Description
Recent research examines the implications of
components of accruals for future profitability.
Because the persistence of earnings varies with
the level of company profitability
Accounting Research Journal
Components of Accruals, Losses and Future Profitability
Hai Wu Neil Fargher
Article information:
To cite this document:
Hai Wu Neil Fargher, (2007),"Components of Accruals, Losses and Future Profitability", Accounting Research J ournal, Vol.
20 Iss 2 pp. 96 - 110
Permanent link to this document:
http://dx.doi.org/10.1108/10309610780000695
Downloaded on: 24 January 2016, At: 21:02 (PT)
References: this document contains references to 0 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 469 times since 2007*
Users who downloaded this article also downloaded:
Arifur Khan, Paul Mather, (2013),"The value of executive director share ownership and discretionary accruals", Accounting
Research J ournal, Vol. 26 Iss 1 pp. 35-55 http://dx.doi.org/10.1108/ARJ -02-2012-0011
Prodosh Simlai, (2014),"Firm characteristics, distress risk and average stock returns", Accounting Research J ournal, Vol. 27
Iss 2 pp. 101-123 http://dx.doi.org/10.1108/ARJ -06-2012-0046
Kim Watty, Satoshi Sugahara, Nadana Abayadeera, Luckmika Perera, J ade McKay, (2014),"Towards a Global Model of
Accounting Education", Accounting Research J ournal, Vol. 27 Iss 3 pp. 286-300 http://dx.doi.org/10.1108/ARJ -08-2013-0054
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
96
Components of Accruals, Losses and Future
Profitability
Hai Wu and Neil Fargher
Department of Accounting and Finance
Division of Economic and Financial Studies
Macquarie University
Abstract
Recent research examines the implications of
components of accruals for future profitability.
Because the persistence of earnings varies with
the level of company profitability, we expect
differences between profitable and loss-making
companies in the association between
components of accruals and future profitability.
Using the approach adopted by Richardson,
Sloan, Soliman and Tuna (2006) we find
evidence suggesting that the components of
accruals related to revenue growth and to change
in asset turnover are less persistent than the cash
flow component of earnings for profitable
Australian companies. For loss-making
companies, however, the persistence of the
accrual component of earnings is found to be
higher than for the cash flow component of
earnings, suggesting that the accrual component
is more informative than the cash flow
component in explaining period ahead
profitability for many currently unprofitable
companies.
1. Introduction
A basic premise of accrual accounting is that it
provides a more timely and relevant performance
measure than cash flows through a better
measure of revenues and expenses. Accounting
Acknowledgements: We wish to acknowledge the helpful
comments fromGraeme Harrison, Geoff Loudon, Farshid
Navissi, Alan Ramsay, Edward Watts, Peter Wells, Sue
Wright and participants at the Asian Academic Accounting
Association Conference 2006 and the 2007 UTS Summer
Accounting Research Conference. All errors remain the
responsibility of the authors. We gratefully acknowledge the
use of data supplied by the Securities Industry Research
Centre of Asia-Pacific (SIRCA) on behalf of Aspect
Financial and by the Centre for Research in Finance.
Key Words: Accruals, Losses, Persistence
accruals can reduce uncertainty about future
earnings and allow investors to more accurately
predict future earnings (Kang 2005).
Sloan (1996) shows that the accrual
component of earnings is less persistent1 than
the cash flow component of earnings in
explaining period ahead profitability.
Richardson, Sloan, Soliman and Tuna (2006,
hereafter RSST) suggest that the two primary
factors explaining the lower persistence of the
accrual component of earnings are the
diminishing marginal returns embedded in new
investments and accounting distortions. They
decompose the accrual component of earnings
into a growth component and an efficiency
component. The growth component measures
the effect of investment growth on the
persistence of accruals. To the extent that the
efficiency component drives the observed
properties of accruals, it is inferred that the
efficiency component measures the effect of
accounting distortions (RSST 2006, page 719).
Their results suggest that both components
contribute to the lower persistence of the
accrual component of earnings relative to the
cash flow component.
Prior research suggests that the persistence of
earnings varies with profitability (Hayn 1995,
Joos and Plesko 2005, Balkrishna et al. 2007). If
earnings persistence varies between profitable
and loss-making companies, then we would also
expect differences in the persistence of the
components of accruals. We extend the
previous research to examine the differential
implications of current components of earnings
1 In this literature, “persistence” is used to refer to the
implication of accruals for future profitability. That is,
persistence refers to the association between this period
accruals and next period profitability.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
97
for future profitability for companies with
current period losses.
For profitable Australian companies, we find
evidence consistent with the accrual component
being less persistent than the cash flow
component of earnings. When the accruals
component is decomposed, we find evidence
suggesting that the growth and efficiency
components of accruals exhibit lower
persistence than the cash flow component of
earnings. For loss-making companies, the
persistence of the accrual component of
earnings is found to be higher than the cash
flow component of earnings. This result extends
to the components of accruals. The results are
consistent with the argument that loss-making
companies in the development stage tend to
have accruals that are more informative to next
period profitability than is the cash flow
component of earnings. We also find that the
higher persistence of the efficiency components
of accruals is mainly attributable to non-current
operating accruals.
This paper is organised as follows. The next
section provides a brief literature review.
Section three defines the variables and outlines
the research design. Section four describes the
sample. Section five discusses the results.
Finally the overall conclusions and implications
for future research are presented.
2. Relation to Previous Research
2.1 Persistence of Accruals
Following Sloan (1996), explanations for the
lower persistence of accruals can be grouped
into two categories: the diminishing marginal
returns embedded in new investment and
accounting distortions. Fairfield et al (2003)
show that diminishing marginal returns affect
accrual earnings to a greater extent than cash
earnings. Diminishing marginal returns imply
that for high profit companies, returns on the
real investment acquired in the current period
will be lower in the next period due to
competition. This will drag down the average
return for the company’s real investments
towards the long-term mean return. On the other
hand, investments generating lower returns in
the current period will be replaced by
investments with better returns in the current or
future period. This will drive up the average
return towards the long-term mean.
Accounting distortions are caused by the use
of inappropriate accounting. The selection of
inappropriate accounting policies can be either
intentional (earnings management) or
unintentional (estimation error). Xie (2001)
shows that the lower persistence of the accrual
component of earnings is attributable to the
discretionary accruals, which measure the level
of earnings management activities (Jones 1991).
Chan et al. (2006) find that the negative relation
between accruals and future stock returns is
mainly caused by the discretionary component
of accruals.
On the other hand, unintentional accrual
estimation error can also cause the lower
persistence of the accrual component of
earnings (RSST 2005; Dechow and Dichev
2002). RSST (2005) argue that the lower
persistence of accruals is caused by the
estimation error embedded in the accrual
accounts. They argue that due to the nature of
accruals, the persistence of accrual earnings
should be theoretically less than cash earnings.
They decompose total accruals into a range of
aggregate accounting items based on their
relative reliability level. Their results suggest
that the accrual components that are less reliable
(higher estimation error) have lower persistence.
RSST (2006) extend the literature by
decomposing total accruals into the growth and
efficiency components in order to discriminate
effects of two alternative explanations on the
persistence of accruals. The growth component
captures the effect of growth in real investment,
while the efficiency component accounts for
the estimation error in accruals caused either
by the intentional or unintentional adoption
of inaccurate accounting valuation policies.
They find that both components contribute
to the lower persistence of accruals. This
decomposition approach forms the basis for our
analysis examining the persistence of accruals
for Australian companies and the influence of
reported losses on the association between
accruals and persistence.
2.2 Loss-making Companies
Evidence from prior research suggests
differences in persistence of accruals between
companies currently making losses and
profitable companies. Hayn (1995) documents
an insignificant earnings-return relation,
suggesting that the market expects negative
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
98
earnings to be transitory for loss-making
companies. Basu (1997) shows that bad news is
recognised in a more timely manner than good
news, due to accounting conservatism. This
suggests that bad news is reported with
relatively larger errors than good news, as more
timely news is usually more relevant but less
reliable (Taylor and Taylor 2003). It follows
that conservatism should result in earnings
being more transitory and less persistent (Basu
1997). Givoly and Hayn (2000) show that the
increase in the frequency of negative earnings is
mainly caused by negative accruals, and that the
cash flow component of earnings is less
influenced by conservative accounting practice
than the accrual components (Givoly and Hayn
2000). If losses reflect excessive conservatism,
then we expect earnings to be less persistence
for loss-making companies, and this should be
partially attributable to the accruals components
arising from conservative accounting.
Many loss-making companies are, however,
small companies in the early stage of their
development with very limited resources (Klein
and Marquardt 2006). These companies are
more likely to make consistent losses in their
early stage of development, and profits are only
expected after several years. The impact is that
these loss-making companies are more likely to
have a higher persistence in earnings and
accruals compared to profitable companies.
While both accounting conservatism and
stage in the investment cycle predict differences
in the persistence level of accruals between
loss-making and profitable companies, the
direction predicted is ambiguous. The results
are determined by the extent to which losses are
caused by these two factors. By explicitly
considering the impact of losses on the
persistence of earnings, and its components, our
study provides evidence on the relative
importance of these explanations.
2.3 Australian Evidence
Australian studies have largely concentrated on
the area of earnings management. Typically
earnings management studies examine
discretionary accruals in relation to a range of
factors that are expected to affect the intensity
of management manipulation. (Eddey and
Taylor 1999, Godfrey et al 2003, Wells 2002,
Koh 2003, Kelley et al 2004, Davidson et al
2005, and Coulton et al 2005).
There are fewer studies examining the
persistence of changes in earnings or accruals.
Taylor and Taylor (2003) examine the presence
of conservatism in Australian accounting
practices.
2
Following Basu (1997), they test
the persistence of changes in earnings, where
lower persistence of negative earnings change
should indicate the presence of accounting
conservatism. The results from Taylor and
Taylor (2003) suggest the possibility that loss-
making companies exhibit differences in
earnings persistence from profitable companies.
Coulton et al (2005) provide evidence for the
persistence of discretionary and non-
discretionary accruals to show the effectiveness
of discretionary accruals in capturing earnings
management. Their results are mixed. The
accrual component of earnings is found to be
less persistent than the cash flow component of
earnings. However, unlike Xie (2001), the
results for the difference in the persistence level
of discretionary and non-discretionary accruals
is only found for one of the three measures of
discretionary accruals used. The authors argue
that the result could be partially explained by
the relatively large portion of loss-making
companies in the sample.
Oei et al (2006) examine the persistence
level of accruals and its components for the top
300 Australian companies by market
capitalisation. Their results suggest that accruals
and the components of accruals are less
persistence than the cash flows component of
earnings and that there are differences between
the persistence levels of the components of
accruals. Kean and Wells (2006) decompose
return on equity on three bases, cash/accruals,
operating/financing/other, and financial ratios.
They show that under all three decomposition
bases, the components of earnings exhibit
different levels of persistence. Specifically, the
accrual component of ROE is shown to be less
persistent than the cash flow component.
Recent Australian research also provides
some evidence regarding loss-making
companies. Balkrishna et al (2007) and
Lonergan and Wells (2006) both document an
increasing portion of companies reporting losses
over the previous 10 to 30-year period. Watson
2 In Taylor and Taylor (2003) accounting conservatismis
defined as the timely reflection of bad news compared to
good news.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
99
and Wells (2005) find that earnings are not
useful in explaining stock returns for loss-
making companies supporting the transitory
nature of losses. Balkrishna et al (2007) find that
accounting practices are more conservative
among loss-making companies than profitable
companies. This provides evidence supporting
the view that loss-making companies’ earnings
potentially include greater accounting distortions
and are less persistence than profitable
companies. We further examine the issue of the
relation between the components of earnings and
period ahead profitability for loss-making
companies.
3. Variable Definitions and Model
Specification
3.1 Defining accruals
We initially adopt the operating accruals
definition used by RSST (2006). They measure
total operating accruals as the change in net
operating assets deflated by lagged operating
assets (RSST, 2006, page 722).
1
1
?
?
?
=
t
t t
t
NOA
NOA NOA
TACC
Net operating asset (NOA) is defined as the
difference between non-cash operating assets
and operating liabilities. Operating assets is total
assets net of cash and equivalents and total
investments, while operating liability is total
liabilities net off short-termand long-termdebts,
as well as income tax liability.
3
This accruals
measure is in contrast to the more widely used
definition of accruals as the changes in non-cash
working capital net of depreciation expenses.
RSST (2006) argue that the traditional accruals
measure omits many non-current operating
accruals, which are sources of growth and are
affected by accounting distortions. We initially
follow the RSST (2006) approach but then
consider the sensitivity of the results to this
definition of accruals.
3 NOA is therefore defined as: Total Assets (Aspect Item
#5090) – Cash and Short Terminvestment (Aspect Item
#4990) – Investment and Advance (Aspect Item#5070) -
[Total Liabilities (Aspect Item#6040) – Debt in Current
Liabilities (Aspect Item#6000) – Long-termdebt (Aspect
Item#6020) – Income tax liability (Aspect Item#319)].
3.2 The growth and efficiency components
of accruals
Following RSST (2006), the growth component
of accruals is defined as change in sales. The
efficiency component is measured as change in
asset turnover. They are defined as follows:
1 ?
?
=
t
t
Sales
Sales
Growth
;
t
t
AT
AT
Efficiency
?
= , where
t
t
NOA
Sales
AT =
This type of analysis assumes that the
increase in real investment in the current period
causes production growth and subsequently
sales growth. Asset turnover measures the
efficient use of net operating assets to generate
revenue. Change in asset turnover is caused by
disproportionate changes in Sales and NOA.
This can be caused by accounting distortions,
which causes NOA to be valued with estimation
error, or change in real efficiency. RSST (2006)
show that total operating accruals can be
parsimoniously decomposed as follow:
t t t t
t
t
t
t
t
t
t
t
t
t
AT SG AT SG
AT
AT
Sales
Sales
AT
AT
Sales
Sales
NOA
NOA
TACC
? ? ? ? ? ? ? =
?
