Description
The purpose of this paper is to investigate the properties of net income (NI) and total
comprehensive income (TCI) of listed companies in New Zealand (NZ). Four properties of TCI and NI
are examined: persistence, variability, predictive ability, and value relevance. Whether the value
relevance of TCI depends on its reporting location is also investigated.
Accounting Research Journal
Properties of net income and total comprehensive income: New Zealand evidence
M. Humayun Kabir Fawzi Laswad
Article information:
To cite this document:
M. Humayun Kabir Fawzi Laswad, (2011),"Properties of net income and total comprehensive income: New
Zealand evidence", Accounting Research J ournal, Vol. 24 Iss 3 pp. 268 - 289
Permanent link to this document:
http://dx.doi.org/10.1108/10309611111187000
Downloaded on: 24 January 2016, At: 21:14 (PT)
References: this document contains references to 45 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1099 times since 2011*
Users who downloaded this article also downloaded:
J acqueline Birt, Greg Shailer, (2011),"Forecasting confidence under segment reporting", Accounting
Research J ournal, Vol. 24 Iss 3 pp. 245-267 http://dx.doi.org/10.1108/10309611111186993
Kevin Clarke, J ack Flanagan, Sharron O'Neill, (2011),"Winning ARC grants: comparing accounting with
other commerce-related disciplines", Accounting Research J ournal, Vol. 24 Iss 3 pp. 213-244 http://
dx.doi.org/10.1108/10309611111186984
Victoria J . Clout, Larelle Chapple, Nilan Gandhi, (2013),"The impact of auditor independence regulations
on established and emerging firms", Accounting Research J ournal, Vol. 26 Iss 2 pp. 88-108 http://
dx.doi.org/10.1108/ARJ -Dec-2011-0045
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Properties of net income
and total comprehensive income:
New Zealand evidence
M. Humayun Kabir
Department of Accounting, Faculty of Business,
Auckland University of Technology, Auckland, New Zealand, and
Fawzi Laswad
School of Accountancy, Massey University, Palmerston North, New Zealand
Abstract
Purpose – The purpose of this paper is to investigate the properties of net income (NI) and total
comprehensive income (TCI) of listed companies in New Zealand (NZ). Four properties of TCI and NI
are examined: persistence, variability, predictive ability, and value relevance. Whether the value
relevance of TCI depends on its reporting location is also investigated.
Design/methodology/approach – A cross-sectional research design is used with data on TCI
reported by NZ listed companies in 2010 under the new disclosure requirement in IAS 1. Ordinary least
squares (OLS) regressions are used with a sample of 86 ?rms to test for persistence, variability, and
predictive ability, and 81 ?rms to test for value relevance of NI and TCI.
Findings – The study ?nds: NI is potentially more persistent than TCI and potentially explains
contemporaneous stock returns better than TCI; no signi?cant difference in the variability and
predictive ability of NI and TCI; little evidence that the value relevance of TCI depends on its reporting
location; other comprehensive income (OCI) has incremental ability to predict one-year-ahead CFO,
although the incremental ability of OCI to predict one-year-ahead NI is not statistically signi?cant;
and OCI is not incrementally value relevant.
Practical implications – The ?ndings would be of interest to securities analysts and other users in
valuing ?rms and when earnings are used in contractual settings (e.g. management compensation).
Further, the results would also be of potential interest to standard-setters.
Originality/value – The literature on comprehensive income is growing. However, the authors are
not aware of any study that investigates the properties of NI and TCI in accordance with the new
requirement to report comprehensive income in the amended IAS 1 which came into effect in NZ on
January 1, 2009. The paper adds current evidence on the properties of NI and TCI under IFRS to the
international literature.
