various cash ?ow components, ?nancial analysts appear to fully recognize their
differential persistence. Both the historical rankings and the analysts’ rankings of the
Panel A: Historical relations of realized earnings to the accrual and cash ?ow components of earnings
E
tþ1
¼ a
0
þa
1
ACCR
t
þa
2
DCASH
t
þa
3
DIST
EQ
t
þa
4
DIST
D
t
þm
tþ1
ð2Þ
Historical weightings Comparison of weightings
(a)
a
1
a
2
a
3
a
4
a
2
¼ a
3
a
2
¼ a
4
a
3
¼ a
4
Coef?cient 0.503 0.526 0.709 0.517 Difference –0.183 0.009 0.192
Standard error 0.006 0.006 0.005 0.007 Likelihood ratio 2216.1 1.52 960.78
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.000) (0.217) (0.000)
Panel B: Investors’ weightings of the accrual and cash ?ow components of prior-year earnings
SAR
tþ1
¼ d
0
þd
1
ðE
tþ1
2g
0
2g
1
ACCR
t
2g
2
DCASH
t
2g
3
DIST
EQ
t
2g
4
DIST
D
t
Þ þm
tþ1
ð5Þ
Investors’ weightings Comparison of weightings
(a)
g
1
g
2
g
3
g
4
g
2
¼ g
3
g
2
¼ g
4
g
3
¼ g
4
Coef?cient 1.072 0.966 0.891 0.677 Difference 0.075 0.289 0.214
Standard error 0.042 0.041 0.034 0.045 Likelihood ratio 7.96 37.21 26.40
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.005) (0.000) (0.000)
Panel C: Financial analysts’ weightings of the accrual and cash ?ow components of prior-year earnings
FAF
tþ1
¼ b
0
þb
1
ACCR
t
þb
2
DCASH
t
þb
3
DIST
EQ
t
þb
4
DIST
D
t
þm
tþ1
ð6Þ
Financial Analysts’ weightings Comparison of weightings
(a)
b
1
b
2
b
3
b
4
b
2
¼ b
3
b
2
¼ b
4
b
3
¼ b
4
Coef?cient 0.577 0.534 0.694 0.564 Difference –0.160 –0.030 0.130
Standard error 0.005 0.005 0.005 0.007 Likelihood ratio 1753.7 18.51 456.95
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.000) (0.000) (0.000)
Notes: (All earnings and earnings components variables are scaled by average total assets.
Compustat item numbers are in parentheses and ?rm speci?c subscripts are omitted): E
tþ1
(E
t
), Annual
income before extraordinary items and discontinued operations available for common stockholders
(A18) for year t þ 1 (t); ACCR
t
, total annual accruals for year t ¼ DNon-Cash Assets 2 DNon-Debt
Liabilities, where Non-Cash Assets ¼ Total assets (A6) 2 Cash and equivalents (A1) and Non-Debt
Liabilities ¼ Total liabilities (A181) 2 Debt (A9 þ A34); DCASH
t
, change in cash for year t. Cash is
de?ned as cash and equivalents (A1); DIST
EQ
t
, cash distributions to equity holders for year t ¼
E
t
2Change in equity for year t ¼ E
t
2 (DTotal Assets 2 DTotal Liabilities) for year t; DIST
D
t
, cash
distributions to debt holders for year t ¼ Reduction in debts for year t ¼ Reduction in long-term debt
(A9) þ Reduction in short-term debt (A34), for year t; FAF
tþ1
, I/B/E/S median (consensus) analyst
forecast of annual earnings, reported in May of the following year, multiplied by shares outstanding
and scaled by average total assets; SAR
tþ1
, size-adjusted security return, measured as the realized
market return in year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the corresponding
median return for all Compustat ?rms in the same market capitalization decile at the start of year
t þ 1.
(a)
The statistical tests are based on simultaneous non-linear least square estimation as used in
Mishkin (1983), Sloan (1996) and Dechow et al. (2006)
Table II.
Historical, investors’
and ?nancial analysts’
weightings of the
prior-year accrual and
cash ?ow components
of earnings (34,205
?rm-years, 1985-2004)
ARJ
21,1
44
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
cash components suggest that ‘cash distributed to equity holders’ is the most
persistent. Recall that Table II, Panel B indicates that investors perceive ‘cash retained
by the ?rm’ to be the most persistent. Thus, analysts’ weightings are more consistent
with historical weightings in terms of the rankings among the three cash components.
Overall, the results in Table II suggest that ?nancial analysts are more
sophisticated than investors in evaluating the persistence of the cash components.
Financial analysts appear to recognize the relative differential persistence of all three
cash components, whereas investors appear to recognize only the higher persistence of
DIST
EQ
t
relative to DIST
D
t
, but not the higher persistence of DIST
EQ
t
relative to
DCASH
t
. Financial analysts correctly recognize that DIST
EQ
t
is the most persistent
among the three cash ?ow components, whereas investors rank DCASH
t
as the most
persistent cash component. Thus, our ?rst hypothesis (H1) is supported: the rankings
of the persistence of the three cash ?ow components by ?nancial analysts are more
consistent with the historical rankings than those by investors.
The tests above focus on directional comparisons (i.e. rankings) of the weightings of
the cash components implicit in historical, investors’ and analysts’ relations. To test
our second hypothesis (H2), the following section focuses on the magnitude of the
weighting differences across these relations.
4.3 Comparisons of weightings of prior-year cash components
Table III reports three sets of formal contrasts of the weighting differences for the cash
components reported in Tables II. First, the historical relations of the cash components
to subsequent-period realized earnings (Table II, Panel A) are compared to investors’
weightings (Table II, Panel B), to assess the magnitude of investors’ mis-weighting of
the three cash components. Second, the historical relations (Table II, Panel A) are
compared to ?nancial analysts’ weightings (Table II, Panel C), to evaluate the
magnitude of analysts’ mis-weighting of the cash components. Last, the investors’
weightings (Table II, Panel B) are compared to ?nancial analysts’ weightings (Table II,
Panel C) to address to what extent the analysts’ bias contributes to the investors’ bias
in weighting these cash components. All the three sets of comparisons are examined in
order to test our second hypothesis (H2).
The comparisons between historical and investors’ weightings in Table III showthat
investors overestimate the persistence of all earnings components. For example,
investors’ weighting of DIST
EQ
t
(0.891) is signi?cantly higher than its historical
weighting (0.709) both statistically (Likelihood ratio statistic ¼ 28.90, p-value ¼ 0.000)
and economically (investors mis-weight DIST
EQ
t
by 20.182)[4]. Among the cash
components, the overestimation in magnitude is most severe for DCASH
t
(with a
difference of 20.440) and least severe for DIST
D
t
(with a difference of 20.160).
Although this paper focuses oncash ?ows rather than accruals, Table III also shows that
investors signi?cantly overweight accruals, consistent with the ?ndings in prior studies.
The evidence that investors overestimate the persistence of all earnings components
suggests that they rely too heavily on past earnings in forecasting annual earnings.
The comparisons between historical and ?nancial analysts’ weightings reported in
Table III show that analysts appear to correctly weight DCASH
t
: there is no signi?cant
difference between the analysts’ weighting (0.534) and the historical weighting (0.526).
Analysts’ weighting of DIST
D
t
(0.564) is signi?cantly higher than its historical
weighting (0.517), and analysts’ weighting of DIST
EQ
t
(0.694) is signi?cantly lower
Forecasting
annual earnings
45
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
I
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
I
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
A
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
I
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
5
0
3
1
.
0
7
2
0
.
5
7
7
D
i
f
f
e
r
e
n
c
e
2
0
.
5
6
9
2
0
.
0
7
4
2
0
.
4
9
5
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
3
3
.
9
7
1
6
4
.
9
9
1
7
9
.
6
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
2
6
0
.
9
6
6
0
.
5
3
4
D
i
f
f
e
r
e
n
c
e
2
0
.
4
4
0
2
0
.
0
0
8
2
0
.
4
3
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
3
5
.
2
8
2
.
0
3
1
3
2
.
7
9
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
1
5
4
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
7
0
9
0
.
8
9
1
0
.
6
9
4
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
2
0
.
0
1
5
2
0
.
1
9
7
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
8
.
9
0
8
.
2
0
3
4
.
5
4
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
D
I
S
T
Dt
0
.
5
1
7
0
.
6
7
7
0
.
5
6
4
D
i
f
f
e
r
e
n
c
e
2
0
.
1
6
0
2
0
.
0
4
7
2
0
.
1
1
3
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
2
.
6
0
4
5
.
5
9
6
.
3
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
1
2
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
;
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
e
d
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table III.
Comparisons of
historical, investors’
and ?nancial analysts’
weightings of the cash
components of earnings
(34,205 ?rm-years,
1985-2004)
ARJ
21,1
46
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
than its historical weighting (0.709). The magnitudes of these differences, however, do
not seem economically signi?cant[5]. For example, analysts’ overweighting of DIST
D
t
is about 9 percent (0.047/0.517), and analysts’ underweighting of DIST
EQ
t
is only about
2 percent (0.015/0.709).
Lastly, the comparisons betweenanalysts andinvestors reported inTable III showthat
investors’ weightings consistently exceed analysts’ weightings for all cash components.
There are two interesting observations regarding these comparisons. First, analysts’
mis-weightings of DIST
EQ
t
and DIST
D
t
are substantially smaller than investors’
mis-weightings. For example, analysts’ mis-weighting of DIST
EQ
t
accounts for only
8 percent of investors’ mis-weighting (0.015/0.182 ¼ 8.24 percent). Analysts’
overweighting of DIST
D
t
accounts for about 29 percent of investors’ overweighting
(0.047/0.160 ¼ 29.37 percent). This indicates that analysts’ bias inweightingDIST
EQ
t
and
DIST
D
t
is at best a partial explanation for investors’ mis-weighting. Second, analysts’
weighting of DCASH
t
is unbiased whereas investors overweight DCASH
t
substantially.
Investors’ weighting for DCASH
t
is almost twice as large as the historical weighting
(0.966/0.526 ¼ 1.84). Both observations support the ?nding in prior literature that
?nancial analysts, as information intermediaries in the capital market, process certain
types of information more effectively than investors do. Thus, our second hypothesis (H2)
is supported: the magnitudes of the weightings of the three cash ?ow components by
?nancial analysts are more consistent with the historical weightings than are the
corresponding weightings by investors.
Overall, the evidence provided in Tables II and III suggest that ?nancial analysts do
not mislead investors in recognizing the differential persistence of cash components in
forecasting annual earnings. Analysts’ bias in weighting the cash components is at
best a partial explanation for investors’ bias. Other security market inef?ciencies that
are unrelated to ?nancial analysts’ earnings forecasts underlie at least part of
investors’ mis-weightings of cash components.
To present a visual summary of the above comparisons, Figure 1 depicts the
weightings for ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
based upon the historical, ?nancial
analysts’ and investors’ relations (in Table II) for all ?rm-years. Note that analysts’
forecasts re?ect the differential persistence of the earnings components much better
than investors do. Analysts appear to recognize that DIST
EQ
t
is the most persistent
component while ACCR
t
, DCASH
t
, and DIST
D
t
have lower weights, consistent with the
historical relations. In contrast, investors weight ACCR
t
and DCASH
t
higher than
DIST
EQ
t
. Second, analysts’ weightings closely follow those for the historical relations
both in direction and in magnitude. On the other hand, investors’ and historical
weightings are substantially different (the two lines are further apart). Third, the
difference in the weightings between the historical and investors’ relations is greater for
ACCR
t
and DCASH
t
and smaller for DIST
EQ
t
and DIST
D
t
, suggesting that investors’
mis-weightings are more severe for ACCR
t
and DCASH
t
.
4.4 Sample partition based on analyst coverage
This section evaluates whether the level of ?nancial analyst coverage affects investors’
and/or analysts’ differential weightings of the cash components of earnings. We use
?nancial analysts following as a proxy for the quality of the information environment
and investors’ sophistication in processing information. We partition our sample into
the lower and higher analyst coverage groups, de?ned as follows. We ?rst sort the
Forecasting
annual earnings
47
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sample ?rms in each year on the number of analysts’ forecasts included in the I/B/E/S
consensus forecasts. We then assign cases below the annual cross-sectional medians to
the lower analyst coverage group. The remaining cases are assigned to the higher
analyst coverage group. Because analyst coverage is a discrete variable, the two
analyst coverage groups have uneven number of cases. There are 15,603 and 18,602
cases for the lower and higher analyst coverage groups, respectively.
This section uses the same empirical models as in Section 4.2 but the relations are
estimated separately for ?rms in lower and higher analysts coverage groups. Section
4.4.1 examines the differential persistence (or rankings) of the cash components, and
section 4.4.2 assesses their absolute persistence (or magnitudes).
4.4.1 Historical, investors’ and analysts’ weightings of prior-year cash components
for the lower and higher analyst coverage groups. Table IV presents the weightings of
the earnings components reported separately within the lower and higher analyst
coverage groups. Overall, the results reveal that the quality of the information
environment has a minor impact on the historical relations and analysts’ weightings of
the cash ?ow components. Investors’ weightings, however, are affected to a greater
extent. For the historical relations, both panels in Table IV show that DIST
EQ
t
has the
highest persistence parameter among the three cash components, although the relative
persistence between DIST
D
t
and DCASH
t
differs across the two ?nancial analyst
coverage groups. For closely (lightly) followed ?rms, DIST
D
t
has signi?cantly higher
(lower) persistence than DCASH
t
. This evidence suggests that for ?rms in poorer (richer)
information environment, DCASH
t
is signi?cantly more (less) persistent than DIST
D
t
.