?
?
?
?
?
?
=
?
=
? ?
?
) ( ) (
1 1
1
In this decomposition, sales growth is
positively related to total operating accruals,
while the change in the efficient use of assets is
negatively related to total operating accruals. In
the absence of sales growth, reduction in asset
turnover is solely caused by increase in net
operating assets, which measures accruals.
There is a third term in the decomposition,
which is an interaction term. It is argued that
economies of scale imply a positive correlation
between sales growth and change in efficiency.
4
3.3 Model specification
Following RSST (2006, page 733) we model
the period ahead accounting rate of return as:
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ?
(1)
4 Economies of scaleimply a diminishing marginal cost to
the production in the long term. As sales grow, production
levels also grow. The cost of producing additional
products will drop, requiring less capital to be used. This
will effectively increase asset turnover. Therefore, with an
increase in sales growth, asset turnover also increase.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
100
where RNOA
t
is return on NOA for period t,
defined as Net Operating Income (NOI) in
period t deflated by lagged NOA. NOI is
sourced from the Aspect Financial database as
EBIT (Aspect Item #8012) minus Other
Revenue (Aspect Item #7080). This specification
is explained by RSST (2005, page 457) as a
direct test of the ‘relative persistence’ of
accruals. If the accrual component of earnings is
less persistent than the cash flow component of
earnings, ?
2
is predicted to be less than zero.
5
An alternative formof persistence model is
also considered. An interaction term
(RNOA
t
*TACC
t
) is included to capture the
relation between the interaction of this period
accruals and profitability, and one-year ahead
profitability. That is, the RSST (2006) model
does not allow for the interaction between
profitability and accruals. The basic model
including this interaction term is expressed as
follows:
2 1 0 1 +
+ + =
t t t
TACC RNOA RNOA ? ? ?
1 3
*
+
+ +
t t t
TACC RNOA ? ?
(1a)
RSST (2006) restates equation (1) to separate
the growth and efficiency components of total
operating accruals in order to examine their
individual effects on the differential persistence
of cash and accruals:
3 2 1 0 1 +
? ? + + =
t t t t
AT SG RNOA RNOA ? ? ? ?
1 4 +
+ ? ? ?
t t t
AT SG ? ?
(2)
Based upon U.S. evidence the coefficients on
sales growth and change in asset turnover are
predicted to be less than zero (?
2
<0 and ?
3
<0).
In order to examine the difference in
persistence of accruals between loss-making
and profitable companies, we define an
indicator variable for loss firms. Loss takes the
value of one if current year’s earnings are
negative (RNOA
t
<0) and zero otherwise.
Losses can affect both the intercept and the
slope coefficients. Extending the model to
consider the growth and efficiency components
of accruals, and the differential persistence of
unprofitable companies, results in the following
model:
3 2 1 0 1 +
? ? + + =
t t t t
AT SG RNOA RNOA ? ? ? ?
5 4
+ + ? ? ?
t t
Loss AT SG ? ?
6
*
t
RNOA Loss ?
8 7
* * ? ? +
t t
AT Loss SG Loss ? ?
1 9
*
+
+ ? ? ?
t t t
AT SG Loss ? ?
(3)
To examine the incremental difference in the
persistence of unprofitable companies we test
whether ?
7
and ?
8
are significantly different from
zero.
4. Sample Selection and Description
4.1 Sample Selection
Table 1 summarises the sample selection
process. We identified 22,014 company-year
observations from the ASPECT Financial
Database for the period 1987 to 2005. We
perform four adjustments. (1) We eliminated all
firm-year observations with missing EBIT.
6
(2)
We exclude all firm-year observations with zero
sales revenue.
7
(3) Because all financial
statement variables are measured in terms of
percentage change, and the dependent variable
is one-year ahead accounting rate of return,
observations without matched one-year-lagged
and one-year-ahead data are excluded. (4) Upon
review of the data by year we further excluded
all observations fromthe four years prior to
1990 because of the relatively narrow coverage
in these early years and a clear bias towards
larger companies. We obtain 7,432 company-
year observations representing 1,249
companies. After matching our sample to the
return data obtained from the CRIFS-SPPR
database, our final sample contains 7,285
company-year observations covering 1,240
companies for the period 1990 to 2004. Yearly
sample variation ranges from 305 observations
in 1990 to 637 observations in 2002. Consistent
with previous research of this type it must be
acknowledged that the data requirements tend to
exclude smaller, less profitable companies.
Further, to the extent that companies that do not
survive a consecutive three-year period are
eliminated, the sample reflects more profitable
companies and the results must be interpreted
with respect to this limitation.
5
If
1 2 1 0 1
) (
+ +
+ + ? + =
t t t t t
TACC TACC RNOA RNOA ? ? ? ? ,
where ?
1
is the persistence of cash flows and ?
2
is the
persistence of accruals then this can be rewritten as
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ? where ?
2
>0 is a
test of ?
1
– ?
2
.
6 This step also excludes financial services companies as
EBIT is not meaningful for these entities.
7 The calculations of the growth and efficiency
components of accruals require current and previous
sales revenue to be strictly non-zero.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
101
Table 1
Sample Selection
Companies Company-Years
Total observations from the Aspect Financials Database 2,836 22,014
Excluding observations with missing EBIT 2,342 18,206
Excluding observations with missing Trading Revenue 1,996 14,405
Excluding all firm-year with a zero value for Trading Revenue 1,787 13,156
Excluding all firm-year data without matching lagged data 1,500 9,511
Excluding all firm-year data without matching one-year ahead data 1,249 7,453
Excluding all firm-year data before 1990 (due to small coverage
before 1990) 1,249 7,432
Final sample after excluding all firm-years without the matching
return data from the CRIFS SPPR database. 1,240 7,285
Table 2
Descriptive Statistics
Panel A: Descriptive Statistics
Variables Mean Median Std. Dev % Positive Min Max
Number of
Observations
TACC
t
0.175 0.045 0.559 58.30% -0.586 1.872 7,285
SG
t
0.301 0.092 0.751 67.04% -0.511 2.804 7,285
?AT
t
-0.084 0.038 0.660 55.62% -2.150 0.801 7,285
RNOA
t
-0.138 0.032 0.551 55.68% -1.897 0.509 7,285
Panel B: Pearson Correlation Matrix
TACC
t
SG
t
?AT
t
RNOA
t
RNOA
t+1
TACC
t
–
0.264
(0.0001)
-0.566
(0.0001)
-0.030
(0.0118)
0.085
(0.0001)
SG
t
0.264
(0.0001)
–
0.387
(0.0001)
-0.063
(0.0001)
-0.073
(0.0001)
?AT
t
-0.566
(0.0001)
0.387
(0.0001)
–
0.124
(0.0001)
0.011
(0.3600)
RNOA
t
-0.030
(0.0118)
-0.063
(0.0001)
0.124
(0.0001)
–
0.708
(0.0001)
RNOA
t+1
0.085
(0.0001)
-0.073
(0.0001)
0.011
(0.3600)
0.708
(0.0001)
–
Notes to Table:
Variable Definitions.
TACC
t
=total accruals in period t, defined as the change in Net Operating Asset (NOA) deflated by the lagged NOA
[(NOA
t
– NOA
t-1
)/ NOA
t-1
];
SG
t
=sales growth at period t [(Sales
t
– Sales
t-1
)/Sales
t-1
];
?AT
t
=the change in Asset Turnover [(AT
t
- AT
t-1
)/AT
t
], where AT
t
is equal to Sales
t
/ NOA
t
;
RNOA
t
=return on Net Operating Assets for period t, defined as Net Operating Income (NOI) in period t deflated by
lagged Net Operating Assets (NOA), where NOI is EBIT – Other Revenue;
NOA =Operating Assets – Operating Liabilities, refer text for details;
Loss =1 if RNOA
t
<0, otherwise 0.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
102
4.2 Descriptive Statistics
Table 2 presents the descriptive statistics for the
variables examined in this research. All variables
are winsorized at the 95% and 5% level to
reduce the impact of extreme observations on
the distribution of regression residuals. Panel A
of Table 2 presents the univariate statistics. The
mean values for accruals (TACC) and sales
growth (SG) are positive, while the mean value
for change in efficiency (?AT) is negative. The
mean accruals measured this way is 17.5%. This
compares to the mean for RSST (2006) of
16.4%. Around 55% of the companies in the
sample experience positive return on net
operating assets (RNOA) in either current period
or next period or both. The mean values for
RNOA
t
and RNOA
t+1
are however negative. This
is not the case for U.S. samples (e.g. Sloan 1996;
Xie 2001) and potentially influences the
persistence of accruals.
Panel B in Table 2 presents the pairwise
correlations between variables. Most of the
correlations are significant at the one percent
level, except for the correlation between the
efficiency component of accruals (?AT) and
future earnings (RNOA
t+1
). As would be
expected, there is a strong positive correlation
between current and future earnings. The sales
growth (SG) and change in efficiency (?AT) are
positively correlated. RSST (2006) argue that
this positive correlation supports the argument
for economies of scale.
Table 3
The Distributions of Firm-Year Observations with
Respect to Earnings Patterns
Company size
Size classification 1 Size classification 2
Earnings pattern
(2)
Small
(3)
Large
(4)
Total
(5)
Small
(6)
Medium
(7)
Large
(8)
Total
Consecutive losses
(Row Percentage)/
[Column Percentage]
1,802
(72%)
[49%]
717
(28%)
[20%]
2,519
[35%]
1,108
[62%]
1,138
[31%]
273
[15%]
2,519
[35%]
Reversing losses 381
(54%)
[10%]
330
(46%)
[9%]
711
[10%]
171
[10%]
382
[10%]
158
[9%]
711
[10%]
Consecutive
profits
1,092
(33%)
[30%]
2,218
(67%)
[61%]
3,310
[45%]
349
[20%]
1,754
[48%]
1,207
[66%]
3,310
[45%]
Reversing profits 368
(49%)
[10%]
377
(51%)
[10%]
745
[10%]
152
[9%]
409
[11%]
184
[10%]
745
[10%]
Total 3,643
(50%)
3,642
(50%)
7,285
1,780
(24%)
3683
(51%)
1,822
(25%)
7,285
Notes to Table:
The sample is separated into four categories based upon the following pattern of earnings: 1. The consecutive losses
category contains those firm-year observations that make losses in both the current and next period; 2. The consecutive
profits category is made up of firm-year observations making profit in both the current and next period; 3. Firm-year
observations are classified as reversing losses if the current period earnings are negative but subsequently reverse to
positive in the next period; 4. Firm-year observations are classified as reversing profits if negative earnings are reported
in the current period, but subsequently earning negative earnings in the next period.
Two alternate criteria are used to classify companies by company size. The first criterion classifies companies as large
or small. The top 50% firm-year observations by average net operating assets (NOA) in the current period are classified
as large companies. The bottom50% firm-year observations in the sample are classified as small companies. The second
criterion classified companies into three categories. The top 25% companies are classified as large. The bottom25%
companies are classified as small. The middle 50% companies are therefore classified as mediumsize. Both metrics are
based on the average net operating assets (NOA) calculated as the average between the beginning and ending NOA.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
103
4.3 Loss-making companies
Small companies in their development stage are
more likely to make consistent losses (Klein and
Marquardt 2006), leading to higher persistence
levels of earnings and accruals.
8
Table 3
provides descriptive evidence on the relation
between company size and the pattern of
earnings persistence. Four earnings patterns are
summarised: consecutive losses (RNOA
t
<0 and
RNOA
t+1
<0), loss reversion (RNOA
t
<0 and
RNOA
t+1
>0), consecutive profits (RNOA
t
>0
and RNOA
t+1
>0), and profit reversion
(RNOA
t
>0 and RNOA
t+1
<0). By definition,
companies experiencing profit and loss
reversion have lower earnings persistence than
companies making consecutive profits or losses.
Table 3 shows that small companies are as
likely to experience profit and loss reversion as
large companies. Small companies are however
more likely to report consecutive losses than
large companies, as 72% of the consecutive loss
companies are small companies. In addition,
most companies making consecutive profits are
large companies (67%). A decreasing pattern
can also be identified for companies making
consecutive losses with increasing company
size. Specifically, 62% of small companies
report consecutive losses compared to 31% of
medium size companies and 15% of large size
companies. These sample characteristics
support the argument that small companies are
more likely to report losses and more likely to
report consistent losses. We focus our analysis
on differences in earnings persistence between
profitable and loss making companies.
However, clearly, profitability is also related to
firm size.
5. Results
5.1 Persistence of accruals
Table 4 provides initial estimates of persistence
for comparison with previous research. Results
are reported using the Fama-Macbeth (1973)
procedure. We estimate yearly cross-sectional
regressions and report the mean coefficient
across the annual regressions. Column (1) of
8 Unreported statistics show that only 25.37% of large
companies are making losses compared to 50.85% of
small companies, where small is defined as the average
balance of beginning and ending net operating assets less
than the median value.
Panel A provides results for the basic regression
of the accounting rate of return on lagged rate of
return. The adjusted R-Square for the model is
0.47. The persistence coefficient for earnings
(?
1
) is estimated at 0.76 and is significantly less
than one (F=9.16, p-value <0.0001). A value of
the coefficient ?
1
that is less than one supports
the maintained hypothesis in this type of study
that current earnings are mean reverting.
Based upon previous research we would
expect a lower persistence level for the accrual
component of earnings compared to the cash
flow component. The results including an
accrual component are reported in column (2) of
Panel A of Table 4. The coefficient ?
1
is
estimated at 0.76 for the persistence of the cash
flow component of earnings. The coefficient on
accruals (TACC
t
) is significantly greater than
zero (?