Keywords New Zealand, Earnings, Income, Listed companies, Disclosure, Persistence, Variability,
Predictive ability, Value relevance, Comprehensive income
Paper type Research paper
Introduction
International Accounting Standard (IAS) 1 “Presentation of Financial Statements”
requires entities to report other comprehensive income (OCI) and total comprehensive
income (TCI) in a statement of comprehensive income for periods starting on or after
January 1, 2009 (IASB, 2009, para 10)[1]. IAS 1 allows two reporting formats:
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors wish to thank the two anonymous reviewers for their helpful comments on an earlier
version of this paper.
ARJ
24,3
268
Accounting Research Journal
Vol. 24 No. 3, 2011
pp. 268-289
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111187000
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(1) a single statement of comprehensive income including all revenues and
expenses; or
(2) two statements, a separate income statement and a statement of comprehensive
income (IASB, 2009, para 81).
Most New Zealand (NZ) entities reported comprehensive income during 2010 for the
?rst time in accordance with IAS 1[2].
This study examines the properties of net income (NI) and TCI of listed companies
in NZ. There is a growing body of literature on the properties of comprehensive income
(Dhaliwal et al., 1999; O’Hanlon and Pope, 1999; Cahan et al., 2000; Chambers et al.,
2007; Barton et al., 2009). Most of these studies employed estimates of TCI. Further,
studies investigating properties of NI and TCI under IFRS are sparse. Two studies
(Cahan et al., 2000; Barton et al., 2009) that examine the properties of measures of ?rm
performance in NZ relate to previous NZ GAAP before the adoption of International
Financial Reporting Standards (IFRS) in NZ[3].
There have been two signi?cant changes in NZ that are relevant to this study
subsequent to the period covered in Cahan et al. (2000) and Barton et al. (2009). First, NZ
?rms started preparing their ?nancial statements in accordance with IFRS in 2007[4],
and adoption of IFRS impacted on the ?nancial statement amounts of listed companies
in NZ (Kabir et al., 2010; Stent et al., 2010). Second, the requirement to report TCI in a
performance statement came into effect in NZ for the ?rst time for periods starting on
or after January 1, 2009. Under pre-IFRS NZ GAAP, components of comprehensive
income were reported in a statement of changes in equity (Cahan et al., 2000).
Research suggests that the disclosure of comprehensive income may matter to users
depending on the reporting location (Hirst and Hopkins, 1998; Maines and McDaniel,
2000; Chambers et al., 2007). Further, the relative properties of NI and TCI may vary
under different sets of GAAP (e.g. US GAAP, IFRS) as the components of NI and TCI
may vary under each set[5]. Hence, the results of prior studies on properties of
comprehensive income may not be generalized to the current IFRS setting in NZ.
This study utilizes data on comprehensive income that were reported in 2010 in
accordance with the requirements of IAS 1.
Dechow and Schrand (2004) de?ne earnings as of higher quality if they are:
.
more persistent and less volatile;
.
more strongly associated with future cash ?ows; and
.
more strongly associated with contemporaneous stock price performance or
market value[6].
Following Dechow and Schrand (2004), this paper investigates four properties of NI
and TCI:
(1) persistence;
(2) cross-sectional variation in the income metrics;
(3) ability to predict one-year-ahead cash ?ows from operating activities (CFO) and
NI; and
(4) associations with contemporaneous stock returns.
Properties of
NI and TCI
269
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This paper also examines whether the value relevance of TCI depends on where it is
disclosed – a single statement or two-statement format as permitted under IAS 1.
Investigating the properties of NI and comprehensive income is important because
investors may want to know which metric, NI or TCI, measures ?rm performance
better (Black, 1993). Prior research indicates that earnings are used in a variety of
situations such as bonus contracts, debt covenants, and ?rm valuation (Dechow, 1994).
Since earnings are widely used by analysts in ?rm valuation (Dechow and Schrand,
2004), knowledge of the properties of NI and TCI would help analysts place appropriate
emphasis on the summary income metrics when valuing ?rms.
Investigating the properties of NI and TCI is also important from another
perspective. Given the increasing use of fair value in accounting standards, there are
concerns about the properties of income under fair value accounting (Barth, 2006).