The results for investors’ weightings reported in Table IV Panel A show that
investors anticipate the higher persistence of DIST
EQ
t
relative to DIST
D
t
but incorrectly
Figure 1.
Historical, investors’ and
?nancial analysts’
weightings of earnings
components, (34,205
?rm-years, 1985-2004)
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
Historical
Analysts
Investors
Notes: (All earnings components variables are scaled by average total assets. Compustat item
numbers are in parentheses and firm specific subscripts are omitted):
ACCR
t
: Total annual accruals for year t = ?Non-Cash Assets–?Non-Debt Liabilities
Where Non-Cash Assets = Total assets (A6) –Cash & equivalents (A1), and
Non-Debt Liabilities = Total liabilities (A181) –Debt (A9+A34)
?CASH
t
: Change in cash for year t. Cash is defined as cash and equivalents (A1).
DIST
t
EQ
: Cash distributions to equity holders for year t = E
t
–Change in equity for year t
= E
t
–(?Total Assets–?Total Liabilities) for year t.
DIST
t
D
: Cash distributions to debt holders for year t = Reduction in debts for year t
= Reduction in long-term debt (A9) + Reduction in short-term debt (A34), for year t.
ACCR
t
?CASH
t
DIST
t
EQ
DIST
t
D
ARJ
21,1
48
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P
a
n
e
l
A
:
L
o
w
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
5
,
6
0
3
)
W
e
i
g
h
t
i
n
g
s
o
n
e
a
r
n
i
n
g
s
c
o
m
p
o
n
e
n
t
s
C
o
m
p
a
r
i
s
o
n
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
A
C
C
R
t
D
C
A
S
H
t
(
1
)
D
I
S
T
E
Q
t
(
2
)
D
I
S
T
Dt
(
3
)
(
1
)
-
(
2
)
(
1
)
-
(
3
)
(
2
)
-
(
3
)
H
i
s
t
o
r
i
c
a
l
C
o
e
f
?
c
i
e
n
t
0
.
5
1
0
0
.
5
2
2
0
.
7
0
8
0
.
4
9
2
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
6
0
.
0
3
0
0
.
2
1
6
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
9
0
.
0
0
8
0
.
0
0
8
0
.
0
1
0
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
9
2
7
.
9
8
7
.
1
1
4
8
1
.
5
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
8
)
(
0
.
0
0
0
)
I
n
v
e
s
t
o
r
s
C
o
e
f
?
c
i
e
n
t
1
.
1
1
2
1
.
0
0
4
0
.
8
9
7
0
.
5
5
6
D
i
f
f
e
r
e
n
c
e
0
.
1
0
7
0
.
4
4
8
0
.
3
4
1
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
6
7
0
.
0
6
3
0
.
0
5
3
0
.
0
7
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
6
.
1
7
3
2
.
3
8
2
4
.
5
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
A
n
a
l
y
s
t
s
C
o
e
f
?
c
i
e
n
t
0
.
6
0
4
0
.
5
5
8
0
.
7
3
5
0
.
5
7
0
D
i
f
f
e
r
e
n
c
e
2
0
.
1
7
7
2
0
.
0
1
2
0
.
1
6
5
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
9
0
.
0
0
9
0
.
0
0
8
0
.
0
1
1
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
7
2
2
.
1
2
0
.
9
1
2
4
0
.
8
1
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
3
4
1
)
(
0
.
0
0
0
)
P
a
n
e
l
B
:
h
i
g
h
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
8
,
6
0
2
)
W
e
i
g
h
t
i
n
g
s
o
n
e
a
r
n
i
n
g
s
c
o
m
p
o
n
e
n
t
s
C
o
m
p
a
r
i
s
o
n
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
A
C
C
R
t
D
C
A
S
H
t
(
1
)
D
I
S
T
E
Q
t
(
2
)
D
I
S
T
Dt
(
3
)
(
1
)
-
(
2
)
(
1
)
-
(
3
)
(
2
)
-
(
3
)
H
i
s
t
o
r
i
c
a
l
C
o
e
f
?
c
i
e
n
t
0
.
4
7
8
0
.
5
1
2
0
.
6
8
3
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
1
7
1
2
0
.
0
2
5
0
.
1
4
6
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
7
0
.
0
0
8
0
.
0
0
7
0
.
0
0
8
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
2
8
4
.
7
8
.
5
6
3
8
6
.
5
2
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
3
)
(
0
.
0
0
0
)
I
n
v
e
s
t
o
r
s
C
o
e
f
?
c
i
e
n
t
0
.
9
3
4
0
.
8
3
1
0
.
7
8
6
0
.
7
5
6
D
i
f
f
e
r
e
n
c
e
0
.
0
4
5
0
.
0
7
5
0
.
0
3
0
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
4
7
0
.
0
4
8
0
.
0
4
2
0
.
0
5
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
.
3
6
2
.
1
9
0
.
4
1
0
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
1
2
5
)
(
0
.
1
4
0
)
(
0
.
5
2
1
)
A
n
a
l
y
s
t
s
C
o
e
f
?
c
i
e
n
t
0
.
5
1
7
0
.
4
7
6
0
.
6
0
9
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
1
3
3
2
0
.
0
6
1
0
.
0
7
2
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
5
0
.
0
0
6
0
.
0
0
5
0
.
0
0
6
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
3
9
9
.
6
9
4
.
9
4
1
6
8
.
7
6
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
;
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table IV.
The impact of analyst
coverage on historical,
investors and ?nancial
analysts weightings of
the cash components of
earnings (34,205
?rm-years, 1985-2004)
Forecasting
annual earnings
49
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
perceive DCASH
t
as the most persistent cash component for the lower analyst
coverage group. These ?ndings are consistent with the results reported in Table II for
the full sample. Table IV Panel B shows that, for the higher analyst coverage group,
investors’ weighting of DIST
EQ
t
is not signi?cantly different from that of DIST
D
t
. It
appears that, contrary to our expectation, investors recognize the differential
persistence between DIST
EQ
t
and DIST
D
t
only for the lower analyst coverage group.
The overestimation of DCASH
t
by investors, however, is reduced in the higher
coverage group. The large variances in investors’ weightings provide one plausible
reason for our inability to detect investors’ better recognition of differential persistence
of cash components in the higher analyst coverage group.
Turning next to the analysts’ weightings, both panels show that analysts appear to
recognize the differential persistence of the cash components. For example, analysts
weight DIST
EQ
t
as the most persistent among the three cash components in both groups,
which is consistent with the historical relations. The persistence rankings of the cash
components by ?nancial analysts vary only slightly across the analysts coverage
groups. In the higher coverage group, analysts’ rankings of the cash components
correspond to their historical rankings, i.e. DIST
EQ
t
(DCASH
t
) is the most (least)
persistent cash component. In the lower coverage group, analysts’ weightings of
DCASH
t
and DIST
D
t
are statistically undistinguishable. These results support the view
that ?nancial analysts’ ability to recognize the relative persistence between DCASH
t
and
DIST
D
t
is positively associated with the quality of the information environment.
In summary, the results fromTable IVsupport hypothesis 3 (H3): the level of ?nancial
analyst coverage affects bothinvestors’ and ?nancial analysts’ rankings of the persistence
of the cash ?ow components. The extent to which investors and analysts are affected,
however, differ. Financial analysts’ rankings are less affectedthanare investors’ rankings.
For lightly followed ?rms, investors fail to recognize that DIST
EQ
t
is the most persistent
among the three cash ?ow components. For closely followed ?rms, there is no signi?cant
difference among the weightings on the three cash components by investors. On the other
hand, the level of analyst coverage has a positive effect on ?nancial analysts’ rankings of
the three cash components. For lightly followed ?rms, analysts fail to recognize the higher
persistence of DCASH
t
thanDIST
D
t
. For closely followed ?rms, analysts’ rankings of cash
components fully re?ect their historical rankings.
4.4.2 Comparisons of weightings of prior-year cash components for the lower and
higher analyst coverage groups. Table V presents the formal contrasts of the magnitudes
of the weightings reportedinTable IV. The contrasts are providedseparatelyfor the lower
andthe higher analyst coverage groups, inorder totest H4. First, we contrast the historical
and investors’ weightings across the two analyst coverage groups. Panel A of Table V
shows that for the lower analyst coverage group, there is no signi?cant difference (with
p-value of 0.383) between the historical and investors’ weightings of DIST
D
t
, i.e. investors
appear to correctly weight DIST
D
t
. Investors, however, overweight both the DCASH
t
and
the DIST
EQ
t
components. Panel B of Table V shows that investors overweight all three
cash components, for ?rms in the higher analyst coverage group. These results indicate
that the level of ?nancial analyst coverage affects the magnitudes of investors’ weightings
of the persistence of the cash ?ow components.
The comparisons between the historical and analysts’ weightings of the cash
components in Panel A of Table V show that, for the lower analyst coverage group,
analysts overweight all the cash components. Although the over-weightings are
ARJ
21,1
50
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P
a
n
e
l
A
:
L
o
w
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
5
,
6
0
3
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
V
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
I
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
A
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
I
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
5
1
0
1
.
1
1
2
0
.
6
0
4
D
i
f
f
e
r
e
n
c
e
2
0
.
6
0
2
2
0
.
0
9
4
2
0
.
5
0
8
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
0
2
.
8
2
9
5
.
6
4
7
3
.
8
9
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
2
2
1
.
0
0
4
0
.
5
5
8
D
i
f
f
e
r
e
n
c
e
2
0
.
4
8
2
2
0
.
0
3
6
2
0
.
4
4
6
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
6
8
.
4
3
1
3
.
9
8
5
9
.
7
4
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
7
0
8
0
.
8
9
7
0
.
7
3
5
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
9
2
0
.
0
2
7
2
0
.
1
6
2
L
-
R
s
t
a
t
i
s
t
i
c
1
2
.
9
4
9
.
9
4
9
.
6
8
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
2
)
(
0
.
0
0
2
)
D
I
S
T
Dt
0
.
4
9
2
0
.
5
5
6
0
.
5
7
0
D
i
f
f
e
r
e
n
c
e
2
0
.
0
6
4
2
0
.
0
7
8
0
.
0
1
4
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
0
.
7
6
4
3
.
8
3
0
.
0
3
(
p
-
v
a
l
u
e
)
(
0
.
3
8
3
)
(
0
.
0
0
0
)
(
0
.
8
5
5
)
P
a
n
e
l
B
:
H
i
g
h
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
8
,
6
0
2
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
V
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
i
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
a
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
i
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
4
7
8
0
.
9
3
4
0
.
5
1
7
D
i
f
f
e
r
e
n
c
e
2
0
.
4
5
6
2
0
.
0
3
9
2
0
.
4
1
7
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
0
9
.
7
5
3
9
.
7
1
9
4
.
0
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
1
2
0
.
8
3
1
0
.
4
7
6
D
i
f
f
e
r
e
n
c
e
2
0
.
3
1
9
0
.
0
3
6
2
0
.
3
5
5
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
4
6
.
9
7
3
0
.
5
9
5
9
.
8
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
6
8
3
0
.
7
8
6
0
.
6
0
9
D
i
f
f
e
r
e
n
c
e
2
0
.
1
0
3
0
.
0
7
4
2
0
.
1
7
7
L
-
R
s
t
a
t
i
s
t
i
c
5
.
9
8
1
5
7
.
6
3
1
8
.
2
1
(
p
-
v
a
l
u
e
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
Dt
0
.
5
3
7
0
.
7
5
6
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
2
1
9
0
.
0
0
0
2
0
.
2
1
9
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
8
.
2
6
0
.
0
0
0
1
8
.
7
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
9
6
8
)
(
0
.
0
0
0
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
e
d
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table V.
Contrast of historical,
?nancial analysts, and
investors weightings of
the cash components of
earnings by analyst
coverage group (34,205
?rm-years, 1985-2004)
Forecasting
annual earnings
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant statistically, the magnitudes of analysts’ biases are in general insigni?cant
economically. For example, analysts overweight DIST
EQ
t
by 4 percent ( 2 0.027/
0.708 ¼ 4 percent). Panel B of Table V reports the corresponding results for the higher
analyst coverage group and reveals that analysts correctly weight DIST
D
t
but
underweight DCASH
t
and DIST
EQ
t
. Note that analysts’ mis-weightings behave
differently – overweighting (underweighting) for the lower (higher) analyst coverage.
Since the level of analyst coverage proxies for the quality of the information
environment, one plausible explanation of this difference is that, in a rich information
environment with many competing information sources, ?nancial analysts are able to
reduce their reliance on prior-year earnings components in forecasting annual
earnings.
The ?nal part of Table V compares analysts’ and investors’ weightings. The results
in general provide evidence that ?nancial analysts are better able than investors at
processing the cash components information regardless of the quality of the information
environment. For example, in the lower coverage group, investors’ mis-weighting of
DCASH
t
is 13 times (20.482/20.036 ¼ 13.39) as much as analysts’ mis-weighting. In
the higher coverage group, investors’ mis-weighting of DCASH
t
is reduced to 9 times
(20.319/0.036 ¼ 8.86) as muchas analysts’ mis-weighting. These ?ndings complement
prior literature and suggest that ?nancial analysts, as information intermediaries, are
less biased than investors in processing not only the accruals but also the cash
components of earnings.
Overall, the results reported in Table V support our hypothesis 4 (H4): the level of
analyst coverage affects the magnitudes of investors’ and?nancial analysts’ weightings of
the cash?owcomponents whencomparedto the historical weightings. The magnitudes of
bothinvestors’ andanalysts’ mis-weightings of the cashcomponents are generallysmaller
for ?rms in the higher analyst coverage group. Consistent with prior literature, our results
show that higher analyst coverage indicates more ef?cient use of information.