2
=0.12, p-value <0.0001) indicating that
the persistence level of the accrual component is
higher than the cash flow component of
earnings. This is contrary to expectations. The
high incidence of loss-making companies in our
sample is a potential explanation. To investigate
the effects of loss-making companies and
observations with extreme change in the
accounting rate of return, we re-estimate the
regression using a reduced sample. We exclude
loss-making companies, which include all
observations with negative RNOA
t
or negative
RNOA
t+1
. The reduced sample size is 3,310. By
excluding all companies with losses, we
systematically reduce the number of
observations with extreme changes in the
accounting rate of return, of which many
observations have low earnings persistence.
Results for the reduced sample set excluding
companies making losses are reported in column
(3) of Table 4 Panel A. The coefficient on TACC
is now significantly negative at less than the one
percent level (?
2
=-0.15), indicating that the
accrual component of earnings is less persistent
than the cash flow component. This is consistent
with prior research using U.S. data, with RSST
(2006) finding a value of -0.131 for this
coefficient. Overall, the results for the reduced
sample restricted to profitable companies, are
generally consistent with those found using U.S.
data.
The specification used by RSST (2006) does
however ignore the interaction between
profitability and accruals. Column (4) reports
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
104
Table 4
Basic measures of persistence
Panel A:
1 1 0 1 + +
+ + =
t t t
RNOA RNOA ? ? ?
(1)
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ?
(2), (3)
1 3 2 1 0 1
*
+ +
+ + + + =
t t t t t t
TACC RNOA TACC RNOA RNOA ? ? ? ? ?
(4)
Full sample Consistent profits sub sample†
Pred.
(1) (2) (3) (4)
RNOA
t
<1 0.7575
**
0.7616
**
0.6930
***
0.7548
**
TACC
t
– 0.1190
***
-0.1480
***
-0.1024
***
RNOA
t
* TACC
t
-0.1139
***
Intercept -0.0596
***
-0.0787
***
0.0620
***
0.0545
***
R
2
Adj. R
2
46.6%
46.5%
48.2%
48.0%
54.0%
53.9%
55.3%
55.2%
Sample Size 7,285 7,285 3,310 3,310
Panel B:
1 4 3 2 1 0 1 + +
+ ? ? ? ? ? + + =
t t t t t t t
AT SG AT SG RNOA RNOA ? ? ? ? ? ?
(1), (2)
1 7 6 5
4 3 2 1 0 1
* * *
+
+
+ ? ? ? ? ?
+ ? ? ? ? ? + + =
t t t t t t t
t t t t t t
AT SG RNOA AT RNOA SG RNOA
AT SG AT SG RNOA RNOA
? ? ? ?
? ? ? ? ?
(3)
Full sample Consistent profits sub sample †
Pred.
(1) (2) (3)
RNOA
t
<1 0.7517
***
0.7290
***
0.7445
***
SG
t
– 0.0521
***
-0.0531
***
-0.0567
***
–?AT
t
– 0.0939*** -0.0625
***
-0.0663
***
–SG
t
*?AT
t
0.0566
***
-0.0321
***
-0.0439
***
RNOA
t
*SG
t
-0.0448
–RNOA
t
*?AT
t
-0.0534
–RNOA
t
*SG
t
*?AT
t
0.0405
Intercept -0.0646
***
0.0553
***
-0.0551
***
R
2
Adj. R
2
48.2%
47.7%
54.5%
54.4%
54.8%
54.7%
Sample Size 7,285 3,310 3,310
Notes to table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for definition of variables.
† The reduced sample set excludes companies making losses in the current or next period.
results for equation (1a) including an interaction
term estimated for the reduced sample. The
coefficient on TACC
t
is still significantly
smaller than zero (?
2
=-0.10). The interaction
term, RNOA
t
*TACC
t
, is negative and significant
(p <0.01) indicating that higher accruals are
associated with lower current profitability and
that this is negatively related to period ahead
profitability.
5.2 Persistence of the growth and efficiency
components of accruals
We extend the analysis to consider the causes of
the differential persistence level between
accruals and cash earnings by decomposing
accruals into three components. Panel B of Table
4 provides empirical results for equation (2). As
shown in column (1), coefficient estimates for
sales growth (SG
t
) and change in asset turnover
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
105
(-?AT
t
) are both unexpectedly positive and
significant using the full sample. Again, the
model is re-estimated using the reduced sample
of profitable companies. For the reduced sample,
both ?
2
and ?
3
become significantly negative at
less than 1% level of significance, suggesting
that both growth and efficiency components
contribute to the lower persistence of accrual
earnings. The expectations for lower persistence
of these components are therefore supported
using the reduced sample of profitable
companies. The values for ?
2
and ?
3
are -0.05 and
-0.06 respectively. RSST (2006) report ?
2
and?
3
to be -0.10 and -0.15 respectively. The results
found in the reduced sample set of Australian
companies are consistent with those reported in
RSST (2006). For completeness we also report a
model including interaction terms in column (3).
Three interaction terms measuring the
interactions between the components of accruals
and earnings are added to equation (2). The
coefficients on SG
t
and -?AT
t
are still
significantly negative. The coefficients on
RNOA
t
*SG
t
and RNOA
t
*?AT
t
are found to be
insignificantly different from zero. For the
reduced sample, the components of accruals
affect the current-future earnings relation
collectively not individually.
To summarise, the pattern of persistence
found in the U.S. data is only supported in
Australia for the reduced sample set of profitable
companies. This suggests that earnings
persistence of profitable companies might
behave differently fromearnings persistence of
loss-making companies.
5.3 Results considering the differential
persistence of accruals for loss-making
companies
To explore the difference in the informativeness
of accruals between loss-making and profitable
companies, we provide regression results for
equation (3) allowing for variation in the
coefficients between profitable and loss-making
companies. These results are reported in Tables
5 and 6. Column (1) of Table 5 reports the
results for the basic earnings persistence
specification. The coefficient on RNOA
t
is
estimated at 0.48 and is significantly less than
one. The incremental intercept term for loss
companies is negative and significant. The
coefficient attached to Loss*RNOA
t
is found to
be significantly positive indicating earnings are
more persistent for loss-making companies.
Table 5
Analysis of the Impacts of Losses on the Persistence of Accruals Earnings
1 3 2 1 0 1
*
+ +
+ + + + =
t t t t
RNOA Loss Loss RNOA RNOA ? ? ? ? ?
(1)
1 5 4 3 2 1 0 1
* *
+ +
+ + + + + + =
t t t t t t
TACC Loss RNOA Loss Loss TACC RNOA RNOA ? ? ? ? ? ? ?
(2)
1 7 6
5 4 3 2 1 0 1
* * *
* *
+
+
+ + +
+ + + + + =
t t t t
t t t t t t
RNOA TACC Loss TACC Loss
RNOA Loss Loss RNOA TACC TACC RNOA RNOA
? ? ?
? ? ? ? ? ?
(3)
Pred. (1) (2) (3)
RNOA
t
<1 0.4796
***
0.5014
***
0.6166
***
TACC
t
–
-0.0170 0.0401
*
RNOA
t
*TACC
t
-0.2854
**
Loss +/– -0.1225
***
-0.1209
***
-0.1064
***
Loss* RNOA
t
+/– 0.2405
***
0.2637
***
0.1566
**
Loss* TACC
t
+/– 0.3842
***
0.1105
**
Loss* TACC
t
* RNOA
t
-0.0838
Intercept 0.0137 0.0141 0.0006
R
2
Adj. R
2
47.4%
47.0%
51.0%
50.5%
52.4%
51.6%
Sample Size 7,285 7,285 7,285
Notes to Table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for definition of variables.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
106
Column (2) reports the estimated coefficients
for a regression allowing differing persistence
between cash and accrual earnings. Firstly, the
coefficient on TACC is estimated to be negative
but not significantly different from zero.
Secondly, the coefficient on Loss*RNOA
t
is
found to be significantly positive. This suggests
that the cash flow component of earnings is more
persistent for loss-making companies. Thirdly,
the coefficient attached to Loss*TACC
t
is
estimated at 0.38 and is significantly different
fromzero. A null hypothesis that there is no
difference in the persistence of the accrual
component of earnings between loss-making and
profitable companies can be rejected (t=11.89,
p<0.0001). The accrual component of earnings is
more persistent than the cash flow component for
loss-making companies.
The analysis is extended to incorporate
interaction terms. In addition to RNOA
t
*TACC
t
,
measuring the impact of accruals on current and
future earnings relation for profitable companies,
Loss* RNOA
t
*TACC
t
is also added to capture the
differential impact of accruals between profitable
and loss-making companies. Results for the basic
loss model incorporating interaction terms
between accruals and earnings are reported in
column (3) of Table 5. This time, the coefficient
on TACC
t
is estimated to be significantly positive
at the ten percent level indicating accruals are
more persistent than cash for profitable
companies. Loss*TACC
t
remains significantly
positive consistent with accruals being more
persistent for loss-making companies than
profitable companies. RNOA
t
*TACC
t
is
significantly negative indicating total accruals
negatively affect the relationship between current
and future for profitable companies. The
coefficient on Loss* RNOA
t
*TACC
t
is not
however significantly different fromzero.
Table 6 extends the analysis to consider the
components of accruals and the differences
between loss-making and profitable companies.
Column (1) of Table 6 presents the regression
results. For profitable companies, the coefficients
on both the sales growth (SG
t
) and the efficiency
(-?AT
t
) component of accruals are estimated to
be negative. However, only the coefficient on the
growth component is found to be significantly
different fromzero at the five percent level. For
loss-making companies, it is shown that the
coefficients estimated for Loss* SG
t
and
-Loss*?AT
t
are both significantly greater than
zero (Loss* SG
t
=0.25 and -Loss*?AT
t
=0.27).
The growth and efficiency components of
accruals are both found to be more persistent than
the cash flow component of earnings for loss-
making companies. An F-test for a pooled
regression (not reported) confirms that for loss-
making companies, both the growth and
efficiency components of accruals are more
persistent than the cash flow component of
earnings (SG
t
+Loss* SG
t
>0 and -?AT
t
+
-Loss*?AT
t
>0).
Results for equation (3) incorporating the
interaction terms between accruals components
and earnings are reported in column (2) of Table
6. For profitable companies, both SG
t
and -?AT
t
are not significantly different from zero. The
coefficient on RNOA
t
*SG
t
is significantly
negative, indicating that the negative impact of
accruals on current and future earnings is mainly
attributable to the growth component of earnings.
For loss-making companies, both Loss*SG
t
and -
Loss*?AT
t
are estimated to be significantly
greater than zero (Loss* SG
t
=0.10 and -Loss*
?AT
t
=0.14), consistent with the accruals
components being more persistent for loss-
making companies.
In summary, the growth and efficiency
components of accruals are both found to be
more persistent than the cash flow component of
earnings for loss-making companies. This is
consistent with the expectation that small
development-stage loss-making companies
exhibit higher accruals persistence.
5.4 Further analyses
5.4.1 Current versus non-current accruals
In order to investigate the source of the
differential persistence level observed for the
efficiency components of accruals between loss-
making and profitable companies, we replace
the efficiency component of accruals in
equation (3) with two disaggregated items.
?AT
t
WC
measures the change in working capital
turnover
9
; and ?AT
t
NCO
measures the change in
non-current operating capital turnover
10
. Results
9 AT
t
WC
is calculated as Sales
t
/WC
t
. WC
t
is the working
capital at time t, defined as Current Assets – Cash and
Equivalents – Short-termInvestment – Current Liabilities
+Current Debts +Tax Payable.
10 AT
t
NCO
is calculated as Sales
t
/NCO
t
. NCO
t
is the non-
current operating capital at timet, defined as Non-current
Assets – Long-termInvestment – Non-current Liabilities +
Non-Current Debt.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
107
Table 6
Analysis of the Impact of Losses on the Persistence of the
Components of Accruals
1 9 8 7
6 5 4 3 2 1 0 1
* * *
*
+
+
+ ? ? ? ? ? +
+ + ? ? ? ? ? + + =
t t t t t
t t t t t t t
AT SG Loss AT Loss SG Loss
RNOA Loss Loss AT SG AT SG RNOA RNOA
? ? ? ?
? ? ? ? ? ? ?
(1)
1 15
14 13 12
11 10 9 8 7
6 5 4 3 2 1 0 1
* *
* * * * *
* * * *
* *
+
+
+ ? ? +
? + + ? ? ?
? ? + + + ? ? +
? + + ? ? ? ? ? + + =
t t t t t
t t t t t t
t t t t t t
t t t t t t t t t t
AT SG RNOA Loss
AT RNOA Loss SG RNOA Loss AT SG Loss
AT Loss SG Loss RNOA Loss Loss AT SG RNOA
AT RNOA SG RNOA AT SG AT SG RNOA RNOA
? ?
? ? ?
? ? ? ? ?
? ? ? ? ? ? ?
(2)
1 11 10 9 8
7 6 5 4 3 2 1 0 1
* * * *
*
+
+
+ ? ? ? ? ? ? ? +
+ + ? ? ? ? ? ? ? + + =
t t t
NCO WC
t
t t t
NCO WC
t t t
AT SG Loss AT Loss AT Loss SG Loss
RNOA Loss Loss AT SG AT AT SG RNOA RNOA
t t
t t
? ? ? ? ?
? ? ? ? ? ? ? ?
(3)
Pred. (1) (2) (3)
RNOA
t
<1 0.5362
***
0.6666
***
0.5324
***
SG
t
– -0.0393
**
0.0082 -0.0404
**
–? AT
t
– -0.0128 0.0008
–?AT
t
WC
0.0039
–?AT
t
NCO
-0.0272
–SG
t
*?AT
t
0.0251 0.0171 0.0313
RNOA
t
*SG
t
-0.3429
**
–RNOA
t
*?AT
t
-0.1891
–RNOA
t
*SG
t
*?AT
t
0.1621
**
Loss +/– -0.1350
***
-0.1137
***
-0.1392
***
Loss* RNOA
t
+/– 0.2160
***
0.0971 0.1962
***
Loss* SG
t
+/– 0.2511
***
0.1026
**
0.1817
***
–Loss*?AT
t
+/– 0.2712
***
0.1375
***
–Loss*?AT
t
WC
-0.0008
–Loss*?AT
t
NCO
0.1484
***
–Loss* SG
t
*?AT
t
0.0446 0.0520 0.0141
Loss*RNOA
t
*SG
t
0.1612
–Loss*RNOA
t
*?AT
t
-0.0525
–Loss*RNOA
t
*SG
t
*?AT
t
-0.2244
Intercept 0.0188
**
0.0060 0.0203
**
R
2
Adj. R
2
51.0%
50.0%
53.4%
52.2%
50.0%
48.7%
Sample Size 7,285 7,285 7,285
Notes to Table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for the definition of most variables.
The additional variables for this table are:
1. ?AT
t
WC
=the change in working capital turnover [(AT
t
WC
- AT
t-1
WC
) /AT
t
WC
], where AT
t
WC
is calculated as Sales
t
/ WC
t
;
2. ?AT
t
NCO
=the change in non-current operating capital turnover [(AT
t
NCO
- AT
t-1
NCO
) /AT
t
NCO
], where AT
t
NCO
is
calculated as Sales
t
/ NCO
t
;
3. WC
t
=Working Capitals, defined as Current Assets – Cash and Equivalents – Short-term Investment – Current
Liabilities + Current Debt +Tax Payable;
4. NCO
t
=Non-current operating Capitals, defined as Non-Current Assets – Long-term Investment – Non-Current
Liabilities + Non-Current Debt.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
108
are reported in column (3) of Table 6.
For profitable companies, the coefficients on
both -?AT
t
WC
and -?AT
t
NCO
are not significantly
different from zero. For loss-making
companies, (-Loss*?AT
t
NCO
) is significantly
positive, while (-Loss*?AT
t
WC
) is found to be
insignificantly different from zero. The
differential persistence level of the efficiency
component observed between profitable and
loss-making companies is primarily attributable
to the non-current operating capital component.
5.4.2 The influence of investing and
financing activities
Because the RSST (2006) measure of accruals
does not distinguish between operating,
investing and financing activities, we test the
sensitivity of our results to the inclusion of
current cash flows from investing (CFI) and
financing activities (CFF). We define CFI as
total investing cash flows scaled by lagged NOA
and CFF as total financing cash flows scaled by
lagged NOA. We include both CFI and CFF as
control variables to equation (1), (2) and (3).
For equation (3), we also allow the implications
of CFI and CFF for future profitability to vary
between profitable and loss-making companies.
Results (unreported) show that the inclusion of
CFI and CFF do not significantly alter the
inferences reported above. We do however find
that both CFI and CFF are useful in predicting
future profitability for both profitable and loss-
making companies suggesting that the RSST
(2006) decomposition of accruals could
potentially be extended to consider these aspects
of a company’s activities.
5.4.3 Alternate definition of accruals
We examine the robustness of the results to
using an alternative definition for total accruals.
We defined total accruals as the residual value
of earnings after deducting the cash flow
component (Hribar and Collins 2002). Accruals
are calculated as the difference between RNOA
and cash flow fromoperating activities. When
equation (1) is estimated for the reduced
sample, the coefficients on the persistence of
accruals are qualitatively similar to those
reported.
11
We extend the analysis by
decomposing total accruals based on the
11 The model is estimated for years 1993-2004 only, since
cash flow data obtained before 1993 is considered to be
less reliable.
alternate definition into the growth component
of accruals (change in sales growth) and a
residual accrual term defined as the difference
between total accruals and the growth
component. Unreported results show that the
growth component of accruals is still
significantly less persistent than the cash flow
component of earnings for profitable
companies. The growth component of accruals
is again found to be more persistent for loss-
making companies than profitable companies.
5.4.4 Industry effects
There is a concern that the results reported
above might be industry specific (Francis and
Smith 2005). We therefore re-estimated
equation (3) for the 22 industries identified
using ASX industry code. The mean
coefficients are examined using this approach.
The results are qualitatively similar to those
reported in tables 5 and 6. There is however
substantial variation in coefficients across
industries. Future research might consider the
systematic differences in accruals and
persistence of accruals across industries but that
is beyond the scope of this study.
5.4.5 Relation to returns
For completeness the impact on stock returns
was also explored to a limited extent. Clinch et
al (2006) find that for Australian companies,
investors misinterpret the association between
current cash flows and future earnings to a
greater extent than the association between
current accruals and future earnings. To test
whether accruals, and components of accruals
can predict future stock returns, we replicate the
model controlling for losses by replacing the
dependent variable in equation (3) with stock
returns (not reported for brevity). Returns were
defined as the next year’s annual buy-and-hold
returns measured starting three months after the
current fiscal year end. Monthly returns were
sourced from the SPPR database. The
explanatory power obtained for these return
models is relatively low (the average R-square
for our models is around one percent). The
accrual component of earnings is found to be
negatively related to next year’s return. Both the
growth and efficiency components of accruals
are also found to be significantly negative,
indicating that both components contribute to
the negative relation between accruals and
future stock return. We essentially find no
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
109
significant differences between the profitable
and loss-making companies.
6. Conclusion
This research examines persistence of accruals
for Australian companies. Prior research using
U.S. samples suggests that the accrual
component of earnings is less persistent than the
cash flow component of earnings (Sloan 1996;
Xie 2001; RSST 2005), indicating that the low
earnings persistence is largely attributable to
accruals. Prior research also suggests that two
primary factors contribute to the lower
persistence of accruals: the diminishing
marginal returns on new investments and
accounting distortions (RSST 2006).
This study finds evidence that the accrual
component of earnings is less persistent than the
cash flow component of earnings for profitable
Australian companies. The results indicate that
the growth and efficiency components of
accruals are less persistent than the cash flow
component of earnings for profitable
companies.
The results suggest that the persistence of the
components of accruals are sensitive to
companies’ current profitability. For our
sample, we find that loss-making companies
have higher persistence of the growth and
efficiency components of accruals than the cash
flow component of earnings, suggesting that for
currently unprofitable companies, the accrual
components of earnings are more informative
for period ahead profitability than the cash flow
component of earnings.
The results also suggest that the RSST
(2006) approach to measuring persistence of
accruals requires further consideration. Results
for models allowing interaction between
accruals and earnings suggest that the
interaction is explaining future profitability.
The results must be interpreted with respect
to several limitations. Firstly, consistent with
previous studies in this line of research only a
single future period is examined. The
implications of current earnings, and earnings
components for future periods beyond one year
could be different using a more complete time
series where data permits. Secondly, results
suggest that the higher accruals persistence for
loss-making companies is mostly consistent
with the argument that small companies in the
development stage experience consistent losses
for several years. This does not however mean
that one can ignore the impact of conservative
accounting practices. To test the impacts of
these two factors, new models would need to be
developed. Thirdly, this line of research ignores
any reliability issues associated with the cash
flow component of earnings. Some recent
studies argue that the cash flows are also subject
to managerial manipulation (Roychowdhury
2006; Graham et al 2005) and future research
could also consider the transient components of
operating cash flows.
References:
Balkrishna, H., Coulton, J.J. and Taylor, S.L. (2007),
“Accounting Losses and Earnings Conservatism:
Evidence from Australian GAAP”, Accounting and
Finance, Forthcoming.
Basu, S. (1997), “The Conservatism Principle and the
Asymmetric Timeliness of Earnings”, Journal of
Accounting and Economics, vol. 24, pp. 3–37.
Chan, K., Chan, L., Jegadeesh, N. and Lakonishok, J.
(2006), “Earnings Quality and Stock Returns”, Journal
of Business, vol. 79, pp. 1041–1082.
Clinch, G., Fuller, D., Govendir, B. and Wells, P. (2007),
“The Accrual Anomaly: Australian Evidence”, Working
Paper, University of Technology, Sydney.
Coulton, J., Taylor, S. and Taylor, S.L. (2005), “Is
‘Benchmark Beating’ by Australian Firms Evidence of
Earnings Management”, Accounting and Finance,
vol. 45, pp. 553–576.
Davidson, R., Stewart, J.G. and Kent, P. (2005), “Internal
Governance Structures and Earnings Management”,
Accounting and Finance, vol. 45, pp. 241–267.
Dechow, P.M. and Dichev, I.D. (2002), “The Quality of
Accruals and Earnings: The Role of Accrual Estimation
Errors”, The Accounting Review, vol. 77 (Supplement),
pp. 35–59.
Eddey, P.H. and Taylor, S.L. (1999), “Directors’
Recommendations on Takeover Bids and the
Management of Earnings: Evidence from Australian
Takeovers”, ABACUS, vol. 35, pp. 29–45.
Fairfield, P., Whisenant, S. and Yohn, T.L. (2003), “The
Differential Persistence of Accruals and Cash Flows for
Future Operating Income versus Future Profitability”,
Review of Accounting Studies, vol. 8, pp. 221–243.
Fama, E.F. and MacBeth, J.D. (1973), “Risk, Return and
Equilibrium: Empirical Tests”, The Journal of Political
Economy, vol. 81, pp. 607–636.
Francis, J. and Smith, M. (2005), “A Re-examination of the
Persistence of Accruals and Cash Flow”, Journal of
Accounting Research, vol. 43, pp. 413–451.
Givoly, D. and Hayn, C. (2000), “The Changing Time-
Series Properties of Earnings, Cash Flows and Accruals:
Has Financial Reporting Become More Conservative”,
Journal of Accounting and Economics, vol. 29,
pp. 287–320.
Godfrey, J.M., Mather, P. and Ramsay, A. (2003), “Earnings
and Impression Management in Financial Reports: The
Case of CEO Changes”, Abacus, vol. 39, pp. 95–123.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
110
Graham, J.R., Harvey, C.R. and Rajgopal, S. (2005), “The
Economic Implications of Corporate Financial
Reporting”, Journal of Accounting and Economics,
vol. 40, pp. 3–73.
Hayn, C. (1995), “The Information Content of Losses”,
Journal of Accounting and Economics, vol. 20,
pp. 125–154.
Hribar, P. and Collins, D.W. (2002), “Errors in Estimating
Accruals: Implications for Empirical Research”,
Journal of Accounting Research, vol. 40, pp. 105–134.
Jones, J. J. (1991), “Earnings Management during Import
Relief Investigations”, Journal of Accounting Research,
vol. 29(2), pp. 1–31.
Joos, P. and Plesko, G.A. (2005), “Valuing Loss Firms”, The
Accounting Review, vol. 80, pp. 847–870.
Kang, T. (2005), “Earnings prediction and the role of
accrual-related disclosure: international evidence”,
Accounting Research Journal, vol. 18, pp. 6–12.
Kean, S. and Wells, P. (2006), “Earnings Decomposition and
the Persistence of Earnings”, Working Paper, University
of Technology, Sydney.
Kelley, S.O., Koh, P.S., and Tong, Y.H. (2004),
“Governance Structures and Abnormal Accruals
Characteristics”, Working Paper, University of
Queensland.
Klein, A. and Marquardt, C.A. (2006), "Fundamentals of
Accounting Losses", The Accounting Review, vol. 81,
pp. 179–206.
Koh, P.S. (2003), “On the Association between Institutional
Ownership and Aggressive Corporate Earnings
Management in Australia”, The British Accounting
Review, vol. 35, pp. 105–128.
Lonergan, W. and Wells, P. (2006), “Changes in the
Association Between Stock Prices and Book Value of
Equity and Earnings over the Period 1979–1997”,
Working Paper, University of Technology, Sydney.
Oei, R., Ramsay, A., and Mather, P. (2006), “Earnings
persistence, accruals and managerial share ownership”,
Working Paper, Monash University.
Richardson, S.A., Sloan, R.G., Soliman, M.T. and Tuna, I.
(2005), “Accrual Reliability, Earnings Persistence and
Stock Prices”, Journal of Accounting and Economics,
vol. 39, pp. 437–485.
Richardson, S.A., Sloan, R.G., Soliman, M.T. and Tuna, I.
(2006), “The Implication of Accounting Distortions and
Growth for Accruals and Profitability”, The Accounting
Review, vol. 81, pp. 713–743.
Roychowdhury, S. (2006), “Earnings Management through
Real Activities Manipulation”, Journal of Accounting
and Economics, vol. 42, pp. 335–370.
Sloan, R.G. (1996), “Do Stock Prices Fully Reflect
Information in Accruals and Cash Flows about Future
Earnings”, The Accounting Review, vol. 71,
pp. 289–315.
Taylor, S. and Taylor, S.L. (2003), “Earnings Conservatism
in a Continuous Disclosure Environment: Empirical
Evidence”, Working Paper, University of Melbourne
and University of New South Wales.
Watson, J. and Wells, P. (2005), “The Association Between
Various Earnings and Cash Flow Measures of Firm
Performance and Stock Returns: Some Australian
Evidence”, Working Paper, University of Technology,
Sydney.
Wells, P. (2002), “Earnings Management Surrounding CEO
Changes”, Accounting and Finance, vol. 42,
pp. 169–193.
Xie, H. (2001), “The Mispricing of Abnormal Accruals”,
The Accounting Review, vol. 76, pp. 357–373.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Greg Clinch, Damian Fuller, Brett Govendir, Peter Wells. 2012. The accrual anomaly: Australian evidence. Accounting &
Finance 52:10.1111/acfi.2012.52.issue-2, 377-394. [CrossRef]
2. Kristen Anderson, Kerrie Woodhouse, Alan Ramsay, Robert Faff. 2009. Testing for asymmetric effects in the accrual anomaly
using piecewise linear regressions. Pacific Accounting Review 21:1, 5-25. [Abstract] [Full Text] [PDF]
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_506171159.pdf
Recent research examines the implications of
components of accruals for future profitability.