5. Summary and conclusions
This study examines whether ?nancial analysts mislead investors in recognizing the
differential persistence of the cash ?ow components of earnings, de?ned by Dechowet al.
(2006), in forecasting annual earnings. The three cash components examined are cash
retained by the ?rm, cash distributed to equity holders and cash distributed to debt
holders. Dechow et al. (2006) document that investors correctly recognize the differential
persistence of cash distributed to equity holders and debt holders. On the other hand,
investors overestimate the persistence of cash retained in the ?rm. Our study extends that
inquiry and examines how?nancial analysts utilize the information contained in the three
cash components in forming their earnings forecasts, in order to determine whether
analysts’ biases contribute to investors’ biases. This inquiry is important to academics as
well as to the investment community. As information intermediaries, ?nancial analysts
play a prominent role in the ?nancial market. Consequently, the ability of analysts to
incorporate value-relevant information in their published expectations may impact
securities prices. Furthermore, our study partitions the sample based on the quality of the
information environment (proxied by the level of analyst coverage) in order to evaluate
whether the quality of the information environment affects ?nancial analysts’ and
investors’ weightings of differential cash components.
ARJ
21,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Consistent with our predictions, we ?nd that ?nancial analysts’ weightings of the three
cash ?ow components are more closely aligned with the historical relations than are the
corresponding weightings by investors, both in direction and in magnitude. Speci?cally,
?nancial analysts correctly recognize that cash distributed to equity holders is the most
persistent among the three cash ?ow components, whereas investors rank cash retained
by the ?rm as the most persistent cash component. In addition, the degree of analysts’
mis-weightings is economically small and much lower than the degree of investors’
mis-weightings. These results support our ?rst and second hypotheses.
Moreover, we ?nd that the level of ?nancial analyst coverage has an impact on the
weightings of the cash ?ow components by both investors and analysts, both in
direction and in magnitude, supporting our third and fourth hypotheses. The extent of
both investors’ and analysts’ mis-weightings of the cash components is generally
smaller for ?rms with greater levels of analyst following, which is widely used as a
proxy for the quality of a given ?rm’s information environment. Finally, we show that
cash distributed to equity holders is most persistent to subsequent year’s earnings,
regardless of the quality of the information environment.
Our ?ndings suggest that ?nancial analysts donot misleadinvestors inrecognizingthe
differential persistence of the cash components in forecasting annual earnings. Rather,
analysts’ bias inweightingthe cashcomponents of earnings is at best a partial explanation
for investors’ bias. Other security market inef?ciencies that are unrelated to ?nancial
analysts’ earnings forecasts underlie at least part of investors’ mis-weightings of cash
components. Earlier studies have shown that analysts are less biased than investors in
weightings the accrual component of earnings. This study extends that research, and
indicates that ?nancial analysts, as information intermediaries, are less biased than
investors in processing not only the accrual but also the cash components of earnings.
Notes
1. See Dechow et al. (2006), pp. 5-7 for a detailed explanation of this decomposition of income.
They scale all earnings components variables by average total assets.
2. Our ?ndings in both Panel A and Panel B of Table II are directionally the same as the main
persistence results reported by Dechow et al. (2006).
3. Note that the standard errors of the weightings by investors are about seven times as much
as those of the historical weightings, while the standard errors of analysts’ weightings are
nearly identical to those of the historical weightings. This suggests that the weightings of
earnings components by investors vary much more widely than either historical or analysts’
weightings.
4. Dechow et al. (2006) present similar results. Their Table V Panel D shows the investor’s
weightings (valuation coef?cients) of the three cash components are all signi?cantly greater
than the historical weightings (forecasting coef?cients). For example, the historical weight of
cash distributed to equity holders is 0.784, investors’ weight is 0.839, and the L-R statistic for
testing the equality of the two coef?cients is 18.35 ( p-value ¼ 0.001).
5. As noted earlier, the standard errors of analysts’ weightings are much lower (about 7 times)
than those of investors’ weightings. The lowstandard error may contribute to the statistically
signi?cant difference between historical and analysts’ weightings of cash distributed to
equity holders and cash distributed to debt holders, even though the magnitudes of the
differences are rather small.
Forecasting
annual earnings
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Abarbanell, J. and Bernard, V. (1992), “Tests of analysts’ overreaction/underreaction to earnings
information as an explanation for anomalous stock price behavior”, Journal of Finance,
Vol. 47 No. 3, pp. 1181-207.
Abarbanell, J. and Bushee, B. (1998), “Abnormal returns to a fundamental analysis strategy”,
The Accounting Review, Vol. 73 No. 1, pp. 19-45.
Bernard, V. and Thomas, J. (1989), “Post-earnings-announcement drift: delayed price response or
risk premium?”, Journal of Accounting Research, Vol. 27 No. 3, pp. 1-36.
Bernard, V. and Thomas, J. (1990), “Evidence that stock prices do not fully re?ect the implication
f current earnings for future earnings”, Journal of Accounting and Economics, Vol. 13,
pp. 305-40.
Bhattacharya, N. (2001), “April Investors’ trade size and trading responses around earnings
announcements: anempirical investigation”, The Accounting Review, Vol. 76 No. 2, pp. 221-44.
Bradshaw, M., Richardson, S. and Sloan, R. (2001), “June Do analysts and auditors use
information in accruals?”, Journal of Accounting Research, Vol. 39 No. 1, pp. 45-74.
Dechow, P. (1994), “Accounting earnings and cash ?ows as measures of ?rm performance:
the role of accounting accruals”, Journal of Accounting and Economics, Vol. 18 No. 1,
pp. 3-42.
Dechow, P., Richardson, S. and Sloan, R. (2006), “The persistence and pricing of the cash
component of earnings”, working paper, University of Michigan, Ann Arbor, MI.
Elgers, P., Lo, M. and Pfeiffer, R. (2001), “Delayed security price adjustments to ?nancial
analysts’ forecasts of annual earnings”, The Accounting Review, Vol. 76 No. 4, pp. 613-32.
Elgers, P., Lo, M. andPfeiffer, R. (2003), “Analysts’ vs investors’ weightings of accruals inforecasting
annual earnings”, Journal of Accounting and Public Policy, Vol. 22 No. 3, pp. 255-80.
Frankel, R. and Lee, C. (1998), “Accounting valuation, market expectation, and cross-sectional
stock returns”, Journal of Accounting and Economics, Vol. 25 No. 3, pp. 283-319.
Hong, H., Lim, T. and Stein, J. (2000), “Bad news travels slowly: size, analyst coverage, and the
pro?tability of momentum strategies”, Journal of Finance, Vol. 55 No. 1, pp. 265-95.
Mishkin, F. (1983), A Rational Expectations Approach to Macroeconomics, The University of
Chicago Press, Chicago, IL.
Ohlson, J. (1995), “Earnings, book values and dividends in equity valuation”, Contemporary
Accounting Research, Vol. 11 No. 2, pp. 661-87.
Sloan, R. (1996), “Do stock prices fully re?ect information in accruals and cash ?ows about future
earnings?”, The Accounting Review, Vol. 71 No. 3, pp. 289-315.
Walther, B. (1997), “Investor sophistication and market earnings expectations”, Journal of
Accounting Research, Vol. 35 No. 2, pp. 157-79.
Corresponding author
Le (Emily) Xu can be contacted at: [email protected]
ARJ
21,1
54
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. May H. Lo, Le (Emily) Xu. 2013. Regulation FD and analysts’ vs. investors’ weightings of the cash
components of earnings. Research in Accounting Regulation 25, 169-184. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_251893383.pdf
The purpose of this study is to examine whether financial analysts mislead investors in
recognizing the differential persistence of the three cash flow components of earnings, defined by
Dechow et al., in forecasting annual earnings.
Accounting Research Journal
Do analysts mislead investors?: A comparison of analysts' and investors' weightings of
cash components in forecasting annual earnings
May H. Lo Le (Emily) Xu
Article information:
To cite this document:
May H. Lo Le (Emily) Xu, (2008),"Do analysts mislead investors?", Accounting Research J ournal, Vol. 21
Iss 1 pp. 33 - 54
Permanent link to this document:
http://dx.doi.org/10.1108/10309610810891337
Downloaded on: 24 January 2016, At: 21:06 (PT)
References: this document contains references to 16 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 486 times since 2008*
Users who downloaded this article also downloaded:
Mostafa M. Maksy, Lin Zheng, (2008),"Factors associated with student performance in advanced
accounting and auditing: An empirical study in a public university", Accounting Research J ournal, Vol. 21
Iss 1 pp. 16-32 http://dx.doi.org/10.1108/10309610810891328
Raymond da Silva Rosa, J ennifer Filippetto, Ann Tarca, (2008),"ASIC actions: canaries for
poor corporate governance?", Accounting Research J ournal, Vol. 21 Iss 1 pp. 67-86 http://
dx.doi.org/10.1108/10309610810891355
Gerry Gallery, Natalie Gallery, Matthew Supranowicz, (2008),"Cash-based related party
transactions in new economy firms", Accounting Research J ournal, Vol. 21 Iss 2 pp. 147-166 http://
dx.doi.org/10.1108/10309610810905935
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Do analysts mislead investors?
A comparison of analysts’ and investors’
weightings of cash components in forecasting
annual earnings
May H. Lo
School of Business, Western New England College,
Spring?eld, Massachusetts, USA, and
Le (Emily) Xu
Department of Accounting and Finance,
Whittemore School of Business and Economics,
University of New Hampshire, Durham, New Hampshire, USA
Abstract
Purpose – The purpose of this study is to examine whether ?nancial analysts mislead investors in
recognizing the differential persistence of the three cash ?ow components of earnings, de?ned by
Dechow et al., in forecasting annual earnings.
Design/methodology/approach – The paper uses Mishkin’s econometric approach to compare the
persistence of the cash ?ow components within and across the historical, analysts’ and investors’
weightings.
Findings – It is found that ?nancial analysts’ weightings of the cash ?ow components are more
closely aligned with the historical relations than are investors’ weightings, both in direction and in
magnitude. The degree of analysts’ mis-weighting is economically small and much lower than the
degree of investors’ mis-weighting. Moreover, the extent of both investors’ and analysts’
mis-weightings of the cash components is generally smaller for ?rms with greater levels of analyst
following, a proxy for the quality of the information environment.
Research limitations/implications – The ?ndings suggest that ?nancial analysts’ bias in
weighting the cash components of earnings is at best a partial explanation for investors’ bias.
Practical implications – This study is important to academics and the investment community that
relies upon ?nancial analysts as information intermediaries, because the ability of analysts to
incorporate value-relevant information in their published expectations may impact securities prices.
Originality/value – The study is the ?rst to document the weightings of the cash components of
earnings by ?nancial analysts. In addition, this paper provides evidence that ?nancial analysts, as
information intermediaries, are less biased than investors in processing not only the accrual but also
the cash components of earnings.
Keywords Financial analysis, Investors, Earnings, Financial forecasting
Paper type Research paper
1. Introduction
Sloan (1996) demonstrates that operating cash ?ows are more persistent than accruals, in
the sense that operating cash ?ows have a stronger linear relation with subsequent-year
earnings than do accruals. Securities prices, however, act as if investors’ earnings
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors acknowledge I/B/E/S International, Inc. for providing earnings per share forecast
data available through the Institutional Brokers Estimate System.
Forecasting
annual earnings
33
Accounting Research Journal
Vol. 21 No. 1, 2008
pp. 33-54
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309610810891337
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
expectations fail to re?ect the greater persistence of the cash ?ow component of current
earnings. Following Sloan (1996), several studies show that investors under-weight the
persistence of operating cash ?ows and over-weight the accrual component of earnings.
For example, Bradshaw et al. (2001) use ?nancial analysts’ forecasts of earnings as
empirical proxies for investors’ expectations of earnings. They ?nd that the subsequently
revealed analysts’ forecast bias is positively related to the current period’s earnings
accruals. They also document that the apparent investor bias is driven solely by the
working capital component of the earnings accruals.
In order to examine whether ?nancial analysts’ forecasts of annual earnings re?ect
an over-weighting of working capital accruals that is comparable to the mis-weighting
documented in Sloan (1996) and Bradshaw et al. (2001), Elgers et al. (2003) ?nd that the
over-weighting of working capital accruals in analysts’ earnings forecasts is less than
one-third of the over-weighting by investors that is implicit in stock prices. Moreover,
they are able to attribute less than 40 percent of the delayed securities returns
associated with working capital accruals to subsequent errors in analysts’ annual
earnings forecasts, for ?rms with less than median levels of analyst coverage. Their
?ndings suggest that securities market inef?ciencies that are unrelated to ?nancial
analysts’ earnings forecasts underlie at least part of the accruals-related anomaly.
Recently, Dechow et al. (2006) investigate the persistence and pricing of the cash
component of earnings, as distinct from the accrual-related anomaly. They decompose
the cash component of earnings into three sub-components:
(1) cash that is retained by the ?rm;
(2) cash that is distributed to equity holders as a result of equity ?nancing; and
(3) cash that is distributed to debt holders as a result of debt ?nancing.
They demonstrate that the higher persistence of the cash component (relative to
accruals) is entirely attributable to net cash distributions to equity holders. Moreover,
investors appear to correctly anticipate the lower persistence of cash distributed to
debt holders relative to cash distributed to equity holders. However, investors also
appear to overestimate the persistence of cash retained by the ?rm relative to cash
distributed to equity holders and debt holders.