Because the persistence of earnings varies with
the level of company profitability
Accounting Research Journal
Components of Accruals, Losses and Future Profitability
Hai Wu Neil Fargher
Article information:
To cite this document:
Hai Wu Neil Fargher, (2007),"Components of Accruals, Losses and Future Profitability", Accounting Research J ournal, Vol.
20 Iss 2 pp. 96 - 110
Permanent link to this document:
http://dx.doi.org/10.1108/10309610780000695
Downloaded on: 24 January 2016, At: 21:02 (PT)
References: this document contains references to 0 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 469 times since 2007*
Users who downloaded this article also downloaded:
Arifur Khan, Paul Mather, (2013),"The value of executive director share ownership and discretionary accruals", Accounting
Research J ournal, Vol. 26 Iss 1 pp. 35-55 http://dx.doi.org/10.1108/ARJ -02-2012-0011
Prodosh Simlai, (2014),"Firm characteristics, distress risk and average stock returns", Accounting Research J ournal, Vol. 27
Iss 2 pp. 101-123 http://dx.doi.org/10.1108/ARJ -06-2012-0046
Kim Watty, Satoshi Sugahara, Nadana Abayadeera, Luckmika Perera, J ade McKay, (2014),"Towards a Global Model of
Accounting Education", Accounting Research J ournal, Vol. 27 Iss 3 pp. 286-300 http://dx.doi.org/10.1108/ARJ -08-2013-0054
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
96
Components of Accruals, Losses and Future
Profitability
Hai Wu and Neil Fargher
Department of Accounting and Finance
Division of Economic and Financial Studies
Macquarie University
Abstract
Recent research examines the implications of
components of accruals for future profitability.
Because the persistence of earnings varies with
the level of company profitability, we expect
differences between profitable and loss-making
companies in the association between
components of accruals and future profitability.
Using the approach adopted by Richardson,
Sloan, Soliman and Tuna (2006) we find
evidence suggesting that the components of
accruals related to revenue growth and to change
in asset turnover are less persistent than the cash
flow component of earnings for profitable
Australian companies. For loss-making
companies, however, the persistence of the
accrual component of earnings is found to be
higher than for the cash flow component of
earnings, suggesting that the accrual component
is more informative than the cash flow
component in explaining period ahead
profitability for many currently unprofitable
companies.
1. Introduction
A basic premise of accrual accounting is that it
provides a more timely and relevant performance
measure than cash flows through a better
measure of revenues and expenses. Accounting
Acknowledgements: We wish to acknowledge the helpful
comments fromGraeme Harrison, Geoff Loudon, Farshid
Navissi, Alan Ramsay, Edward Watts, Peter Wells, Sue
Wright and participants at the Asian Academic Accounting
Association Conference 2006 and the 2007 UTS Summer
Accounting Research Conference. All errors remain the
responsibility of the authors. We gratefully acknowledge the
use of data supplied by the Securities Industry Research
Centre of Asia-Pacific (SIRCA) on behalf of Aspect
Financial and by the Centre for Research in Finance.
Key Words: Accruals, Losses, Persistence
accruals can reduce uncertainty about future
earnings and allow investors to more accurately
predict future earnings (Kang 2005).
Sloan (1996) shows that the accrual
component of earnings is less persistent1 than
the cash flow component of earnings in
explaining period ahead profitability.
Richardson, Sloan, Soliman and Tuna (2006,
hereafter RSST) suggest that the two primary
factors explaining the lower persistence of the
accrual component of earnings are the
diminishing marginal returns embedded in new
investments and accounting distortions. They
decompose the accrual component of earnings
into a growth component and an efficiency
component. The growth component measures
the effect of investment growth on the
persistence of accruals. To the extent that the
efficiency component drives the observed
properties of accruals, it is inferred that the
efficiency component measures the effect of
accounting distortions (RSST 2006, page 719).
Their results suggest that both components
contribute to the lower persistence of the
accrual component of earnings relative to the
cash flow component.
Prior research suggests that the persistence of
earnings varies with profitability (Hayn 1995,
Joos and Plesko 2005, Balkrishna et al. 2007). If
earnings persistence varies between profitable
and loss-making companies, then we would also
expect differences in the persistence of the
components of accruals. We extend the
previous research to examine the differential
implications of current components of earnings
1 In this literature, “persistence” is used to refer to the
implication of accruals for future profitability. That is,
persistence refers to the association between this period
accruals and next period profitability.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
97
for future profitability for companies with
current period losses.
For profitable Australian companies, we find
evidence consistent with the accrual component
being less persistent than the cash flow
component of earnings. When the accruals
component is decomposed, we find evidence
suggesting that the growth and efficiency
components of accruals exhibit lower
persistence than the cash flow component of
earnings. For loss-making companies, the
persistence of the accrual component of
earnings is found to be higher than the cash
flow component of earnings. This result extends
to the components of accruals. The results are
consistent with the argument that loss-making
companies in the development stage tend to
have accruals that are more informative to next
period profitability than is the cash flow
component of earnings. We also find that the
higher persistence of the efficiency components
of accruals is mainly attributable to non-current
operating accruals.
This paper is organised as follows. The next
section provides a brief literature review.
Section three defines the variables and outlines
the research design. Section four describes the
sample. Section five discusses the results.
Finally the overall conclusions and implications
for future research are presented.
2. Relation to Previous Research
2.1 Persistence of Accruals
Following Sloan (1996), explanations for the
lower persistence of accruals can be grouped
into two categories: the diminishing marginal
returns embedded in new investment and
accounting distortions. Fairfield et al (2003)
show that diminishing marginal returns affect
accrual earnings to a greater extent than cash
earnings. Diminishing marginal returns imply
that for high profit companies, returns on the
real investment acquired in the current period
will be lower in the next period due to
competition. This will drag down the average
return for the company’s real investments
towards the long-term mean return. On the other
hand, investments generating lower returns in
the current period will be replaced by
investments with better returns in the current or
future period. This will drive up the average
return towards the long-term mean.
Accounting distortions are caused by the use
of inappropriate accounting. The selection of
inappropriate accounting policies can be either
intentional (earnings management) or
unintentional (estimation error). Xie (2001)
shows that the lower persistence of the accrual
component of earnings is attributable to the
discretionary accruals, which measure the level
of earnings management activities (Jones 1991).
Chan et al. (2006) find that the negative relation
between accruals and future stock returns is
mainly caused by the discretionary component
of accruals.
On the other hand, unintentional accrual
estimation error can also cause the lower
persistence of the accrual component of
earnings (RSST 2005; Dechow and Dichev
2002). RSST (2005) argue that the lower
persistence of accruals is caused by the
estimation error embedded in the accrual
accounts. They argue that due to the nature of
accruals, the persistence of accrual earnings
should be theoretically less than cash earnings.
They decompose total accruals into a range of
aggregate accounting items based on their
relative reliability level. Their results suggest
that the accrual components that are less reliable
(higher estimation error) have lower persistence.
RSST (2006) extend the literature by
decomposing total accruals into the growth and
efficiency components in order to discriminate
effects of two alternative explanations on the
persistence of accruals. The growth component
captures the effect of growth in real investment,
while the efficiency component accounts for
the estimation error in accruals caused either
by the intentional or unintentional adoption
of inaccurate accounting valuation policies.
They find that both components contribute
to the lower persistence of accruals. This
decomposition approach forms the basis for our
analysis examining the persistence of accruals
for Australian companies and the influence of
reported losses on the association between
accruals and persistence.
2.2 Loss-making Companies
Evidence from prior research suggests
differences in persistence of accruals between
companies currently making losses and
profitable companies. Hayn (1995) documents
an insignificant earnings-return relation,
suggesting that the market expects negative
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
98
earnings to be transitory for loss-making
companies. Basu (1997) shows that bad news is
recognised in a more timely manner than good
news, due to accounting conservatism. This
suggests that bad news is reported with
relatively larger errors than good news, as more
timely news is usually more relevant but less
reliable (Taylor and Taylor 2003). It follows
that conservatism should result in earnings
being more transitory and less persistent (Basu
1997). Givoly and Hayn (2000) show that the
increase in the frequency of negative earnings is
mainly caused by negative accruals, and that the
cash flow component of earnings is less
influenced by conservative accounting practice
than the accrual components (Givoly and Hayn
2000). If losses reflect excessive conservatism,
then we expect earnings to be less persistence
for loss-making companies, and this should be
partially attributable to the accruals components
arising from conservative accounting.
Many loss-making companies are, however,
small companies in the early stage of their
development with very limited resources (Klein
and Marquardt 2006). These companies are
more likely to make consistent losses in their
early stage of development, and profits are only
expected after several years. The impact is that
these loss-making companies are more likely to
have a higher persistence in earnings and
accruals compared to profitable companies.
While both accounting conservatism and
stage in the investment cycle predict differences
in the persistence level of accruals between
loss-making and profitable companies, the
direction predicted is ambiguous. The results
are determined by the extent to which losses are
caused by these two factors. By explicitly
considering the impact of losses on the
persistence of earnings, and its components, our
study provides evidence on the relative
importance of these explanations.
2.3 Australian Evidence
Australian studies have largely concentrated on
the area of earnings management. Typically
earnings management studies examine
discretionary accruals in relation to a range of
factors that are expected to affect the intensity
of management manipulation. (Eddey and
Taylor 1999, Godfrey et al 2003, Wells 2002,
Koh 2003, Kelley et al 2004, Davidson et al
2005, and Coulton et al 2005).
There are fewer studies examining the
persistence of changes in earnings or accruals.
Taylor and Taylor (2003) examine the presence
of conservatism in Australian accounting
practices.
2
Following Basu (1997), they test
the persistence of changes in earnings, where
lower persistence of negative earnings change
should indicate the presence of accounting
conservatism. The results from Taylor and
Taylor (2003) suggest the possibility that loss-
making companies exhibit differences in
earnings persistence from profitable companies.
Coulton et al (2005) provide evidence for the
persistence of discretionary and non-
discretionary accruals to show the effectiveness
of discretionary accruals in capturing earnings
management. Their results are mixed. The
accrual component of earnings is found to be
less persistent than the cash flow component of
earnings. However, unlike Xie (2001), the
results for the difference in the persistence level
of discretionary and non-discretionary accruals
is only found for one of the three measures of
discretionary accruals used. The authors argue
that the result could be partially explained by
the relatively large portion of loss-making
companies in the sample.
Oei et al (2006) examine the persistence
level of accruals and its components for the top
300 Australian companies by market
capitalisation. Their results suggest that accruals
and the components of accruals are less
persistence than the cash flows component of
earnings and that there are differences between
the persistence levels of the components of
accruals. Kean and Wells (2006) decompose
return on equity on three bases, cash/accruals,
operating/financing/other, and financial ratios.
They show that under all three decomposition
bases, the components of earnings exhibit
different levels of persistence. Specifically, the
accrual component of ROE is shown to be less
persistent than the cash flow component.
Recent Australian research also provides
some evidence regarding loss-making
companies. Balkrishna et al (2007) and
Lonergan and Wells (2006) both document an
increasing portion of companies reporting losses
over the previous 10 to 30-year period. Watson
2 In Taylor and Taylor (2003) accounting conservatismis
defined as the timely reflection of bad news compared to
good news.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
99
and Wells (2005) find that earnings are not
useful in explaining stock returns for loss-
making companies supporting the transitory
nature of losses. Balkrishna et al (2007) find that
accounting practices are more conservative
among loss-making companies than profitable
companies. This provides evidence supporting
the view that loss-making companies’ earnings
potentially include greater accounting distortions
and are less persistence than profitable
companies. We further examine the issue of the
relation between the components of earnings and
period ahead profitability for loss-making
companies.
3. Variable Definitions and Model
Specification
3.1 Defining accruals
We initially adopt the operating accruals
definition used by RSST (2006). They measure
total operating accruals as the change in net
operating assets deflated by lagged operating
assets (RSST, 2006, page 722).
1
1
?
?
?
=
t
t t
t
NOA
NOA NOA
TACC
Net operating asset (NOA) is defined as the
difference between non-cash operating assets
and operating liabilities. Operating assets is total
assets net of cash and equivalents and total
investments, while operating liability is total
liabilities net off short-termand long-termdebts,
as well as income tax liability.
3
This accruals
measure is in contrast to the more widely used
definition of accruals as the changes in non-cash
working capital net of depreciation expenses.
RSST (2006) argue that the traditional accruals
measure omits many non-current operating
accruals, which are sources of growth and are
affected by accounting distortions. We initially
follow the RSST (2006) approach but then
consider the sensitivity of the results to this
definition of accruals.
3 NOA is therefore defined as: Total Assets (Aspect Item
#5090) – Cash and Short Terminvestment (Aspect Item
#4990) – Investment and Advance (Aspect Item#5070) -
[Total Liabilities (Aspect Item#6040) – Debt in Current
Liabilities (Aspect Item#6000) – Long-termdebt (Aspect
Item#6020) – Income tax liability (Aspect Item#319)].
3.2 The growth and efficiency components
of accruals
Following RSST (2006), the growth component
of accruals is defined as change in sales. The
efficiency component is measured as change in
asset turnover. They are defined as follows:
1 ?
?
=
t
t
Sales
Sales
Growth
;
t
t
AT
AT
Efficiency
?
= , where
t
t
NOA
Sales
AT =
This type of analysis assumes that the
increase in real investment in the current period
causes production growth and subsequently
sales growth. Asset turnover measures the
efficient use of net operating assets to generate
revenue. Change in asset turnover is caused by
disproportionate changes in Sales and NOA.
This can be caused by accounting distortions,
which causes NOA to be valued with estimation
error, or change in real efficiency. RSST (2006)
show that total operating accruals can be
parsimoniously decomposed as follow:
t t t t
t
t
t
t
t
t
t
t
t
t
AT SG AT SG
AT
AT
Sales
Sales
AT
AT
Sales
Sales
NOA
NOA
TACC
? ? ? ? ? ? ? =
?
?
?
?
?
?
?
=
?
=
? ?
?