To extend this line of inquiry, our study examines whether or not ?nancial analysts
contribute to investors’ mis-weighting of cash ?ow components. We ?rst determine
whether ?nancial analysts recognize the differential persistence of the three cash ?ow
components in forming their forecasts of annual earnings. We next compare investors’
and analysts’ weightings of cash components using the historical persistence as a
benchmark. Our aim is to assess whether ?nancial analysts’ bias, if any, in weighting
cash components may contribute to investors’ bias. Moreover, Dechow et al. (2006) do
not partition their sample according to the quality of the information environment,
although recent evidence suggests that richer information environments are associated
with more ef?cient securities pricing. This study examines the impact of the quality of
the information environment on both investors’ and analysts’ weightings of cash
components. We achieve this purpose by grouping the cases based on the level of
?nancial analyst coverage. In summary, the purpose of this paper is to evaluate:
(1) whether analysts’ weightings of the three cash components are similar in
direction and magnitude to their historical weightings;
ARJ
21,1
34
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(2) whether investors’ and analysts’ weightings (and bias, if any) are similar; and
(3) whether the level of sophistication of the information environment affects
investors’ and ?nancial analysts’ weightings of differential cash components.
We examine differential persistence from two aspects:
(1) relative persistence among the three cash ?ow components compared within
each of the historical, investors’ and analysts’ weightings; and
(2) absolute persistence of the three cash ?ow components compared across the
historical, investors’ and analysts’ weightings.
The benchmark for assessing persistence is based upon the historical linear relations
between realized earnings and prior-year earnings components (detailed model
speci?cation is indicated in the next section). The coef?cients obtained from the
historical relations are historical weightings, and the rankings of the historical
weightings based on their magnitudes are historical rankings. With respect to relative
persistence, we examine the relative weightings (or rankings) of cash components that
are implicit in analysts’ earnings forecasts as well as in securities’ prices. With respect
to absolute persistence, we focus on the magnitudes of ?nancial analysts’ weightings
and compare these weightings to both the historical and investors’ weightings.
The focus of the paper is on ?nancial analysts’ (in contrast to investors’) utilization
of the information contained in the three cash components. This inquiry is important to
academics as well as to the investment community that relies upon ?nancial analysts
as information intermediaries. As information intermediaries, ?nancial analysts play a
prominent role in the ?nancial market. Consequently, the ability of analysts to
incorporate value-relevant information in their published expectations may impact
securities prices. If we provide evidence that ?nancial analysts mis-weight the cash
?ow components in the same manner, as do investors, this would be consistent with
the interpretation that analysts mislead investors in weighting the cash components.
Such evidence would suggest that ?nancial analysts do not fully impound the
earnings-relevant information in historical accounting information, and that this
aspect of their behavior may contribute at least in part to market inef?ciency. On the
other hand, if ?nancial analysts’ mis-weightings of cash components are statistically
less than those of investors’, this evidence would suggest that analysts are relatively
more ef?cient in processing ?nancial information, and that either investors fail to
impound earnings-relevant information contained in the analysts’ forecasts, or other
factors contribute to investors’ mis-weightings of the cash ?ow components.
Our initial empirical results show that cash distributed to equity holders is the most
persistent among all three cash components of earnings, and that investors correctly
anticipate the relatively higher (lower) persistence of cash distributed to equity holders
(debt holders). These results are consistent with Dechow et al. (2006). We then
document that ?nancial analysts’ forecasts incorporate the differential persistence of
the cash components better than investors do. Speci?cally, ?nancial analysts’ rankings
of the cash components of earnings are more consistent with the historical rankings
than the corresponding rankings by investors. Furthermore, the degree of analysts’
mis-weighting is economically small and much lower than the degree of investors’
mis-weighting that is implied in security returns. Our ?ndings suggest that ?nancial
analysts do not mislead investors in recognizing the differential persistence of cash
Forecasting
annual earnings
35
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components in forecasting annual earnings. Analysts’ bias in weighting the cash
components is at best a partial explanation for investors’ bias. We also ?nd that
?nancial analysts’ coverage has an impact on the weightings of the cash components
by investors. Consistent with our expectations, investors’ overweighting of accruals,
cash retained by the ?rm, and cash distributed to equity holders are greater for the
lower analyst coverage group than for the higher analyst coverage group.
The evidence in this study contributes to the literature in several ways. First, we
document that cash distributed to equity holders is the most persistent among the three
cash components across different information environments. Using ?nancial analysts’
following as proxy for the quality of the information environment, we ?nd that the
“cash distributed to equity-holders” component is more persistent than the “cash
retained by the ?rm” or “cash distributed to debt-holders” components in both lower
and higher analyst coverage groups. This ?nding suggests that the quality of the
information environment is not associated with factors that affect the higher
persistence of cash distributed to equity holders. This result is consistent with the
?nding in Elgers et al. (2003) on the relative persistence of the operating cash ?ows,
working capital accrual and other accrual components. Second, we show that investors’
overweighting of the cash components of earnings, as documented by Dechow et al.
(2006), is conditioned by the quality of the information environment, for which the level
of analyst following is a proxy. Our results indicate that investors’ overweighting of
cash components is in general greater for the lower analyst coverage group.
Finally, this study is the ?rst to document the weightings of the cash components of
earnings by ?nancial analysts. Prior research has examined both investors’ and
analysts’ weightings of the accrual component of annual earnings. For example,
Bradshaw et al. (2001) ?nd that investors overweight working capital accruals, and
Elgers et al. (2003) document that ?nancial analysts’ overweighting of working capital
accruals is much less than the overweighting by investors. In terms of the weightings
of the cash components of annual earnings, Dechow et al. (2006) examine investors’
weightings as compared to historical relations. Our study ?lls a gap in this line of
research by investigating analysts’ weightings of the cash components as compared to
both historical relations and investors’ weightings. Such an examination allows us to
assess whether analysts mislead investors in estimating the persistence of the three
cash components. Our empirical results indicate that ?nancial analysts’ weightings of
the cash ?ow components are more closely aligned with the historical relations than
are the corresponding weightings by investors, both in direction and in magnitude.
Our ?ndings complement prior literature and suggest that ?nancial analysts, as
information intermediaries, are less biased than are investors in processing not only
the accrual but also the cash components of earnings.
The rest of the paper is organized as follows. Section 2 develops our main
hypotheses and describes the research design. Section 3 explains our sample selection
and the measurement of the analysis variables. Section 4 presents the empirical results
and Section 5 provides a summary and conclusions.
2. Hypotheses development
This paper addresses the research question: “do analysts mislead investors?” in the
context of recognizing the differential persistence of the cash components of annual
earnings. To pursue this question, we ?rst examine whether the weightings of cash
ARJ
21,1
36
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components implied in analysts’ earnings forecasts are consistent with the historical
relations. We then compare analysts’ and investors’ weightings using historical
persistence estimate as a benchmark. Following the approach in Dechow et al. (2006),
we decompose annual earnings (E) into total accruals (ACCR), changes in cash
(DCash), cash distributions to equity holders (DIST
EQ
), and cash distributions to debt
holders (DIST
D
)[1]:
E ¼ ACCR þDCash þ DIST
EQ
þ DIST
D
ð1Þ
As noted earlier, Dechow et al. (2006) document that the higher persistence of the cash
component of earnings is attributable to cash distributions to equity holders, and that
investors tend to overestimate the persistence of cash retained by the ?rms, although
they correctly anticipate the relatively higher (lower) persistence of cash distributed to
equity (debt) holders. That evidence suggests that investors fail to fully recognize the
differential persistence of the cash components. As information intermediaries,
?nancial analysts presumably in?uence investors’ expectations. A primary aim of this
paper is to assess to what extent analysts’ bias contributes to investors’ bias in
weighting these cash components.
Prior literature has shown that in some contexts ?nancial analysts utilize ?nancial
information more effectively than investors do, and that market mispricing appears to
re?ect bias in investors’ earnings expectations that is more pronounced than any
related bias in ?nancial analysts’ forecasts. For example, Bernard and Thomas (1989,
1990) report a failure of stock prices to re?ect fully the implications of current quarterly
earnings for future earnings. Subsequently, Abarbanell and Bernard (1992) examine
whether ?nancial analysts’ under-reaction to prior earnings information could explain
these anomalous securities returns, and ?nd that under-reactions in analysts’
forecasts are only about half as large as necessary to explain the lagged securities price
adjustments. Elgers et al. (2003) ?nd that the over-weighting of working capital
accruals in analysts’ earnings forecasts is less than one-third of the over-weighting by
investors that is implicit in stock prices. Frankel and Lee (1998) develop pro?table
trading rules based on the ratio of “intrinsic value” (measured using a variant of
Ohlson’s (1995) residual income valuation model) to market value. The intrinsic value
measures are based upon ?nancial analysts’ earnings forecasts and growth
predictions, implying that investors do not ef?ciently incorporate these analysts’
expectations in securities prices. Similarly, Elgers et al. (2001) report delayed securities
price adjustments to share-price scaled analysts’ earnings forecasts. Collectively, these
studies suggest that in some contexts ?nancial analysts suffer less from information
inef?ciencies than do investors.
The above evidence leads us to hypothesize that ?nancial analysts are better able
than investors to recognize the differential persistence of various cash ?ow components
in forming their forecasts of annual earnings. We examine differential persistence from
two aspects: relative persistence among the three cash ?ow components compared
withineachof the historical investors’ and analysts’ weightings andabsolute persistence
of the three cash ?ow components compared across the historical, investors’ and
analysts’ weightings. With respect to relative persistence, we examine the relative
weightings (or rankings) of these cash components that are implicit in analysts’ earnings
forecasts as well as in securities’ prices. We expect that ?nancial analysts rank the
persistence of the cash components in a manner that is more consistent with their
Forecasting
annual earnings
37
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
historical rankings than do investors. With respect to absolute persistence, we focus on
the magnitudes of ?nancial analysts’ weightings and compare these weightings to both
the historical weightings and investors’ weightings. Speci?cally, we expect that
?nancial analysts’ weightings of the cash components are more consistent with their
historical weightings than the weightings by investors. Formally, we state our ?rst two
hypotheses as follows:
H1. The rankings of the persistence of the three cash ?ow components by
?nancial analysts are more consistent with the historical rankings than are
the corresponding rankings by investors.
H2. The magnitudes of the weightings of the three cash ?ow components by
?nancial analysts are more consistent with the historical weightings than are
the corresponding weightings by investors.
To evaluate whether ?nancial analysts mislead investors in weighting the cash
components in forecasting annual earnings, this study compares historical relations,
analysts’ weightings, and investors’ weightings. A ?nding that analysts perform better
than investors in weighting the cash ?ow components would complement the prior
literature and suggest that:
(1) analysts are more ef?cient than investors in processing not only accruals but
also cash ?ow information; and
(2) investors’ mis-weightings of the cash ?ow components are not caused entirely
by analysts’ bias.
Recent evidence indicates that ?nancial analyst coverage is positively associated with
the ef?ciency of investors’ information usage. For example, Walther (1997) reports that
investors in the securities of lightly followed ?rms over-rely on time-series predictions
of earnings, relative to ?nancial analysts’ earnings forecasts. Similarly, Bhattacharya
(2001) ?nds that, for ?rms with little to moderate analyst following, trading around
earnings announcements is more closely related to seasonal random-walk forecast
errors than to analysts’ forecast errors. Hong et al. (2000) ?nd that momentum-based
investment strategies are substantially more pro?table for ?rms with low analyst
coverage, because the initial under-reaction to value-relevant information, and
subsequent returns momentum, are stronger for these ?rms. Elgers et al. (2001)
document delayed security price adjustments to the information in ?nancial analysts’
earnings forecasts, and that the magnitudes of the delayed abnormal returns are larger
for ?rms with lower analyst coverage. These studies provide persuasive evidence that
analyst coverage is likely to be an effective partitioning variable. Given the premise
that there is more competition in information acquisition for ?rms with higher analyst
coverage, the quality of the information environment for highly followed ?rms is
expected to be better. Hence investors’ and analysts’ earnings forecasts for such ?rms
are expected to be more informed, or less biased. Moreover, to the extent that there is a
greater prevalence of less-sophisticated investors for ?rms with lower analyst
following, there is likely to be additional security pricing inef?ciency by investors,
beyond that induced by analysts’ forecasts. Accordingly, we adopt analyst coverage as
a partitioning device and formulate the following hypothesis:
ARJ
21,1
38
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H3. The level of ?nancial analyst coverage affects ?nancial analysts’ and
investors’ rankings of the persistence of the cash ?ow components.
H4. The level of ?nancial analyst coverage affects the magnitudes of ?nancial
analysts’ and investors’ weightings of the cash ?ow components.
Based on the evidence in prior research, we expect that the mis-weightings of the cash
components of earnings by investors or analysts will vary inversely with the level of
analyst coverage.
3. Research design
3.1 Sample selection and measurement of variables
The empirical analysis uses all December ?scal-year ?rm-years available from the
intersection of the I/B/E/S US Summary data base (consensus forecasts), the CRSP
monthly stock return data base and the Compustat annual data base (active and
research companies) with suf?cient information to measure the following required
variables. The speci?cations of the earnings, accruals, the cash ?ows components,
securities returns and analysts’ forecasts measures are indicated below (Compustat
item numbers are in parentheses and ?rm speci?c subscripts are omitted):
E
tþ1
¼ Annual income before extraordinary items and discontinued operations
available for common stockholders (A18) for year t þ 1.
ACCR
t
¼ Total annual accruals for year t.
¼ DNon-Cash Assets 2 DNon-Debt Liabilities
Where Non-Cash Assets ¼ Total assets (A6) 2 Cash & equivalents (A1)
Non-Debt Liabilities ¼ Total liabilities (A181) 2 Debt (A9 þ A34).
DCASH
t
¼ Change in cash for year t. Cash is de?ned as cash and equivalents (A1).
DIST
EQ
t
¼ Cash distribution to equity holders for year t.
¼ E
t
2 Change in equity for year t.
¼ E
t
(DTotal Assets 2 DTotal Liabilities) for year t.
DIST
D
t
¼ Cash distributions to debt holders for year t.
¼ Reduction in debts for year t.
¼ Reduction in long-term debt (A9) þ Reduction in short-term debt (A34),
for year t.
SAR
tþ1
¼ Size-adjusted security return, measured as the realized market return in
year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the
corresponding median return for all Compustat ?rms in the same
market capitalization decile at the start of year t þ 1.
FAF
tþ1
¼ I/B/E/S median (consensus) analyst forecast of annual earnings,
reported in May of the following year, multiplied by shares outstanding
and scaled by average total assets.
Because the start of the cumulating period for size-adjusted securities returns, and
hence the date at which the implicit securities market earnings expectations are
measured, is May 1 of the earnings year, we use the May consensus forecasts to ensure
comparability between analysts’ and market’s implicit earnings expectations.
All earnings and components variables are scaled by average total assets, as in
Sloan (1996), Bradshaw et al. (2001) and Elgers et al. (2003).
Forecasting
annual earnings
39
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
To enter our sample, we require ?rm-years to have complete data during our
analysis period for each of the variables described above. Speci?cally, we ?rst
eliminate ?rm-years with missing required data from COMPUSTAT and CRSP. This
screen yields 94,760 observations over the time span 1952-2004. We then eliminate
?rm-years with missing analysts’ forecasts of annual earnings from I/B/E/S, yielding
34,731 ?rm-year observations over the time span 1976-2004. Because I/B/E/S analyst
data was sparse prior to year 1985, we eliminate cases between 1976 and 1984 in order
to minimize self selection bias. Our ?nal sample contains 34,205 ?rm-years over the
sample period 1985-2004. Lastly, to mitigate the potential effects of outliers on our
analyses, we winsorize the extreme cases for the asset-scaled earnings components
variables at þ1 and 21 in each year, as used in Dechow et al. (2006).
3.2 Empirical models
To examine the historical persistence of earnings components, we relate realized
earnings to the earnings components from the previous year and estimate the following
model:
E
tþ1
¼ a
0
þa
1
ACCR
t
þa
2
DCASH
t
þa
3
DIST
EQ
t
þa
4
DIST
D
t
þm
tþ1
ð2Þ
where E
tþ1
is year t þ 1 earnings, ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
are the
accruals, change in cash, cash distributions to equity holders, and cash distributions to
debt holders, respectively, of prior-year’s earnings. The parameter estimates for the three
cash components of earnings (i.e.
_
a
i
, i ¼ 2; 3 and 4) indicate the historical persistence or
weightings of these cash components. These historical weightings are the benchmarks
against which the investors’ and ?nancial analysts’ weightings are compared.
We next examine investors’ weighting of these prior-year earnings components in
their earnings expectations. Because the earnings expectations impounded in securities
prices cannot be directly observed, our analysis relies on relations of cash components
and accruals to investor earnings expectations that are inferred from empirical
relations between abnormal security returns and “unexpected” earnings. We follow the
approach developed in Mishkin (1983) and subsequently adapted by Sloan (1996)
to assess the market’s weighting of earnings components. Following this general
approach, we estimate the historical linear relations of realized earnings, E
tþ1
, to
prior-year earnings components, as speci?ed in equation (2). To measure the implicit
linear weighting of these four earnings components in investor’s earnings expectations,
we specify the relation between abnormal returns and unexpected earnings as follows:
SAR
tþ1
¼ d
0
þd
1
UE
tþ1
þm
tþ1
ð3Þ
where SAR
tþ1
represents abnormal (size-adjusted) security returns, a commonly used
measure of unexpected returns (e.g. Abarbanell and Bushee, 1998; Elgers et al., 2003),
and UE
tþ1
represents unexpected earnings in year t þ 1. The unexpected earnings
variable, UE
tþ1
is then decomposed into realized earnings, E
tþ1
and expected earnings,
E(E
tþ1
):
UE
tþ1
¼ E
tþ1
2EðE
tþ1
Þ
we substitute the linear prediction of E
tþ1
in equation (2) for expected earnings E(E
tþ1
)
in equation (3), and rearrange equation (3) as follows:
ARJ
21,1
40
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
SAR
tþ1
¼ d
0
þd
1
½E
tþ1
2EðE
tþ1
Þ? þm
tþ1
ð4aÞ
¼d
0
þd
1
E
tþ1
2d
1
_
g
0
þ
_
g
1
ACCR
t
þ
_
g
2
DCASH
t
þ
_
g
3
DIST
EQ
1
þ
_
g
4
DIST
D
t
þm
tþ1
ð4bÞ
¼f
0
þd
1
E
tþ1
þf
1
ACCR
t
þf
2
DCASHþf
3
DIST
EQ
t
þf
4
DIST
D
t
þm
tþ1
ð4cÞ
where the parameter estimates
_
g
i
(i ¼ 1, 2, 3 and 4) represent the slope parameters of
the linear relation of the earnings components to the security market’s expectations,
E(E
tþ1
). In other words, they indicate investor’s implicit weightings of prior periods’
earnings components in forecasting earnings. Empirical estimation of equation (4c)
yields estimates of d
1
and f
i
. We then de?ne investors’ weightings as
_
g
i
¼2
^
f
i
=
_
d
1
. If
investors utilize the information contained in these earnings components
effectively, then investors’ weightings of earnings components in equation (4b)
would be consistent with the historical weightings in equation (2), i.e.
_
g
i
¼
_
a
i
. We use
the following equation to indicate investors’ weighting model hereafter to simplify the
expression:
SAR
tþ1
¼d
0
þd
1
E
tþ1
2g
0
2g
1
ACCR
t
2g
2
DCASH
t
2g
3
DIST
EQ
t
2g
4
DIST
D
t
þm
tþ1
ð5Þ
To examine the weightings of the prior-year earnings components by ?nancial
analysts, we regress ?nancial analysts’ earnings forecasts on the accrual and cash
components of earnings from the previous year and estimate the following model:
FAF
tþ1
¼b
0
þb
1
ACCR
t
þb
2
DCASH
t
þb
3
DIST
EQ
t
þb
4
DIST
D
t
þm
tþ1
ð6Þ
where FAF
tþ1
is analyst forecasts of annual earnings in year t þ 1. This model is
parallel to the historical relation in equation (2). The parameter estimates
_
b
i
(i ¼ 1, 2, 3
and 4) indicate analysts’ weightings of prior-year earnings components implied in their
earnings forecasts. To test our ?rst two hypotheses, we compare three sets of weightings
of the cash components, i.e. the historical persistence (the coef?cient estimates of a
2
, a
3
,
and a
4
in expression (2)), the investors’ weighting (the coef?cient estimates of g
2
, g
3
, and
g
4
in expression (5)), and the analysts’ weighting (the coef?cient estimates of b
2
, b
3
, and
b
4
in expression (6)) both in direction and in magnitude.
Speci?cally, to test H1, we examine the rankings of the cash ?ow components by
both investors and analysts, and compare their rankings to the historical rankings
observed from expression (2). To test H2, we formally contrast investors’ and analysts’
weightings to historical weightings separately for the three cash ?ow components.
equations (2), (5) and (6) are also estimated separately for both the lightly and closely
followed ?rms, in order to examine how the quality of the information environment
(for which analyst coverage is a proxy) affects investors’ and analysts’ weightings of
the cash components (H3 and H4).
Following Sloan (1996) and Dechow et al. (2006), we use Mishkin’s (1983)
econometric approach, i.e. simultaneous non-linear least squares regression, that
allows us to simultaneously estimate the historical persistence of the accrual and cash
Forecasting
annual earnings
41
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
components of earnings (in expression (2)), the investors’ weightings of the
corresponding components (in expression (5)) and the ?nancial analysts’ weightings
of the corresponding components (in expression (6)). For more detailed discussion
about this framework, see Mishkin (1983) and Sloan (1996).
4. Empirical results
We organize our empirical results as follows. Section 4.1 presents descriptive statistics
for the analysis variables. Section 4.2 discusses historical, investors’ and ?nancial
analysts’ weightings of prior-year accruals and cash ?ow components. Section 4.3
presents the results from the comparisons of historical, investors’ and analysts’
weightings of the cash ?ow components. Section 4.4 describes the weighting results for
?rm-years in different ?nancial analyst coverage groups.
4.1 Descriptive statistics
Table I presents the univariate statistics and Pearson correlations between the key
variables for the ?rm-years included in the subsequent analysis. These descriptive
statistics re?ect various regularities reported in recent research. The univariate
statistics for the earnings components variables display the same characteristics shown
by Dechow et al. (2006): positive means for both Accruals, ACCR
t
, and Change in cash,
DCASH
t
, indicating that our sample ?rms have been growing during our sample years;
negative means for DIST
EQ
t
and DIST
D
t
, indicating that the amount of capital raised by
those ?rms from their capital holders is more than the amounts distributed to capital
holders. The standard deviations of individual components of earnings show that each
Mean SD
Pearson correlations
ACCR
t
DCASH
t
DIST
EQ
t
DIST
D
t
SAR
tþ1
FAF
tþ1
E
t
0.017 0.163 0.231 0.217 0.226 0.128 20.043 0.581
ACCR
t
0.081 0.198 1 0.028 20.358 20.521 20.082 0.071
DCASH
t
0.040 0.195 1 20.659 20.008 20.060 20.041
DIST
EQ
t
20.075 0.249 1 20.085 0.062 0.330
DIST
D
t
20.030 0.147 1 0.037 0.046
SAR
tþ1
0.024 0.652 1 20.001
FAF
tþ1
0.028 0.170 1
Notes: (All earnings and earnings components variables are scaled by average total assets. Compustat
item numbers are in parentheses and ?rm speci?c subscripts are omitted): E
t
, annual income before
extraordinary items and discontinued operations available for common stockholders (A18) for year t;
ACCR
t
, total annual accruals for year t ¼ DNon-Cash Assets 2 DNon-Debt Liabilities, where
Non-Cash Assets ¼ Total assets (A6) 2 Cash and equivalents (A1), and Non-Debt Liabilities ¼ Total
liabilities (A181) 2 Debt (A9 þ A34); DCASH
t
, change in cash for year t. Cash is de?ned as cash and
equivalents (A1); DIST
EQ
t
, cash distributions to equity holders for year t ¼ E
t
2 Change in equity for
year t ¼ E
t
2 (DTotal Assets 2 DTotal Liabilities) for year t; DIST
D
t
, cash distributions to debt
holders for year t ¼ Reduction in debts for year t ¼ Reduction in long-term debt (A9) þ Reduction in
short-term debt (A34), for year t; SAR
tþ1
, size-adjusted security return, measured as the realized
market return in year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the corresponding
median return for all Compustat ?rms in the same market capitalization decile at the start of year t þ 1;
FAF
tþ1
, I/B/E/S median (consensus) analyst forecast of annual earnings for year t þ 1, reported in
May of year t þ 1, multiplied by shares outstanding and scaled by average total assets
Table I.
Descriptive statistics
(34,205 ?rm-years,
1985-2004)
ARJ
21,1
42
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
component, i.e. ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
, represents an important source of
the variation in earnings.
Pearson correlations reported in Table I are also consistent with prior studies. For
example, size-adjusted returns, SAR
tþ1
, and prior-year accruals, ACCR
t
, are negatively
correlated (with a correlation coef?cient of 20.082), consistent with the lagged security
price adjustments to accruals reported in Sloan (1996), Bradshaw et al. (2001) and Elgers
et al. (2003). The correlation between accruals, ACCR
t,
and cash distributed to equity
holders, DIST
EQ
t
, as well as to debt holders, DIST
D
t
, are negative (with a correlation
coef?cient of 20.358 and 20.521, respectively), consistent with the role of accruals in
mitigating timing problems in cash ?ow measures of earnings (Dechow, 1994). Not
surprisingly, DCASH
t
and DIST
EQ
t
are also negatively correlated (with a correlation
coef?cient of 20.659), since cashdistributions toequityholders consume a ?rm’s free cash.
4.2 Historical, investors’ and analysts’ weightings of prior-year cash components
This section presents comparisons of the estimated linear relations of accruals and
cash ?ow components to subsequent-year realized earnings, investor expectations of
earnings inferred from returns/earnings regressions, and ?nancial analysts’ forecasts
of earnings. We ?rst examine the historical persistence of earnings components. This is
followed by deriving inferred investors’ weightings of the earnings components.
Market ef?ciency requires that the weightings of earnings components by investors
re?ect the historical weightings without bias. Finally we investigate the weightings of
the earnings components by ?nancial analysts and compare investors’ and analysts’
weightings, in order to test our ?rst hypothesis (H1).
Panel A of Table II presents the estimated historical persistence of accruals and
cash ?ow components from equation (2). The historical results correspond closely to
those reported by Dechow et al. (2006). The results show that DIST
EQ
t
has the highest
persistence among the three cash components. The persistence coef?cient for DIST
EQ
t
(0.709) is signi?cantly higher than those for DCASH
t
(0.526) and DIST
D
t
(0.517). There
is, however, no signi?cant difference in the historical persistence between DCASH
t
and
DIST
D
t
. Notice that the persistence coef?cients for DCASH
t
and DIST
D
t
are both
similar to the persistence coef?cient for ACCR
t
(0.503), which suggest that the higher
persistence of cash ?ows vis-a`-vis accruals documented in prior literature is entirely
driven by cash distributed to equity holders. These historical persistence measures will
be used as benchmarks to examine the corresponding weightings by ?nancial analysts
and investors.