) ( ) (
1 1
1
In this decomposition, sales growth is
positively related to total operating accruals,
while the change in the efficient use of assets is
negatively related to total operating accruals. In
the absence of sales growth, reduction in asset
turnover is solely caused by increase in net
operating assets, which measures accruals.
There is a third term in the decomposition,
which is an interaction term. It is argued that
economies of scale imply a positive correlation
between sales growth and change in efficiency.
4
3.3 Model specification
Following RSST (2006, page 733) we model
the period ahead accounting rate of return as:
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ?
(1)
4 Economies of scaleimply a diminishing marginal cost to
the production in the long term. As sales grow, production
levels also grow. The cost of producing additional
products will drop, requiring less capital to be used. This
will effectively increase asset turnover. Therefore, with an
increase in sales growth, asset turnover also increase.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
100
where RNOA
t
is return on NOA for period t,
defined as Net Operating Income (NOI) in
period t deflated by lagged NOA. NOI is
sourced from the Aspect Financial database as
EBIT (Aspect Item #8012) minus Other
Revenue (Aspect Item #7080). This specification
is explained by RSST (2005, page 457) as a
direct test of the ‘relative persistence’ of
accruals. If the accrual component of earnings is
less persistent than the cash flow component of
earnings, ?
2
is predicted to be less than zero.
5
An alternative formof persistence model is
also considered. An interaction term
(RNOA
t
*TACC
t
) is included to capture the
relation between the interaction of this period
accruals and profitability, and one-year ahead
profitability. That is, the RSST (2006) model
does not allow for the interaction between
profitability and accruals. The basic model
including this interaction term is expressed as
follows:
2 1 0 1 +
+ + =
t t t
TACC RNOA RNOA ? ? ?
1 3
*
+
+ +
t t t
TACC RNOA ? ?
(1a)
RSST (2006) restates equation (1) to separate
the growth and efficiency components of total
operating accruals in order to examine their
individual effects on the differential persistence
of cash and accruals:
3 2 1 0 1 +
? ? + + =
t t t t
AT SG RNOA RNOA ? ? ? ?
1 4 +
+ ? ? ?
t t t
AT SG ? ?
(2)
Based upon U.S. evidence the coefficients on
sales growth and change in asset turnover are
predicted to be less than zero (?
2
<0 and ?
3
<0).
In order to examine the difference in
persistence of accruals between loss-making
and profitable companies, we define an
indicator variable for loss firms. Loss takes the
value of one if current year’s earnings are
negative (RNOA
t
<0) and zero otherwise.
Losses can affect both the intercept and the
slope coefficients. Extending the model to
consider the growth and efficiency components
of accruals, and the differential persistence of
unprofitable companies, results in the following
model:
3 2 1 0 1 +
? ? + + =
t t t t
AT SG RNOA RNOA ? ? ? ?
5 4
+ + ? ? ?
t t
Loss AT SG ? ?
6
*
t
RNOA Loss ?
8 7
* * ? ? +
t t
AT Loss SG Loss ? ?
1 9
*
+
+ ? ? ?
t t t
AT SG Loss ? ?
(3)
To examine the incremental difference in the
persistence of unprofitable companies we test
whether ?
7
and ?
8
are significantly different from
zero.
4. Sample Selection and Description
4.1 Sample Selection
Table 1 summarises the sample selection
process. We identified 22,014 company-year
observations from the ASPECT Financial
Database for the period 1987 to 2005. We
perform four adjustments. (1) We eliminated all
firm-year observations with missing EBIT.
6
(2)
We exclude all firm-year observations with zero
sales revenue.
7
(3) Because all financial
statement variables are measured in terms of
percentage change, and the dependent variable
is one-year ahead accounting rate of return,
observations without matched one-year-lagged
and one-year-ahead data are excluded. (4) Upon
review of the data by year we further excluded
all observations fromthe four years prior to
1990 because of the relatively narrow coverage
in these early years and a clear bias towards
larger companies. We obtain 7,432 company-
year observations representing 1,249
companies. After matching our sample to the
return data obtained from the CRIFS-SPPR
database, our final sample contains 7,285
company-year observations covering 1,240
companies for the period 1990 to 2004. Yearly
sample variation ranges from 305 observations
in 1990 to 637 observations in 2002. Consistent
with previous research of this type it must be
acknowledged that the data requirements tend to
exclude smaller, less profitable companies.
Further, to the extent that companies that do not
survive a consecutive three-year period are
eliminated, the sample reflects more profitable
companies and the results must be interpreted
with respect to this limitation.
5
If
1 2 1 0 1
) (
+ +
+ + ? + =
t t t t t
TACC TACC RNOA RNOA ? ? ? ? ,
where ?
1
is the persistence of cash flows and ?
2
is the
persistence of accruals then this can be rewritten as
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ? where ?
2
>0 is a
test of ?
1
– ?
2
.
6 This step also excludes financial services companies as
EBIT is not meaningful for these entities.
7 The calculations of the growth and efficiency
components of accruals require current and previous
sales revenue to be strictly non-zero.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
101
Table 1
Sample Selection
Companies Company-Years
Total observations from the Aspect Financials Database 2,836 22,014
Excluding observations with missing EBIT 2,342 18,206
Excluding observations with missing Trading Revenue 1,996 14,405
Excluding all firm-year with a zero value for Trading Revenue 1,787 13,156
Excluding all firm-year data without matching lagged data 1,500 9,511
Excluding all firm-year data without matching one-year ahead data 1,249 7,453
Excluding all firm-year data before 1990 (due to small coverage
before 1990) 1,249 7,432
Final sample after excluding all firm-years without the matching
return data from the CRIFS SPPR database. 1,240 7,285
Table 2
Descriptive Statistics
Panel A: Descriptive Statistics
Variables Mean Median Std. Dev % Positive Min Max
Number of
Observations
TACC
t
0.175 0.045 0.559 58.30% -0.586 1.872 7,285
SG
t
0.301 0.092 0.751 67.04% -0.511 2.804 7,285
?AT
t
-0.084 0.038 0.660 55.62% -2.150 0.801 7,285
RNOA
t
-0.138 0.032 0.551 55.68% -1.897 0.509 7,285
Panel B: Pearson Correlation Matrix
TACC
t
SG
t
?AT
t
RNOA
t
RNOA
t+1
TACC
t
–
0.264
(0.0001)
-0.566
(0.0001)
-0.030
(0.0118)
0.085
(0.0001)
SG
t
0.264
(0.0001)
–
0.387
(0.0001)
-0.063
(0.0001)
-0.073
(0.0001)
?AT
t
-0.566
(0.0001)
0.387
(0.0001)
–
0.124
(0.0001)
0.011
(0.3600)
RNOA
t
-0.030
(0.0118)
-0.063
(0.0001)
0.124
(0.0001)
–
0.708
(0.0001)
RNOA
t+1
0.085
(0.0001)
-0.073
(0.0001)
0.011
(0.3600)
0.708
(0.0001)
–
Notes to Table:
Variable Definitions.
TACC
t
=total accruals in period t, defined as the change in Net Operating Asset (NOA) deflated by the lagged NOA
[(NOA
t
– NOA
t-1
)/ NOA
t-1
];
SG
t
=sales growth at period t [(Sales
t
– Sales
t-1
)/Sales
t-1
];
?AT
t
=the change in Asset Turnover [(AT
t
- AT
t-1
)/AT
t
], where AT
t
is equal to Sales
t
/ NOA
t
;
RNOA
t
=return on Net Operating Assets for period t, defined as Net Operating Income (NOI) in period t deflated by
lagged Net Operating Assets (NOA), where NOI is EBIT – Other Revenue;
NOA =Operating Assets – Operating Liabilities, refer text for details;
Loss =1 if RNOA
t
<0, otherwise 0.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
102
4.2 Descriptive Statistics
Table 2 presents the descriptive statistics for the
variables examined in this research. All variables
are winsorized at the 95% and 5% level to
reduce the impact of extreme observations on
the distribution of regression residuals. Panel A
of Table 2 presents the univariate statistics. The
mean values for accruals (TACC) and sales
growth (SG) are positive, while the mean value
for change in efficiency (?AT) is negative. The
mean accruals measured this way is 17.5%. This
compares to the mean for RSST (2006) of
16.4%. Around 55% of the companies in the
sample experience positive return on net
operating assets (RNOA) in either current period
or next period or both. The mean values for
RNOA
t
and RNOA
t+1
are however negative. This
is not the case for U.S. samples (e.g. Sloan 1996;
Xie 2001) and potentially influences the
persistence of accruals.
Panel B in Table 2 presents the pairwise
correlations between variables. Most of the
correlations are significant at the one percent
level, except for the correlation between the
efficiency component of accruals (?AT) and
future earnings (RNOA
t+1
). As would be
expected, there is a strong positive correlation
between current and future earnings. The sales
growth (SG) and change in efficiency (?AT) are
positively correlated. RSST (2006) argue that
this positive correlation supports the argument
for economies of scale.
Table 3
The Distributions of Firm-Year Observations with
Respect to Earnings Patterns
Company size
Size classification 1 Size classification 2
Earnings pattern
(2)
Small
(3)
Large
(4)
Total
(5)
Small
(6)
Medium
(7)
Large
(8)
Total
Consecutive losses
(Row Percentage)/
[Column Percentage]
1,802
(72%)
[49%]
717
(28%)
[20%]
2,519
[35%]
1,108
[62%]
1,138
[31%]
273
[15%]
2,519
[35%]
Reversing losses 381
(54%)
[10%]
330
(46%)
[9%]
711
[10%]
171
[10%]
382
[10%]
158
[9%]
711
[10%]
Consecutive
profits
1,092
(33%)
[30%]
2,218
(67%)
[61%]
3,310
[45%]
349
[20%]
1,754
[48%]
1,207
[66%]
3,310
[45%]
Reversing profits 368
(49%)
[10%]
377
(51%)
[10%]
745
[10%]
152
[9%]
409
[11%]
184
[10%]
745
[10%]
Total 3,643
(50%)
3,642
(50%)
7,285
1,780
(24%)
3683
(51%)
1,822
(25%)
7,285
Notes to Table:
The sample is separated into four categories based upon the following pattern of earnings: 1. The consecutive losses
category contains those firm-year observations that make losses in both the current and next period; 2. The consecutive
profits category is made up of firm-year observations making profit in both the current and next period; 3. Firm-year
observations are classified as reversing losses if the current period earnings are negative but subsequently reverse to
positive in the next period; 4. Firm-year observations are classified as reversing profits if negative earnings are reported
in the current period, but subsequently earning negative earnings in the next period.
Two alternate criteria are used to classify companies by company size. The first criterion classifies companies as large
or small. The top 50% firm-year observations by average net operating assets (NOA) in the current period are classified
as large companies. The bottom50% firm-year observations in the sample are classified as small companies. The second
criterion classified companies into three categories. The top 25% companies are classified as large. The bottom25%
companies are classified as small. The middle 50% companies are therefore classified as mediumsize. Both metrics are
based on the average net operating assets (NOA) calculated as the average between the beginning and ending NOA.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
103
4.3 Loss-making companies
Small companies in their development stage are
more likely to make consistent losses (Klein and
Marquardt 2006), leading to higher persistence
levels of earnings and accruals.
8
Table 3
provides descriptive evidence on the relation
between company size and the pattern of
earnings persistence. Four earnings patterns are
summarised: consecutive losses (RNOA
t
<0 and
RNOA
t+1
<0), loss reversion (RNOA
t
<0 and
RNOA
t+1
>0), consecutive profits (RNOA
t
>0
and RNOA
t+1
>0), and profit reversion
(RNOA
t
>0 and RNOA
t+1
<0). By definition,
companies experiencing profit and loss
reversion have lower earnings persistence than
companies making consecutive profits or losses.
Table 3 shows that small companies are as
likely to experience profit and loss reversion as
large companies. Small companies are however
more likely to report consecutive losses than
large companies, as 72% of the consecutive loss
companies are small companies. In addition,
most companies making consecutive profits are
large companies (67%). A decreasing pattern
can also be identified for companies making
consecutive losses with increasing company
size. Specifically, 62% of small companies
report consecutive losses compared to 31% of
medium size companies and 15% of large size
companies. These sample characteristics
support the argument that small companies are
more likely to report losses and more likely to
report consistent losses. We focus our analysis
on differences in earnings persistence between
profitable and loss making companies.
However, clearly, profitability is also related to
firm size.
5. Results
5.1 Persistence of accruals
Table 4 provides initial estimates of persistence
for comparison with previous research. Results
are reported using the Fama-Macbeth (1973)
procedure. We estimate yearly cross-sectional
regressions and report the mean coefficient
across the annual regressions. Column (1) of
8 Unreported statistics show that only 25.37% of large
companies are making losses compared to 50.85% of
small companies, where small is defined as the average
balance of beginning and ending net operating assets less
than the median value.
Panel A provides results for the basic regression
of the accounting rate of return on lagged rate of
return. The adjusted R-Square for the model is
0.47. The persistence coefficient for earnings
(?
1
) is estimated at 0.76 and is significantly less
than one (F=9.16, p-value <0.0001). A value of
the coefficient ?
1
that is less than one supports
the maintained hypothesis in this type of study
that current earnings are mean reverting.
Based upon previous research we would
expect a lower persistence level for the accrual
component of earnings compared to the cash
flow component. The results including an
accrual component are reported in column (2) of
Panel A of Table 4. The coefficient ?
1
is
estimated at 0.76 for the persistence of the cash
flow component of earnings. The coefficient on
accruals (TACC
t
) is significantly greater than
zero (?
2
=0.12, p-value <0.0001) indicating that
the persistence level of the accrual component is
higher than the cash flow component of
earnings. This is contrary to expectations. The
high incidence of loss-making companies in our
sample is a potential explanation. To investigate
the effects of loss-making companies and
observations with extreme change in the
accounting rate of return, we re-estimate the
regression using a reduced sample. We exclude
loss-making companies, which include all
observations with negative RNOA
t
or negative
RNOA
t+1
. The reduced sample size is 3,310. By
excluding all companies with losses, we
systematically reduce the number of
observations with extreme changes in the
accounting rate of return, of which many
observations have low earnings persistence.