Panel B of Table II reports the results from estimating equation (5) to determine
investors’ weightings of the accruals and the cash components in their earnings
expectations. Descriptively, the results in Panel B indicate that investors appear to
weight ACCR
t
(DIST
D
t
) highest (lowest) among the four earnings components,
consistent with Dechow et al. (2006). The test of equality of the coef?cients of DIST
EQ
t
and DIST
D
t
shows that investors’ weighting of DIST
EQ
t
(0.891) is signi?cantly higher
than that of DIST
D
t
(0.677), suggesting that investors correctly anticipate the higher
persistence of cash distributed to equity holders relative to cash distributed to debt
holders. This result is consistent with the historical relations shown in Panel A of
Table II. Contrary to the historical relations, however, investors’ weighting of DCASH
t
(0.966) is signi?cantly higher than their weighting of DIST
EQ
t
(0.891), suggesting that
investors overestimate the persistence of DCASH
t
[2]. Investors perceive DCASH
t
as
Forecasting
annual earnings
43
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the most persistent cash component, while the historical relations indicate that DIST
EQ
t
is the most persistent. In this respect, investors’ rankings of the cash components fail to
fully re?ect the historical rankings[3].
Panel C of Table II presents ?nancial analysts’ weightings of accruals and cash
components obtained from estimating equation (6). The estimation results show that
the coef?cient for DIST
EQ
t
(0.694) is signi?cantly greater than those for DIST
D
t
(0.564)
and DCASH
t
(0.534). This contrast indicates that ?nancial analysts weight the cash
component of earnings that is distributed to equity holders more heavily than either of
the other cash components. These directional results are consistent with the historical
relations reported in Table II, Panel A. Judging from the rankings of the persistence of
various cash ?ow components, ?nancial analysts appear to fully recognize their
differential persistence. Both the historical rankings and the analysts’ rankings of the
Panel A: Historical relations of realized earnings to the accrual and cash ?ow components of earnings
E
tþ1
¼ a
0
þa
1
ACCR
t
þa
2
DCASH
t
þa
3
DIST
EQ
t
þa
4
DIST
D
t
þm
tþ1
ð2Þ
Historical weightings Comparison of weightings
(a)
a
1
a
2
a
3
a
4
a
2
¼ a
3
a
2
¼ a
4
a
3
¼ a
4
Coef?cient 0.503 0.526 0.709 0.517 Difference –0.183 0.009 0.192
Standard error 0.006 0.006 0.005 0.007 Likelihood ratio 2216.1 1.52 960.78
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.000) (0.217) (0.000)
Panel B: Investors’ weightings of the accrual and cash ?ow components of prior-year earnings
SAR
tþ1
¼ d
0
þd
1
ðE
tþ1
2g
0
2g
1
ACCR
t
2g
2
DCASH
t
2g
3
DIST
EQ
t
2g
4
DIST
D
t
Þ þm
tþ1
ð5Þ
Investors’ weightings Comparison of weightings
(a)
g
1
g
2
g
3
g
4
g
2
¼ g
3
g
2
¼ g
4
g
3
¼ g
4
Coef?cient 1.072 0.966 0.891 0.677 Difference 0.075 0.289 0.214
Standard error 0.042 0.041 0.034 0.045 Likelihood ratio 7.96 37.21 26.40
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.005) (0.000) (0.000)
Panel C: Financial analysts’ weightings of the accrual and cash ?ow components of prior-year earnings
FAF
tþ1
¼ b
0
þb
1
ACCR
t
þb
2
DCASH
t
þb
3
DIST
EQ
t
þb
4
DIST
D
t
þm
tþ1
ð6Þ
Financial Analysts’ weightings Comparison of weightings
(a)
b
1
b
2
b
3
b
4
b
2
¼ b
3
b
2
¼ b
4
b
3
¼ b
4
Coef?cient 0.577 0.534 0.694 0.564 Difference –0.160 –0.030 0.130
Standard error 0.005 0.005 0.005 0.007 Likelihood ratio 1753.7 18.51 456.95
( p-value) (0.000) (0.000) (0.000) (0.000) ( p-value) (0.000) (0.000) (0.000)
Notes: (All earnings and earnings components variables are scaled by average total assets.
Compustat item numbers are in parentheses and ?rm speci?c subscripts are omitted): E
tþ1
(E
t
), Annual
income before extraordinary items and discontinued operations available for common stockholders
(A18) for year t þ 1 (t); ACCR
t
, total annual accruals for year t ¼ DNon-Cash Assets 2 DNon-Debt
Liabilities, where Non-Cash Assets ¼ Total assets (A6) 2 Cash and equivalents (A1) and Non-Debt
Liabilities ¼ Total liabilities (A181) 2 Debt (A9 þ A34); DCASH
t
, change in cash for year t. Cash is
de?ned as cash and equivalents (A1); DIST
EQ
t
, cash distributions to equity holders for year t ¼
E
t
2Change in equity for year t ¼ E
t
2 (DTotal Assets 2 DTotal Liabilities) for year t; DIST
D
t
, cash
distributions to debt holders for year t ¼ Reduction in debts for year t ¼ Reduction in long-term debt
(A9) þ Reduction in short-term debt (A34), for year t; FAF
tþ1
, I/B/E/S median (consensus) analyst
forecast of annual earnings, reported in May of the following year, multiplied by shares outstanding
and scaled by average total assets; SAR
tþ1
, size-adjusted security return, measured as the realized
market return in year t þ 1 (May 1, year t þ 1 through April 30, year t þ 2), less the corresponding
median return for all Compustat ?rms in the same market capitalization decile at the start of year
t þ 1.
(a)
The statistical tests are based on simultaneous non-linear least square estimation as used in
Mishkin (1983), Sloan (1996) and Dechow et al. (2006)
Table II.
Historical, investors’
and ?nancial analysts’
weightings of the
prior-year accrual and
cash ?ow components
of earnings (34,205
?rm-years, 1985-2004)
ARJ
21,1
44
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
cash components suggest that ‘cash distributed to equity holders’ is the most
persistent. Recall that Table II, Panel B indicates that investors perceive ‘cash retained
by the ?rm’ to be the most persistent. Thus, analysts’ weightings are more consistent
with historical weightings in terms of the rankings among the three cash components.
Overall, the results in Table II suggest that ?nancial analysts are more
sophisticated than investors in evaluating the persistence of the cash components.
Financial analysts appear to recognize the relative differential persistence of all three
cash components, whereas investors appear to recognize only the higher persistence of
DIST
EQ
t
relative to DIST
D
t
, but not the higher persistence of DIST
EQ
t
relative to
DCASH
t
. Financial analysts correctly recognize that DIST
EQ
t
is the most persistent
among the three cash ?ow components, whereas investors rank DCASH
t
as the most
persistent cash component. Thus, our ?rst hypothesis (H1) is supported: the rankings
of the persistence of the three cash ?ow components by ?nancial analysts are more
consistent with the historical rankings than those by investors.
The tests above focus on directional comparisons (i.e. rankings) of the weightings of
the cash components implicit in historical, investors’ and analysts’ relations. To test
our second hypothesis (H2), the following section focuses on the magnitude of the
weighting differences across these relations.
4.3 Comparisons of weightings of prior-year cash components
Table III reports three sets of formal contrasts of the weighting differences for the cash
components reported in Tables II. First, the historical relations of the cash components
to subsequent-period realized earnings (Table II, Panel A) are compared to investors’
weightings (Table II, Panel B), to assess the magnitude of investors’ mis-weighting of
the three cash components. Second, the historical relations (Table II, Panel A) are
compared to ?nancial analysts’ weightings (Table II, Panel C), to evaluate the
magnitude of analysts’ mis-weighting of the cash components. Last, the investors’
weightings (Table II, Panel B) are compared to ?nancial analysts’ weightings (Table II,
Panel C) to address to what extent the analysts’ bias contributes to the investors’ bias
in weighting these cash components. All the three sets of comparisons are examined in
order to test our second hypothesis (H2).
The comparisons between historical and investors’ weightings in Table III showthat
investors overestimate the persistence of all earnings components. For example,
investors’ weighting of DIST
EQ
t
(0.891) is signi?cantly higher than its historical
weighting (0.709) both statistically (Likelihood ratio statistic ¼ 28.90, p-value ¼ 0.000)
and economically (investors mis-weight DIST
EQ
t
by 20.182)[4]. Among the cash
components, the overestimation in magnitude is most severe for DCASH
t
(with a
difference of 20.440) and least severe for DIST
D
t
(with a difference of 20.160).
Although this paper focuses oncash ?ows rather than accruals, Table III also shows that
investors signi?cantly overweight accruals, consistent with the ?ndings in prior studies.
The evidence that investors overestimate the persistence of all earnings components
suggests that they rely too heavily on past earnings in forecasting annual earnings.
The comparisons between historical and ?nancial analysts’ weightings reported in
Table III show that analysts appear to correctly weight DCASH
t
: there is no signi?cant
difference between the analysts’ weighting (0.534) and the historical weighting (0.526).
Analysts’ weighting of DIST
D
t
(0.564) is signi?cantly higher than its historical
weighting (0.517), and analysts’ weighting of DIST
EQ
t
(0.694) is signi?cantly lower
Forecasting
annual earnings
45
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
I
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
I
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
A
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
I
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
5
0
3
1
.
0
7
2
0
.
5
7
7
D
i
f
f
e
r
e
n
c
e
2
0
.
5
6
9
2
0
.
0
7
4
2
0
.
4
9
5
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
3
3
.
9
7
1
6
4
.
9
9
1
7
9
.
6
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
2
6
0
.
9
6
6
0
.
5
3
4
D
i
f
f
e
r
e
n
c
e
2
0
.
4
4
0
2
0
.
0
0
8
2
0
.
4
3
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
3
5
.
2
8
2
.
0
3
1
3
2
.
7
9
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
1
5
4
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
7
0
9
0
.
8
9
1
0
.
6
9
4
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
2
0
.
0
1
5
2
0
.
1
9
7
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
8
.
9
0
8
.
2
0
3
4
.
5
4
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
D
I
S
T
Dt
0
.
5
1
7
0
.
6
7
7
0
.
5
6
4
D
i
f
f
e
r
e
n
c
e
2
0
.
1
6
0
2
0
.
0
4
7
2
0
.
1
1
3
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
2
.
6
0
4
5
.
5
9
6
.
3
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
1
2
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
;
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
e
d
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table III.
Comparisons of
historical, investors’
and ?nancial analysts’
weightings of the cash
components of earnings
(34,205 ?rm-years,
1985-2004)
ARJ
21,1
46
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
than its historical weighting (0.709). The magnitudes of these differences, however, do
not seem economically signi?cant[5]. For example, analysts’ overweighting of DIST
D
t
is about 9 percent (0.047/0.517), and analysts’ underweighting of DIST
EQ
t
is only about
2 percent (0.015/0.709).
Lastly, the comparisons betweenanalysts andinvestors reported inTable III showthat
investors’ weightings consistently exceed analysts’ weightings for all cash components.
There are two interesting observations regarding these comparisons. First, analysts’
mis-weightings of DIST
EQ
t
and DIST
D
t
are substantially smaller than investors’
mis-weightings. For example, analysts’ mis-weighting of DIST
EQ
t
accounts for only
8 percent of investors’ mis-weighting (0.015/0.182 ¼ 8.24 percent). Analysts’
overweighting of DIST
D
t
accounts for about 29 percent of investors’ overweighting
(0.047/0.160 ¼ 29.37 percent). This indicates that analysts’ bias inweightingDIST
EQ
t
and
DIST
D
t
is at best a partial explanation for investors’ mis-weighting. Second, analysts’
weighting of DCASH
t
is unbiased whereas investors overweight DCASH
t
substantially.
Investors’ weighting for DCASH
t
is almost twice as large as the historical weighting
(0.966/0.526 ¼ 1.84). Both observations support the ?nding in prior literature that
?nancial analysts, as information intermediaries in the capital market, process certain
types of information more effectively than investors do. Thus, our second hypothesis (H2)
is supported: the magnitudes of the weightings of the three cash ?ow components by
?nancial analysts are more consistent with the historical weightings than are the
corresponding weightings by investors.
Overall, the evidence provided in Tables II and III suggest that ?nancial analysts do
not mislead investors in recognizing the differential persistence of cash components in
forecasting annual earnings. Analysts’ bias in weighting the cash components is at
best a partial explanation for investors’ bias. Other security market inef?ciencies that
are unrelated to ?nancial analysts’ earnings forecasts underlie at least part of
investors’ mis-weightings of cash components.
To present a visual summary of the above comparisons, Figure 1 depicts the
weightings for ACCR
t
, DCASH
t
, DIST
EQ
t
and DIST
D
t
based upon the historical, ?nancial
analysts’ and investors’ relations (in Table II) for all ?rm-years. Note that analysts’
forecasts re?ect the differential persistence of the earnings components much better
than investors do. Analysts appear to recognize that DIST
EQ
t
is the most persistent
component while ACCR
t
, DCASH
t
, and DIST
D
t
have lower weights, consistent with the
historical relations. In contrast, investors weight ACCR
t
and DCASH
t
higher than
DIST
EQ
t
. Second, analysts’ weightings closely follow those for the historical relations
both in direction and in magnitude. On the other hand, investors’ and historical
weightings are substantially different (the two lines are further apart). Third, the
difference in the weightings between the historical and investors’ relations is greater for
ACCR
t
and DCASH
t
and smaller for DIST
EQ
t
and DIST
D
t
, suggesting that investors’
mis-weightings are more severe for ACCR
t
and DCASH
t
.
4.4 Sample partition based on analyst coverage
This section evaluates whether the level of ?nancial analyst coverage affects investors’
and/or analysts’ differential weightings of the cash components of earnings. We use
?nancial analysts following as a proxy for the quality of the information environment
and investors’ sophistication in processing information. We partition our sample into
the lower and higher analyst coverage groups, de?ned as follows. We ?rst sort the
Forecasting
annual earnings
47
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sample ?rms in each year on the number of analysts’ forecasts included in the I/B/E/S
consensus forecasts. We then assign cases below the annual cross-sectional medians to
the lower analyst coverage group. The remaining cases are assigned to the higher
analyst coverage group. Because analyst coverage is a discrete variable, the two
analyst coverage groups have uneven number of cases. There are 15,603 and 18,602
cases for the lower and higher analyst coverage groups, respectively.
This section uses the same empirical models as in Section 4.2 but the relations are
estimated separately for ?rms in lower and higher analysts coverage groups. Section
4.4.1 examines the differential persistence (or rankings) of the cash components, and
section 4.4.2 assesses their absolute persistence (or magnitudes).
4.4.1 Historical, investors’ and analysts’ weightings of prior-year cash components
for the lower and higher analyst coverage groups. Table IV presents the weightings of
the earnings components reported separately within the lower and higher analyst
coverage groups. Overall, the results reveal that the quality of the information
environment has a minor impact on the historical relations and analysts’ weightings of
the cash ?ow components. Investors’ weightings, however, are affected to a greater
extent. For the historical relations, both panels in Table IV show that DIST
EQ
t
has the
highest persistence parameter among the three cash components, although the relative
persistence between DIST
D
t
and DCASH
t
differs across the two ?nancial analyst
coverage groups. For closely (lightly) followed ?rms, DIST
D
t
has signi?cantly higher
(lower) persistence than DCASH
t
. This evidence suggests that for ?rms in poorer (richer)
information environment, DCASH
t
is signi?cantly more (less) persistent than DIST
D
t
.
The results for investors’ weightings reported in Table IV Panel A show that
investors anticipate the higher persistence of DIST
EQ
t
relative to DIST
D
t
but incorrectly
Figure 1.
Historical, investors’ and
?nancial analysts’
weightings of earnings
components, (34,205
?rm-years, 1985-2004)
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
Historical
Analysts
Investors
Notes: (All earnings components variables are scaled by average total assets. Compustat item
numbers are in parentheses and firm specific subscripts are omitted):
ACCR
t
: Total annual accruals for year t = ?Non-Cash Assets–?Non-Debt Liabilities
Where Non-Cash Assets = Total assets (A6) –Cash & equivalents (A1), and
Non-Debt Liabilities = Total liabilities (A181) –Debt (A9+A34)
?CASH
t
: Change in cash for year t. Cash is defined as cash and equivalents (A1).
DIST
t
EQ
: Cash distributions to equity holders for year t = E
t
–Change in equity for year t
= E
t
–(?Total Assets–?Total Liabilities) for year t.
DIST
t
D
: Cash distributions to debt holders for year t = Reduction in debts for year t
= Reduction in long-term debt (A9) + Reduction in short-term debt (A34), for year t.
ACCR
t
?CASH
t
DIST
t
EQ
DIST
t
D
ARJ
21,1
48
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P
a
n
e
l
A
:
L
o
w
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
5
,
6
0
3
)
W
e
i
g
h
t
i
n
g
s
o
n
e
a
r
n
i
n
g
s
c
o
m
p
o
n
e
n
t
s
C
o
m
p
a
r
i
s
o
n
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
A
C
C
R
t
D
C
A
S
H
t
(
1
)
D
I
S
T
E
Q
t
(
2
)
D
I
S
T
Dt
(
3
)
(
1
)
-
(
2
)
(
1
)
-
(
3
)
(
2
)
-
(
3
)
H
i
s
t
o
r
i
c
a
l
C
o
e
f
?
c
i
e
n
t
0
.
5
1
0
0
.
5
2
2
0
.
7
0
8
0
.
4
9
2
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
6
0
.
0
3
0
0
.
2
1
6
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
9
0
.
0
0
8
0
.
0
0
8
0
.
0
1
0
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
9
2
7
.
9
8
7
.
1
1
4
8
1
.
5
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
8
)
(
0
.
0
0
0
)
I
n
v
e
s
t
o
r
s
C
o
e
f
?
c
i
e
n
t
1
.
1
1
2
1
.
0
0
4
0
.
8
9
7
0
.
5
5
6
D
i
f
f
e
r
e
n
c
e
0
.
1
0
7
0
.
4
4
8
0
.
3
4
1
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
6
7
0
.
0
6
3
0
.
0
5
3
0
.
0
7
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
6
.
1
7
3
2
.
3
8
2
4
.
5
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
A
n
a
l
y
s
t
s
C
o
e
f
?
c
i
e
n
t
0
.
6
0
4
0
.
5
5
8
0
.
7
3
5
0
.
5
7
0
D
i
f
f
e
r
e
n
c
e
2
0
.
1
7
7
2
0
.
0
1
2
0
.
1
6
5
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
9
0
.
0
0
9
0
.
0
0
8
0
.
0
1
1
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
7
2
2
.
1
2
0
.
9
1
2
4
0
.
8
1
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
3
4
1
)
(
0
.
0
0
0
)
P
a
n
e
l
B
:
h
i
g
h
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
8
,
6
0
2
)
W
e
i
g
h
t
i
n
g
s
o
n
e
a
r
n
i
n
g
s
c
o
m
p
o
n
e
n
t
s
C
o
m
p
a
r
i
s
o
n
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
A
C
C
R
t
D
C
A
S
H
t
(
1
)
D
I
S
T
E
Q
t
(
2
)
D
I
S
T
Dt
(
3
)
(
1
)
-
(
2
)
(
1
)
-
(
3
)
(
2
)
-
(
3
)
H
i
s
t
o
r
i
c
a
l
C
o
e
f
?
c
i
e
n
t
0
.
4
7
8
0
.
5
1
2
0
.
6
8
3
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
1
7
1
2
0
.
0
2
5
0
.
1
4
6
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
7
0
.
0
0
8
0
.
0
0
7
0
.
0
0
8
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
2
8
4
.
7
8
.
5
6
3
8
6
.
5
2
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
3
)
(
0
.
0
0
0
)
I
n
v
e
s
t
o
r
s
C
o
e
f
?
c
i
e
n
t
0
.
9
3
4
0
.
8
3
1
0
.
7
8
6
0
.
7
5
6
D
i
f
f
e
r
e
n
c
e
0
.
0
4
5
0
.
0
7
5
0
.
0
3
0
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
4
7
0
.
0
4
8
0
.
0
4
2
0
.
0
5
2
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
2
.
3
6
2
.
1
9
0
.
4
1
0
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
1
2
5
)
(
0
.
1
4
0
)
(
0
.
5
2
1
)
A
n
a
l
y
s
t
s
C
o
e
f
?
c
i
e
n
t
0
.
5
1
7
0
.
4
7
6
0
.
6
0
9
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
1
3
3
2
0
.
0
6
1
0
.
0
7
2
S
t
a
n
d
a
r
d
e
r
r
o
r
0
.
0
0
5
0
.
0
0
6
0
.
0
0
5
0
.
0
0
6
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
3
9
9
.
6
9
4
.
9
4
1
6
8
.
7
6
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
;
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table IV.
The impact of analyst
coverage on historical,
investors and ?nancial
analysts weightings of
the cash components of
earnings (34,205
?rm-years, 1985-2004)
Forecasting
annual earnings
49
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
perceive DCASH
t
as the most persistent cash component for the lower analyst
coverage group. These ?ndings are consistent with the results reported in Table II for
the full sample. Table IV Panel B shows that, for the higher analyst coverage group,
investors’ weighting of DIST
EQ
t
is not signi?cantly different from that of DIST
D
t
. It
appears that, contrary to our expectation, investors recognize the differential
persistence between DIST
EQ
t
and DIST
D
t
only for the lower analyst coverage group.
The overestimation of DCASH
t
by investors, however, is reduced in the higher
coverage group. The large variances in investors’ weightings provide one plausible
reason for our inability to detect investors’ better recognition of differential persistence
of cash components in the higher analyst coverage group.
Turning next to the analysts’ weightings, both panels show that analysts appear to
recognize the differential persistence of the cash components. For example, analysts
weight DIST
EQ
t
as the most persistent among the three cash components in both groups,
which is consistent with the historical relations. The persistence rankings of the cash
components by ?nancial analysts vary only slightly across the analysts coverage
groups. In the higher coverage group, analysts’ rankings of the cash components
correspond to their historical rankings, i.e. DIST
EQ
t
(DCASH
t
) is the most (least)
persistent cash component. In the lower coverage group, analysts’ weightings of
DCASH
t
and DIST
D
t
are statistically undistinguishable. These results support the view
that ?nancial analysts’ ability to recognize the relative persistence between DCASH
t
and
DIST
D
t
is positively associated with the quality of the information environment.
In summary, the results fromTable IVsupport hypothesis 3 (H3): the level of ?nancial
analyst coverage affects bothinvestors’ and ?nancial analysts’ rankings of the persistence
of the cash ?ow components. The extent to which investors and analysts are affected,
however, differ. Financial analysts’ rankings are less affectedthanare investors’ rankings.
For lightly followed ?rms, investors fail to recognize that DIST
EQ
t
is the most persistent
among the three cash ?ow components. For closely followed ?rms, there is no signi?cant
difference among the weightings on the three cash components by investors. On the other
hand, the level of analyst coverage has a positive effect on ?nancial analysts’ rankings of
the three cash components. For lightly followed ?rms, analysts fail to recognize the higher
persistence of DCASH
t
thanDIST
D
t
. For closely followed ?rms, analysts’ rankings of cash
components fully re?ect their historical rankings.
4.4.2 Comparisons of weightings of prior-year cash components for the lower and
higher analyst coverage groups. Table V presents the formal contrasts of the magnitudes
of the weightings reportedinTable IV. The contrasts are providedseparatelyfor the lower
andthe higher analyst coverage groups, inorder totest H4. First, we contrast the historical
and investors’ weightings across the two analyst coverage groups. Panel A of Table V
shows that for the lower analyst coverage group, there is no signi?cant difference (with
p-value of 0.383) between the historical and investors’ weightings of DIST
D
t
, i.e. investors
appear to correctly weight DIST
D
t
. Investors, however, overweight both the DCASH
t
and
the DIST
EQ
t
components. Panel B of Table V shows that investors overweight all three
cash components, for ?rms in the higher analyst coverage group. These results indicate
that the level of ?nancial analyst coverage affects the magnitudes of investors’ weightings
of the persistence of the cash ?ow components.
The comparisons between the historical and analysts’ weightings of the cash
components in Panel A of Table V show that, for the lower analyst coverage group,
analysts overweight all the cash components. Although the over-weightings are
ARJ
21,1
50
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
P
a
n
e
l
A
:
L
o
w
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
5
,
6
0
3
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
V
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
I
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
A
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
I
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
5
1
0
1
.
1
1
2
0
.
6
0
4
D
i
f
f
e
r
e
n
c
e
2
0
.
6
0
2
2
0
.
0
9
4
2
0
.
5
0
8
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
0
2
.
8
2
9
5
.
6
4
7
3
.
8
9
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
2
2
1
.
0
0
4
0
.
5
5
8
D
i
f
f
e
r
e
n
c
e
2
0
.
4
8
2
2
0
.
0
3
6
2
0
.
4
4
6
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
6
8
.
4
3
1
3
.
9
8
5
9
.
7
4
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
7
0
8
0
.
8
9
7
0
.
7
3
5
D
i
f
f
e
r
e
n
c
e
2
0
.
1
8
9
2
0
.
0
2
7
2
0
.
1
6
2
L
-
R
s
t
a
t
i
s
t
i
c
1
2
.
9
4
9
.
9
4
9
.
6
8
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
2
)
(
0
.
0
0
2
)
D
I
S
T
Dt
0
.
4
9
2
0
.
5
5
6
0
.
5
7
0
D
i
f
f
e
r
e
n
c
e
2
0
.
0
6
4
2
0
.
0
7
8
0
.
0
1
4
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
0
.
7
6
4
3
.
8
3
0
.
0
3
(
p
-
v
a
l
u
e
)
(
0
.
3
8
3
)
(
0
.
0
0
0
)
(
0
.
8
5
5
)
P
a
n
e
l
B
:
H
i
g
h
e
r
a
n
a
l
y
s
t
c
o
v
e
r
a
g
e
(
n
¼
1
8
,
6
0
2
)
W
e
i
g
h
t
i
n
g
s
r
e
p
o
r
t
e
d
i
n
T
a
b
l
e
I
V
C
o
n
t
r
a
s
t
s
o
f
w
e
i
g
h
t
i
n
g
s
(
a
)
H
i
s
t
o
r
i
c
a
l
I
n
v
e
s
t
o
r
s
A
n
a
l
y
s
t
s
H
i
s
t
o
r
i
c
a
l
v
s
i
n
v
e
s
t
o
r
s
H
i
s
t
o
r
i
c
a
l
v
s
a
n
a
l
y
s
t
s
A
n
a
l
y
s
t
s
v
s
i
n
v
e
s
t
o
r
s
A
C
C
R
t
0
.
4
7
8
0
.
9
3
4
0
.
5
1
7
D
i
f
f
e
r
e
n
c
e
2
0
.
4
5
6
2
0
.
0
3
9
2
0
.
4
1
7
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
0
9
.
7
5
3
9
.
7
1
9
4
.
0
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
C
A
S
H
t
0
.
5
1
2
0
.