Results for the reduced sample set excluding
companies making losses are reported in column
(3) of Table 4 Panel A. The coefficient on TACC
is now significantly negative at less than the one
percent level (?
2
=-0.15), indicating that the
accrual component of earnings is less persistent
than the cash flow component. This is consistent
with prior research using U.S. data, with RSST
(2006) finding a value of -0.131 for this
coefficient. Overall, the results for the reduced
sample restricted to profitable companies, are
generally consistent with those found using U.S.
data.
The specification used by RSST (2006) does
however ignore the interaction between
profitability and accruals. Column (4) reports
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
104
Table 4
Basic measures of persistence
Panel A:
1 1 0 1 + +
+ + =
t t t
RNOA RNOA ? ? ?
(1)
1 2 1 0 1 + +
+ + + =
t t t t
TACC RNOA RNOA ? ? ? ?
(2), (3)
1 3 2 1 0 1
*
+ +
+ + + + =
t t t t t t
TACC RNOA TACC RNOA RNOA ? ? ? ? ?
(4)
Full sample Consistent profits sub sample†
Pred.
(1) (2) (3) (4)
RNOA
t
<1 0.7575
**
0.7616
**
0.6930
***
0.7548
**
TACC
t
– 0.1190
***
-0.1480
***
-0.1024
***
RNOA
t
* TACC
t
-0.1139
***
Intercept -0.0596
***
-0.0787
***
0.0620
***
0.0545
***
R
2
Adj. R
2
46.6%
46.5%
48.2%
48.0%
54.0%
53.9%
55.3%
55.2%
Sample Size 7,285 7,285 3,310 3,310
Panel B:
1 4 3 2 1 0 1 + +
+ ? ? ? ? ? + + =
t t t t t t t
AT SG AT SG RNOA RNOA ? ? ? ? ? ?
(1), (2)
1 7 6 5
4 3 2 1 0 1
* * *
+
+
+ ? ? ? ? ?
+ ? ? ? ? ? + + =
t t t t t t t
t t t t t t
AT SG RNOA AT RNOA SG RNOA
AT SG AT SG RNOA RNOA
? ? ? ?
? ? ? ? ?
(3)
Full sample Consistent profits sub sample †
Pred.
(1) (2) (3)
RNOA
t
<1 0.7517
***
0.7290
***
0.7445
***
SG
t
– 0.0521
***
-0.0531
***
-0.0567
***
–?AT
t
– 0.0939*** -0.0625
***
-0.0663
***
–SG
t
*?AT
t
0.0566
***
-0.0321
***
-0.0439
***
RNOA
t
*SG
t
-0.0448
–RNOA
t
*?AT
t
-0.0534
–RNOA
t
*SG
t
*?AT
t
0.0405
Intercept -0.0646
***
0.0553
***
-0.0551
***
R
2
Adj. R
2
48.2%
47.7%
54.5%
54.4%
54.8%
54.7%
Sample Size 7,285 3,310 3,310
Notes to table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for definition of variables.
† The reduced sample set excludes companies making losses in the current or next period.
results for equation (1a) including an interaction
term estimated for the reduced sample. The
coefficient on TACC
t
is still significantly
smaller than zero (?
2
=-0.10). The interaction
term, RNOA
t
*TACC
t
, is negative and significant
(p <0.01) indicating that higher accruals are
associated with lower current profitability and
that this is negatively related to period ahead
profitability.
5.2 Persistence of the growth and efficiency
components of accruals
We extend the analysis to consider the causes of
the differential persistence level between
accruals and cash earnings by decomposing
accruals into three components. Panel B of Table
4 provides empirical results for equation (2). As
shown in column (1), coefficient estimates for
sales growth (SG
t
) and change in asset turnover
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
105
(-?AT
t
) are both unexpectedly positive and
significant using the full sample. Again, the
model is re-estimated using the reduced sample
of profitable companies. For the reduced sample,
both ?
2
and ?
3
become significantly negative at
less than 1% level of significance, suggesting
that both growth and efficiency components
contribute to the lower persistence of accrual
earnings. The expectations for lower persistence
of these components are therefore supported
using the reduced sample of profitable
companies. The values for ?
2
and ?
3
are -0.05 and
-0.06 respectively. RSST (2006) report ?
2
and?
3
to be -0.10 and -0.15 respectively. The results
found in the reduced sample set of Australian
companies are consistent with those reported in
RSST (2006). For completeness we also report a
model including interaction terms in column (3).
Three interaction terms measuring the
interactions between the components of accruals
and earnings are added to equation (2). The
coefficients on SG
t
and -?AT
t
are still
significantly negative. The coefficients on
RNOA
t
*SG
t
and RNOA
t
*?AT
t
are found to be
insignificantly different from zero. For the
reduced sample, the components of accruals
affect the current-future earnings relation
collectively not individually.
To summarise, the pattern of persistence
found in the U.S. data is only supported in
Australia for the reduced sample set of profitable
companies. This suggests that earnings
persistence of profitable companies might
behave differently fromearnings persistence of
loss-making companies.
5.3 Results considering the differential
persistence of accruals for loss-making
companies
To explore the difference in the informativeness
of accruals between loss-making and profitable
companies, we provide regression results for
equation (3) allowing for variation in the
coefficients between profitable and loss-making
companies. These results are reported in Tables
5 and 6. Column (1) of Table 5 reports the
results for the basic earnings persistence
specification. The coefficient on RNOA
t
is
estimated at 0.48 and is significantly less than
one. The incremental intercept term for loss
companies is negative and significant. The
coefficient attached to Loss*RNOA
t
is found to
be significantly positive indicating earnings are
more persistent for loss-making companies.
Table 5
Analysis of the Impacts of Losses on the Persistence of Accruals Earnings
1 3 2 1 0 1
*
+ +
+ + + + =
t t t t
RNOA Loss Loss RNOA RNOA ? ? ? ? ?
(1)
1 5 4 3 2 1 0 1
* *
+ +
+ + + + + + =
t t t t t t
TACC Loss RNOA Loss Loss TACC RNOA RNOA ? ? ? ? ? ? ?
(2)
1 7 6
5 4 3 2 1 0 1
* * *
* *
+
+
+ + +
+ + + + + =
t t t t
t t t t t t
RNOA TACC Loss TACC Loss
RNOA Loss Loss RNOA TACC TACC RNOA RNOA
? ? ?
? ? ? ? ? ?
(3)
Pred. (1) (2) (3)
RNOA
t
<1 0.4796
***
0.5014
***
0.6166
***
TACC
t
–
-0.0170 0.0401
*
RNOA
t
*TACC
t
-0.2854
**
Loss +/– -0.1225
***
-0.1209
***
-0.1064
***
Loss* RNOA
t
+/– 0.2405
***
0.2637
***
0.1566
**
Loss* TACC
t
+/– 0.3842
***
0.1105
**
Loss* TACC
t
* RNOA
t
-0.0838
Intercept 0.0137 0.0141 0.0006
R
2
Adj. R
2
47.4%
47.0%
51.0%
50.5%
52.4%
51.6%
Sample Size 7,285 7,285 7,285
Notes to Table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for definition of variables.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
106
Column (2) reports the estimated coefficients
for a regression allowing differing persistence
between cash and accrual earnings. Firstly, the
coefficient on TACC is estimated to be negative
but not significantly different from zero.
Secondly, the coefficient on Loss*RNOA
t
is
found to be significantly positive. This suggests
that the cash flow component of earnings is more
persistent for loss-making companies. Thirdly,
the coefficient attached to Loss*TACC
t
is
estimated at 0.38 and is significantly different
fromzero. A null hypothesis that there is no
difference in the persistence of the accrual
component of earnings between loss-making and
profitable companies can be rejected (t=11.89,
p<0.0001). The accrual component of earnings is
more persistent than the cash flow component for
loss-making companies.
The analysis is extended to incorporate
interaction terms. In addition to RNOA
t
*TACC
t
,
measuring the impact of accruals on current and
future earnings relation for profitable companies,
Loss* RNOA
t
*TACC
t
is also added to capture the
differential impact of accruals between profitable
and loss-making companies. Results for the basic
loss model incorporating interaction terms
between accruals and earnings are reported in
column (3) of Table 5. This time, the coefficient
on TACC
t
is estimated to be significantly positive
at the ten percent level indicating accruals are
more persistent than cash for profitable
companies. Loss*TACC
t
remains significantly
positive consistent with accruals being more
persistent for loss-making companies than
profitable companies. RNOA
t
*TACC
t
is
significantly negative indicating total accruals
negatively affect the relationship between current
and future for profitable companies. The
coefficient on Loss* RNOA
t
*TACC
t
is not
however significantly different fromzero.
Table 6 extends the analysis to consider the
components of accruals and the differences
between loss-making and profitable companies.
Column (1) of Table 6 presents the regression
results. For profitable companies, the coefficients
on both the sales growth (SG
t
) and the efficiency
(-?AT
t
) component of accruals are estimated to
be negative. However, only the coefficient on the
growth component is found to be significantly
different fromzero at the five percent level. For
loss-making companies, it is shown that the
coefficients estimated for Loss* SG
t
and
-Loss*?AT
t
are both significantly greater than
zero (Loss* SG
t
=0.25 and -Loss*?AT
t
=0.27).
The growth and efficiency components of
accruals are both found to be more persistent than
the cash flow component of earnings for loss-
making companies. An F-test for a pooled
regression (not reported) confirms that for loss-
making companies, both the growth and
efficiency components of accruals are more
persistent than the cash flow component of
earnings (SG
t
+Loss* SG
t
>0 and -?AT
t
+
-Loss*?AT
t
>0).
Results for equation (3) incorporating the
interaction terms between accruals components
and earnings are reported in column (2) of Table
6. For profitable companies, both SG
t
and -?AT
t
are not significantly different from zero. The
coefficient on RNOA
t
*SG
t
is significantly
negative, indicating that the negative impact of
accruals on current and future earnings is mainly
attributable to the growth component of earnings.
For loss-making companies, both Loss*SG
t
and -
Loss*?AT
t
are estimated to be significantly
greater than zero (Loss* SG
t
=0.10 and -Loss*
?AT
t
=0.14), consistent with the accruals
components being more persistent for loss-
making companies.
In summary, the growth and efficiency
components of accruals are both found to be
more persistent than the cash flow component of
earnings for loss-making companies. This is
consistent with the expectation that small
development-stage loss-making companies
exhibit higher accruals persistence.
5.4 Further analyses
5.4.1 Current versus non-current accruals
In order to investigate the source of the
differential persistence level observed for the
efficiency components of accruals between loss-
making and profitable companies, we replace
the efficiency component of accruals in
equation (3) with two disaggregated items.
?AT
t
WC
measures the change in working capital
turnover
9
; and ?AT
t
NCO
measures the change in
non-current operating capital turnover
10
. Results
9 AT
t
WC
is calculated as Sales
t
/WC
t
. WC
t
is the working
capital at time t, defined as Current Assets – Cash and
Equivalents – Short-termInvestment – Current Liabilities
+Current Debts +Tax Payable.
10 AT
t
NCO
is calculated as Sales
t
/NCO
t
. NCO
t
is the non-
current operating capital at timet, defined as Non-current
Assets – Long-termInvestment – Non-current Liabilities +
Non-Current Debt.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
107
Table 6
Analysis of the Impact of Losses on the Persistence of the
Components of Accruals
1 9 8 7
6 5 4 3 2 1 0 1
* * *
*
+
+
+ ? ? ? ? ? +
+ + ? ? ? ? ? + + =
t t t t t
t t t t t t t
AT SG Loss AT Loss SG Loss
RNOA Loss Loss AT SG AT SG RNOA RNOA
? ? ? ?
? ? ? ? ? ? ?
(1)
1 15
14 13 12
11 10 9 8 7
6 5 4 3 2 1 0 1
* *
* * * * *
* * * *
* *
+
+
+ ? ? +
? + + ? ? ?
? ? + + + ? ? +
? + + ? ? ? ? ? + + =
t t t t t
t t t t t t
t t t t t t
t t t t t t t t t t
AT SG RNOA Loss
AT RNOA Loss SG RNOA Loss AT SG Loss
AT Loss SG Loss RNOA Loss Loss AT SG RNOA
AT RNOA SG RNOA AT SG AT SG RNOA RNOA
? ?
? ? ?
? ? ? ? ?
? ? ? ? ? ? ?
(2)
1 11 10 9 8
7 6 5 4 3 2 1 0 1
* * * *
*
+
+
+ ? ? ? ? ? ? ? +
+ + ? ? ? ? ? ? ? + + =
t t t
NCO WC
t
t t t
NCO WC
t t t
AT SG Loss AT Loss AT Loss SG Loss
RNOA Loss Loss AT SG AT AT SG RNOA RNOA
t t
t t
? ? ? ? ?
? ? ? ? ? ? ? ?
(3)
Pred. (1) (2) (3)
RNOA
t
<1 0.5362
***
0.6666
***
0.5324
***
SG
t
– -0.0393
**
0.0082 -0.0404
**
–? AT
t
– -0.0128 0.0008
–?AT
t
WC
0.0039
–?AT
t
NCO
-0.0272
–SG
t
*?AT
t
0.0251 0.0171 0.0313
RNOA
t
*SG
t
-0.3429
**
–RNOA
t
*?AT
t
-0.1891
–RNOA
t
*SG
t
*?AT
t
0.1621
**
Loss +/– -0.1350
***
-0.1137
***
-0.1392
***
Loss* RNOA
t
+/– 0.2160
***
0.0971 0.1962
***
Loss* SG
t
+/– 0.2511
***
0.1026
**
0.1817
***
–Loss*?AT
t
+/– 0.2712
***
0.1375
***
–Loss*?AT
t
WC
-0.0008
–Loss*?AT
t
NCO
0.1484
***
–Loss* SG
t
*?AT
t
0.0446 0.0520 0.0141
Loss*RNOA
t
*SG
t
0.1612
–Loss*RNOA
t
*?AT
t
-0.0525
–Loss*RNOA
t
*SG
t
*?AT
t
-0.2244
Intercept 0.0188
**
0.0060 0.0203
**
R
2
Adj. R
2
51.0%
50.0%
53.4%
52.2%
50.0%
48.7%
Sample Size 7,285 7,285 7,285
Notes to Table:
***, **, and * denotes significance at the 1%, 5%, 10% levels (two-tailed), respectively.
Refer to Table 2 for the definition of most variables.