8
3
1
0
.
4
7
6
D
i
f
f
e
r
e
n
c
e
2
0
.
3
1
9
0
.
0
3
6
2
0
.
3
5
5
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
4
6
.
9
7
3
0
.
5
9
5
9
.
8
3
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
E
Q
t
0
.
6
8
3
0
.
7
8
6
0
.
6
0
9
D
i
f
f
e
r
e
n
c
e
2
0
.
1
0
3
0
.
0
7
4
2
0
.
1
7
7
L
-
R
s
t
a
t
i
s
t
i
c
5
.
9
8
1
5
7
.
6
3
1
8
.
2
1
(
p
-
v
a
l
u
e
)
(
0
.
0
1
3
)
(
0
.
0
0
0
)
(
0
.
0
0
0
)
D
I
S
T
Dt
0
.
5
3
7
0
.
7
5
6
0
.
5
3
7
D
i
f
f
e
r
e
n
c
e
2
0
.
2
1
9
0
.
0
0
0
2
0
.
2
1
9
L
i
k
e
l
i
h
o
o
d
r
a
t
i
o
1
8
.
2
6
0
.
0
0
0
1
8
.
7
7
(
p
-
v
a
l
u
e
)
(
0
.
0
0
0
)
(
0
.
9
6
8
)
(
0
.
0
0
0
)
N
o
t
e
s
:
A
C
C
R
t
,
t
o
t
a
l
a
n
n
u
a
l
a
c
c
r
u
a
l
s
f
o
r
y
e
a
r
t
¼
D
N
o
n
-
C
a
s
h
A
s
s
e
t
s
2
D
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
,
w
h
e
r
e
N
o
n
-
C
a
s
h
A
s
s
e
t
s
¼
T
o
t
a
l
a
s
s
e
t
s
(
A
6
)
2
C
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
a
n
d
N
o
n
-
D
e
b
t
L
i
a
b
i
l
i
t
i
e
s
¼
T
o
t
a
l
l
i
a
b
i
l
i
t
i
e
s
(
A
1
8
1
)
2
D
e
b
t
(
A
9
þ
A
3
4
)
D
C
A
S
H
t
,
c
h
a
n
g
e
i
n
c
a
s
h
f
o
r
y
e
a
r
t
.
C
a
s
h
i
s
d
e
?
n
e
d
a
s
c
a
s
h
a
n
d
e
q
u
i
v
a
l
e
n
t
s
(
A
1
)
;
D
I
S
T
E
Q
t
,
c
a
s
h
d
i
s
t
r
i
b
u
t
e
d
t
o
e
q
u
i
t
y
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
E
t
2
C
h
a
n
g
e
i
n
e
q
u
i
t
y
f
o
r
y
e
a
r
t
¼
E
t
2
(
D
T
o
t
a
l
A
s
s
e
t
s
2
D
T
o
t
a
l
L
i
a
b
i
l
i
t
i
e
s
)
f
o
r
y
e
a
r
t
;
D
I
S
T
Dt
,
c
a
s
h
d
i
s
t
r
i
b
u
t
i
o
n
s
t
o
d
e
b
t
h
o
l
d
e
r
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
d
e
b
t
s
f
o
r
y
e
a
r
t
¼
R
e
d
u
c
t
i
o
n
i
n
l
o
n
g
-
t
e
r
m
d
e
b
t
(
A
9
)
þ
R
e
d
u
c
t
i
o
n
i
n
s
h
o
r
t
-
t
e
r
m
d
e
b
t
(
A
3
4
)
,
f
o
r
y
e
a
r
t
.
(
a
)
T
h
e
s
t
a
t
i
s
t
i
c
a
l
t
e
s
t
s
a
r
e
b
a
s
e
d
o
n
s
i
m
u
l
t
a
n
e
o
u
s
n
o
n
-
l
i
n
e
a
r
l
e
a
s
t
s
q
u
a
r
e
e
s
t
i
m
a
t
i
o
n
a
s
u
s
e
d
i
n
M
i
s
h
k
i
n
(
1
9
8
3
)
,
S
l
o
a
n
(
1
9
9
6
)
a
n
d
D
e
c
h
o
w
e
t
a
l
.
(
2
0
0
6
)
Table V.
Contrast of historical,
?nancial analysts, and
investors weightings of
the cash components of
earnings by analyst
coverage group (34,205
?rm-years, 1985-2004)
Forecasting
annual earnings
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant statistically, the magnitudes of analysts’ biases are in general insigni?cant
economically. For example, analysts overweight DIST
EQ
t
by 4 percent ( 2 0.027/
0.708 ¼ 4 percent). Panel B of Table V reports the corresponding results for the higher
analyst coverage group and reveals that analysts correctly weight DIST
D
t
but
underweight DCASH
t
and DIST
EQ
t
. Note that analysts’ mis-weightings behave
differently – overweighting (underweighting) for the lower (higher) analyst coverage.
Since the level of analyst coverage proxies for the quality of the information
environment, one plausible explanation of this difference is that, in a rich information
environment with many competing information sources, ?nancial analysts are able to
reduce their reliance on prior-year earnings components in forecasting annual
earnings.
The ?nal part of Table V compares analysts’ and investors’ weightings. The results
in general provide evidence that ?nancial analysts are better able than investors at
processing the cash components information regardless of the quality of the information
environment. For example, in the lower coverage group, investors’ mis-weighting of
DCASH
t
is 13 times (20.482/20.036 ¼ 13.39) as much as analysts’ mis-weighting. In
the higher coverage group, investors’ mis-weighting of DCASH
t
is reduced to 9 times
(20.319/0.036 ¼ 8.86) as muchas analysts’ mis-weighting. These ?ndings complement
prior literature and suggest that ?nancial analysts, as information intermediaries, are
less biased than investors in processing not only the accruals but also the cash
components of earnings.
Overall, the results reported in Table V support our hypothesis 4 (H4): the level of
analyst coverage affects the magnitudes of investors’ and?nancial analysts’ weightings of
the cash?owcomponents whencomparedto the historical weightings. The magnitudes of
bothinvestors’ andanalysts’ mis-weightings of the cashcomponents are generallysmaller
for ?rms in the higher analyst coverage group. Consistent with prior literature, our results
show that higher analyst coverage indicates more ef?cient use of information.
5. Summary and conclusions
This study examines whether ?nancial analysts mislead investors in recognizing the
differential persistence of the cash ?ow components of earnings, de?ned by Dechowet al.
(2006), in forecasting annual earnings. The three cash components examined are cash
retained by the ?rm, cash distributed to equity holders and cash distributed to debt
holders. Dechow et al. (2006) document that investors correctly recognize the differential
persistence of cash distributed to equity holders and debt holders. On the other hand,
investors overestimate the persistence of cash retained in the ?rm. Our study extends that
inquiry and examines how?nancial analysts utilize the information contained in the three
cash components in forming their earnings forecasts, in order to determine whether
analysts’ biases contribute to investors’ biases. This inquiry is important to academics as
well as to the investment community. As information intermediaries, ?nancial analysts
play a prominent role in the ?nancial market. Consequently, the ability of analysts to
incorporate value-relevant information in their published expectations may impact
securities prices. Furthermore, our study partitions the sample based on the quality of the
information environment (proxied by the level of analyst coverage) in order to evaluate
whether the quality of the information environment affects ?nancial analysts’ and
investors’ weightings of differential cash components.
ARJ
21,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Consistent with our predictions, we ?nd that ?nancial analysts’ weightings of the three
cash ?ow components are more closely aligned with the historical relations than are the
corresponding weightings by investors, both in direction and in magnitude. Speci?cally,
?nancial analysts correctly recognize that cash distributed to equity holders is the most
persistent among the three cash ?ow components, whereas investors rank cash retained
by the ?rm as the most persistent cash component. In addition, the degree of analysts’
mis-weightings is economically small and much lower than the degree of investors’
mis-weightings. These results support our ?rst and second hypotheses.
Moreover, we ?nd that the level of ?nancial analyst coverage has an impact on the
weightings of the cash ?ow components by both investors and analysts, both in
direction and in magnitude, supporting our third and fourth hypotheses. The extent of
both investors’ and analysts’ mis-weightings of the cash components is generally
smaller for ?rms with greater levels of analyst following, which is widely used as a
proxy for the quality of a given ?rm’s information environment. Finally, we show that
cash distributed to equity holders is most persistent to subsequent year’s earnings,
regardless of the quality of the information environment.
Our ?ndings suggest that ?nancial analysts donot misleadinvestors inrecognizingthe
differential persistence of the cash components in forecasting annual earnings. Rather,
analysts’ bias inweightingthe cashcomponents of earnings is at best a partial explanation
for investors’ bias. Other security market inef?ciencies that are unrelated to ?nancial
analysts’ earnings forecasts underlie at least part of investors’ mis-weightings of cash
components. Earlier studies have shown that analysts are less biased than investors in
weightings the accrual component of earnings. This study extends that research, and
indicates that ?nancial analysts, as information intermediaries, are less biased than
investors in processing not only the accrual but also the cash components of earnings.
Notes
1. See Dechow et al. (2006), pp. 5-7 for a detailed explanation of this decomposition of income.
They scale all earnings components variables by average total assets.
2. Our ?ndings in both Panel A and Panel B of Table II are directionally the same as the main
persistence results reported by Dechow et al. (2006).
3. Note that the standard errors of the weightings by investors are about seven times as much
as those of the historical weightings, while the standard errors of analysts’ weightings are
nearly identical to those of the historical weightings. This suggests that the weightings of
earnings components by investors vary much more widely than either historical or analysts’
weightings.
4. Dechow et al. (2006) present similar results. Their Table V Panel D shows the investor’s
weightings (valuation coef?cients) of the three cash components are all signi?cantly greater
than the historical weightings (forecasting coef?cients). For example, the historical weight of
cash distributed to equity holders is 0.784, investors’ weight is 0.839, and the L-R statistic for
testing the equality of the two coef?cients is 18.35 ( p-value ¼ 0.001).
5. As noted earlier, the standard errors of analysts’ weightings are much lower (about 7 times)
than those of investors’ weightings. The lowstandard error may contribute to the statistically
signi?cant difference between historical and analysts’ weightings of cash distributed to
equity holders and cash distributed to debt holders, even though the magnitudes of the
differences are rather small.
Forecasting
annual earnings
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Abarbanell, J. and Bernard, V. (1992), “Tests of analysts’ overreaction/underreaction to earnings
information as an explanation for anomalous stock price behavior”, Journal of Finance,
Vol. 47 No. 3, pp. 1181-207.
Abarbanell, J. and Bushee, B. (1998), “Abnormal returns to a fundamental analysis strategy”,
The Accounting Review, Vol. 73 No. 1, pp. 19-45.
Bernard, V. and Thomas, J. (1989), “Post-earnings-announcement drift: delayed price response or
risk premium?”, Journal of Accounting Research, Vol. 27 No. 3, pp. 1-36.
Bernard, V. and Thomas, J. (1990), “Evidence that stock prices do not fully re?ect the implication
f current earnings for future earnings”, Journal of Accounting and Economics, Vol. 13,
pp. 305-40.
Bhattacharya, N. (2001), “April Investors’ trade size and trading responses around earnings
announcements: anempirical investigation”, The Accounting Review, Vol. 76 No. 2, pp. 221-44.
Bradshaw, M., Richardson, S. and Sloan, R. (2001), “June Do analysts and auditors use
information in accruals?”, Journal of Accounting Research, Vol. 39 No. 1, pp. 45-74.
Dechow, P. (1994), “Accounting earnings and cash ?ows as measures of ?rm performance:
the role of accounting accruals”, Journal of Accounting and Economics, Vol. 18 No. 1,
pp. 3-42.
Dechow, P., Richardson, S. and Sloan, R. (2006), “The persistence and pricing of the cash
component of earnings”, working paper, University of Michigan, Ann Arbor, MI.
Elgers, P., Lo, M. and Pfeiffer, R. (2001), “Delayed security price adjustments to ?nancial
analysts’ forecasts of annual earnings”, The Accounting Review, Vol. 76 No. 4, pp. 613-32.
Elgers, P., Lo, M. andPfeiffer, R. (2003), “Analysts’ vs investors’ weightings of accruals inforecasting
annual earnings”, Journal of Accounting and Public Policy, Vol. 22 No. 3, pp. 255-80.
Frankel, R. and Lee, C. (1998), “Accounting valuation, market expectation, and cross-sectional
stock returns”, Journal of Accounting and Economics, Vol. 25 No. 3, pp. 283-319.
Hong, H., Lim, T. and Stein, J. (2000), “Bad news travels slowly: size, analyst coverage, and the
pro?tability of momentum strategies”, Journal of Finance, Vol. 55 No. 1, pp. 265-95.
Mishkin, F. (1983), A Rational Expectations Approach to Macroeconomics, The University of
Chicago Press, Chicago, IL.
Ohlson, J. (1995), “Earnings, book values and dividends in equity valuation”, Contemporary
Accounting Research, Vol. 11 No. 2, pp. 661-87.
Sloan, R. (1996), “Do stock prices fully re?ect information in accruals and cash ?ows about future
earnings?”, The Accounting Review, Vol. 71 No. 3, pp. 289-315.
Walther, B. (1997), “Investor sophistication and market earnings expectations”, Journal of
Accounting Research, Vol. 35 No. 2, pp. 157-79.
Corresponding author
Le (Emily) Xu can be contacted at: [email protected]
ARJ
21,1
54
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. May H. Lo, Le (Emily) Xu. 2013. Regulation FD and analysts’ vs. investors’ weightings of the cash
components of earnings. Research in Accounting Regulation 25, 169-184. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_251893383.pdf