The additional variables for this table are:
1. ?AT
t
WC
=the change in working capital turnover [(AT
t
WC
- AT
t-1
WC
) /AT
t
WC
], where AT
t
WC
is calculated as Sales
t
/ WC
t
;
2. ?AT
t
NCO
=the change in non-current operating capital turnover [(AT
t
NCO
- AT
t-1
NCO
) /AT
t
NCO
], where AT
t
NCO
is
calculated as Sales
t
/ NCO
t
;
3. WC
t
=Working Capitals, defined as Current Assets – Cash and Equivalents – Short-term Investment – Current
Liabilities + Current Debt +Tax Payable;
4. NCO
t
=Non-current operating Capitals, defined as Non-Current Assets – Long-term Investment – Non-Current
Liabilities + Non-Current Debt.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
108
are reported in column (3) of Table 6.
For profitable companies, the coefficients on
both -?AT
t
WC
and -?AT
t
NCO
are not significantly
different from zero. For loss-making
companies, (-Loss*?AT
t
NCO
) is significantly
positive, while (-Loss*?AT
t
WC
) is found to be
insignificantly different from zero. The
differential persistence level of the efficiency
component observed between profitable and
loss-making companies is primarily attributable
to the non-current operating capital component.
5.4.2 The influence of investing and
financing activities
Because the RSST (2006) measure of accruals
does not distinguish between operating,
investing and financing activities, we test the
sensitivity of our results to the inclusion of
current cash flows from investing (CFI) and
financing activities (CFF). We define CFI as
total investing cash flows scaled by lagged NOA
and CFF as total financing cash flows scaled by
lagged NOA. We include both CFI and CFF as
control variables to equation (1), (2) and (3).
For equation (3), we also allow the implications
of CFI and CFF for future profitability to vary
between profitable and loss-making companies.
Results (unreported) show that the inclusion of
CFI and CFF do not significantly alter the
inferences reported above. We do however find
that both CFI and CFF are useful in predicting
future profitability for both profitable and loss-
making companies suggesting that the RSST
(2006) decomposition of accruals could
potentially be extended to consider these aspects
of a company’s activities.
5.4.3 Alternate definition of accruals
We examine the robustness of the results to
using an alternative definition for total accruals.
We defined total accruals as the residual value
of earnings after deducting the cash flow
component (Hribar and Collins 2002). Accruals
are calculated as the difference between RNOA
and cash flow fromoperating activities. When
equation (1) is estimated for the reduced
sample, the coefficients on the persistence of
accruals are qualitatively similar to those
reported.
11
We extend the analysis by
decomposing total accruals based on the
11 The model is estimated for years 1993-2004 only, since
cash flow data obtained before 1993 is considered to be
less reliable.
alternate definition into the growth component
of accruals (change in sales growth) and a
residual accrual term defined as the difference
between total accruals and the growth
component. Unreported results show that the
growth component of accruals is still
significantly less persistent than the cash flow
component of earnings for profitable
companies. The growth component of accruals
is again found to be more persistent for loss-
making companies than profitable companies.
5.4.4 Industry effects
There is a concern that the results reported
above might be industry specific (Francis and
Smith 2005). We therefore re-estimated
equation (3) for the 22 industries identified
using ASX industry code. The mean
coefficients are examined using this approach.
The results are qualitatively similar to those
reported in tables 5 and 6. There is however
substantial variation in coefficients across
industries. Future research might consider the
systematic differences in accruals and
persistence of accruals across industries but that
is beyond the scope of this study.
5.4.5 Relation to returns
For completeness the impact on stock returns
was also explored to a limited extent. Clinch et
al (2006) find that for Australian companies,
investors misinterpret the association between
current cash flows and future earnings to a
greater extent than the association between
current accruals and future earnings. To test
whether accruals, and components of accruals
can predict future stock returns, we replicate the
model controlling for losses by replacing the
dependent variable in equation (3) with stock
returns (not reported for brevity). Returns were
defined as the next year’s annual buy-and-hold
returns measured starting three months after the
current fiscal year end. Monthly returns were
sourced from the SPPR database. The
explanatory power obtained for these return
models is relatively low (the average R-square
for our models is around one percent). The
accrual component of earnings is found to be
negatively related to next year’s return. Both the
growth and efficiency components of accruals
are also found to be significantly negative,
indicating that both components contribute to
the negative relation between accruals and
future stock return. We essentially find no
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Components of Accruals, Losses and Future Profitability
109
significant differences between the profitable
and loss-making companies.
6. Conclusion
This research examines persistence of accruals
for Australian companies. Prior research using
U.S. samples suggests that the accrual
component of earnings is less persistent than the
cash flow component of earnings (Sloan 1996;
Xie 2001; RSST 2005), indicating that the low
earnings persistence is largely attributable to
accruals. Prior research also suggests that two
primary factors contribute to the lower
persistence of accruals: the diminishing
marginal returns on new investments and
accounting distortions (RSST 2006).
This study finds evidence that the accrual
component of earnings is less persistent than the
cash flow component of earnings for profitable
Australian companies. The results indicate that
the growth and efficiency components of
accruals are less persistent than the cash flow
component of earnings for profitable
companies.
The results suggest that the persistence of the
components of accruals are sensitive to
companies’ current profitability. For our
sample, we find that loss-making companies
have higher persistence of the growth and
efficiency components of accruals than the cash
flow component of earnings, suggesting that for
currently unprofitable companies, the accrual
components of earnings are more informative
for period ahead profitability than the cash flow
component of earnings.
The results also suggest that the RSST
(2006) approach to measuring persistence of
accruals requires further consideration. Results
for models allowing interaction between
accruals and earnings suggest that the
interaction is explaining future profitability.
The results must be interpreted with respect
to several limitations. Firstly, consistent with
previous studies in this line of research only a
single future period is examined. The
implications of current earnings, and earnings
components for future periods beyond one year
could be different using a more complete time
series where data permits. Secondly, results
suggest that the higher accruals persistence for
loss-making companies is mostly consistent
with the argument that small companies in the
development stage experience consistent losses
for several years. This does not however mean
that one can ignore the impact of conservative
accounting practices. To test the impacts of
these two factors, new models would need to be
developed. Thirdly, this line of research ignores
any reliability issues associated with the cash
flow component of earnings. Some recent
studies argue that the cash flows are also subject
to managerial manipulation (Roychowdhury
2006; Graham et al 2005) and future research
could also consider the transient components of
operating cash flows.
References:
Balkrishna, H., Coulton, J.J. and Taylor, S.L. (2007),
“Accounting Losses and Earnings Conservatism:
Evidence from Australian GAAP”, Accounting and
Finance, Forthcoming.
Basu, S. (1997), “The Conservatism Principle and the
Asymmetric Timeliness of Earnings”, Journal of
Accounting and Economics, vol. 24, pp. 3–37.
Chan, K., Chan, L., Jegadeesh, N. and Lakonishok, J.
(2006), “Earnings Quality and Stock Returns”, Journal
of Business, vol. 79, pp. 1041–1082.
Clinch, G., Fuller, D., Govendir, B. and Wells, P. (2007),
“The Accrual Anomaly: Australian Evidence”, Working
Paper, University of Technology, Sydney.
Coulton, J., Taylor, S. and Taylor, S.L. (2005), “Is
‘Benchmark Beating’ by Australian Firms Evidence of
Earnings Management”, Accounting and Finance,
vol. 45, pp. 553–576.
Davidson, R., Stewart, J.G. and Kent, P. (2005), “Internal
Governance Structures and Earnings Management”,
Accounting and Finance, vol. 45, pp. 241–267.
Dechow, P.M. and Dichev, I.D. (2002), “The Quality of
Accruals and Earnings: The Role of Accrual Estimation
Errors”, The Accounting Review, vol. 77 (Supplement),
pp. 35–59.
Eddey, P.H. and Taylor, S.L. (1999), “Directors’
Recommendations on Takeover Bids and the
Management of Earnings: Evidence from Australian
Takeovers”, ABACUS, vol. 35, pp. 29–45.
Fairfield, P., Whisenant, S. and Yohn, T.L. (2003), “The
Differential Persistence of Accruals and Cash Flows for
Future Operating Income versus Future Profitability”,
Review of Accounting Studies, vol. 8, pp. 221–243.
Fama, E.F. and MacBeth, J.D. (1973), “Risk, Return and
Equilibrium: Empirical Tests”, The Journal of Political
Economy, vol. 81, pp. 607–636.
Francis, J. and Smith, M. (2005), “A Re-examination of the
Persistence of Accruals and Cash Flow”, Journal of
Accounting Research, vol. 43, pp. 413–451.
Givoly, D. and Hayn, C. (2000), “The Changing Time-
Series Properties of Earnings, Cash Flows and Accruals:
Has Financial Reporting Become More Conservative”,
Journal of Accounting and Economics, vol. 29,
pp. 287–320.
Godfrey, J.M., Mather, P. and Ramsay, A. (2003), “Earnings
and Impression Management in Financial Reports: The
Case of CEO Changes”, Abacus, vol. 39, pp. 95–123.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ACCOUNTING RESEARCH JOURNAL VOLUME 20 NO 2 (2007)
110
Graham, J.R., Harvey, C.R. and Rajgopal, S. (2005), “The
Economic Implications of Corporate Financial
Reporting”, Journal of Accounting and Economics,
vol. 40, pp. 3–73.
Hayn, C. (1995), “The Information Content of Losses”,
Journal of Accounting and Economics, vol. 20,
pp. 125–154.
Hribar, P. and Collins, D.W. (2002), “Errors in Estimating
Accruals: Implications for Empirical Research”,
Journal of Accounting Research, vol. 40, pp. 105–134.
Jones, J. J. (1991), “Earnings Management during Import
Relief Investigations”, Journal of Accounting Research,
vol. 29(2), pp. 1–31.
Joos, P. and Plesko, G.A. (2005), “Valuing Loss Firms”, The
Accounting Review, vol. 80, pp. 847–870.
Kang, T. (2005), “Earnings prediction and the role of
accrual-related disclosure: international evidence”,
Accounting Research Journal, vol. 18, pp. 6–12.
Kean, S. and Wells, P. (2006), “Earnings Decomposition and
the Persistence of Earnings”, Working Paper, University
of Technology, Sydney.
Kelley, S.O., Koh, P.S., and Tong, Y.H. (2004),
“Governance Structures and Abnormal Accruals
Characteristics”, Working Paper, University of
Queensland.
Klein, A. and Marquardt, C.A. (2006), "Fundamentals of
Accounting Losses", The Accounting Review, vol. 81,
pp. 179–206.
Koh, P.S. (2003), “On the Association between Institutional
Ownership and Aggressive Corporate Earnings
Management in Australia”, The British Accounting
Review, vol. 35, pp. 105–128.
Lonergan, W. and Wells, P. (2006), “Changes in the
Association Between Stock Prices and Book Value of
Equity and Earnings over the Period 1979–1997”,
Working Paper, University of Technology, Sydney.
Oei, R., Ramsay, A., and Mather, P. (2006), “Earnings
persistence, accruals and managerial share ownership”,
Working Paper, Monash University.
Richardson, S.A., Sloan, R.G., Soliman, M.T. and Tuna, I.
(2005), “Accrual Reliability, Earnings Persistence and
Stock Prices”, Journal of Accounting and Economics,
vol. 39, pp. 437–485.
Richardson, S.A., Sloan, R.G., Soliman, M.T. and Tuna, I.
(2006), “The Implication of Accounting Distortions and
Growth for Accruals and Profitability”, The Accounting
Review, vol. 81, pp. 713–743.
Roychowdhury, S. (2006), “Earnings Management through
Real Activities Manipulation”, Journal of Accounting
and Economics, vol. 42, pp. 335–370.
Sloan, R.G. (1996), “Do Stock Prices Fully Reflect
Information in Accruals and Cash Flows about Future
Earnings”, The Accounting Review, vol. 71,
pp. 289–315.
Taylor, S. and Taylor, S.L. (2003), “Earnings Conservatism
in a Continuous Disclosure Environment: Empirical
Evidence”, Working Paper, University of Melbourne
and University of New South Wales.
Watson, J. and Wells, P. (2005), “The Association Between
Various Earnings and Cash Flow Measures of Firm
Performance and Stock Returns: Some Australian
Evidence”, Working Paper, University of Technology,
Sydney.
Wells, P. (2002), “Earnings Management Surrounding CEO
Changes”, Accounting and Finance, vol. 42,
pp. 169–193.
Xie, H. (2001), “The Mispricing of Abnormal Accruals”,
The Accounting Review, vol. 76, pp. 357–373.
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Greg Clinch, Damian Fuller, Brett Govendir, Peter Wells. 2012. The accrual anomaly: Australian evidence. Accounting &
Finance 52:10.1111/acfi.2012.52.issue-2, 377-394. [CrossRef]
2. Kristen Anderson, Kerrie Woodhouse, Alan Ramsay, Robert Faff. 2009. Testing for asymmetric effects in the accrual anomaly
using piecewise linear regressions. Pacific Accounting Review 21:1, 5-25. [Abstract] [Full Text] [PDF]
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_506171159.pdf