In particular, Enria et al. (2004) argued that income under fair value accounting is more
volatile than under the historical cost method. Since TCI incorporates all the realised
and unrealized gains and losses recognised under IFRS, comparing its properties with
those of NI would provide insights into the properties of income under fair value
accounting. Investigating whether the value relevance of TCI depends on its reporting
location is important as IAS 1 allows the option to report TCI in a single statement or
two-statement format, and the IASB (2010) recently issued an exposure draft proposing
the elimination of the two-statement format of reporting TCI[7].
The sample comprises 86 ?rms used to test the persistence, variability, and
predictive ability of NI and TCI, and 81 ?rms used to test value relevance of NI and
TCI, drawing on 2010 data. This study ?nds that NI is potentially more persistent than
is TCI and potentially explains contemporaneous stock returns better than TCI. These
results are robust to alternative de?ators and standard errors corrected for industry
clusters. However, the study does not ?nd any signi?cant difference in the variability
and predictive ability of NI and TCI and little evidence that the value relevance of TCI
depends on its reporting location. Further, OCI is found to have incremental ability
to predict one-year-ahead CFO, although the incremental ability of OCI in predicting
one-year-ahead NI is not statistically signi?cant. The results also indicate that OCI
is not incrementally value relevant.
This study makes two contributions to the literature on the properties of NI and
TCI. First, it provides evidence on the properties of NI and TCI under IFRS. Evidence
on the properties of NI and TCI under IFRS is sparse as the IFRS requirement to report
TCI is new. The earlier evidence on comprehensive income relates to pre-IFRS NZ
GAAP and other local GAAP. Thus, this study contributes current evidence on
comprehensive income under a new reporting regime (i.e. IAS 1). Second, the study
provides evidence on whether the value relevance of TCI depends on the reporting
location of TCI. This is of potential interest to standard-setters as the IASB has
proposed the removal of the option of reporting TCI in two statements. The remainder
of the paper is structured as follows. Section 2 provides a brief review of the literature,
Section 3 develops the hypotheses, and Section 4 discusses the research design and
methodology. Section 5 presents the ?ndings and Section 6 concludes the paper.
2. Literature review
There is an extensive literature on the properties of ?rm performance measures.
Early studies examined the time series properties, persistence, predictive ability
ARJ
24,3
270
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and value relevance of earnings and cash ?ows. Ball and Watts (1972) found that
annual accounting income follows a sub-martingale process, and Watts and Leftwich
(1977) provided evidence suggesting that annual income follows a random walk
process[8]. Lipe’s (1986) ?ndings suggest that different components of earnings vary in
persistence, while Bowen et al. (1986) found that current period earnings are not better
predictors of future cash ?ows than current period cash ?ows. On the other hand,
?ndings of Dechow et al. (1998) indicate that current earnings forecast future operating
cash ?ows better than current operating cash ?ows. Barth et al. (2001a, b) found that
cash ?ows and accrual components of current earnings have better predictive ability
for future cash ?ows than do several lags of earnings. Dechow’s (1994) results suggest
that the association between earnings and contemporaneous stock returns is stronger
than that between cash ?ows and contemporaneous stock returns when:
.
the performance measurement interval is short;
.
the absolute magnitude of accruals is large; and
.
the operating cycle is long.
The literature on comprehensive income is growing. However, while most studies have
used estimates of TCI, Chambers et al. (2007) document that estimates of TCI are
subject to measurement errors[9]. Further, the ?ndings of studies comparing the
properties of NI and TCI are not consistent. One set of studies ?nds that NI is superior to
TCI in terms of value relevance, predictive ability and conservatism. Cheng et al. (1993)
found that operating income and NI explain security returns better than comprehensive
income. O’Hanlon and Pope (1999) investigated the value relevance of dirty surplus
accounting ?ows under UK GAAP over the period 1972-92 and document that ordinary
pro?t and extraordinary items are associated with stock returns, but found no evidence
that dirty surplus ?ows are value relevant. Using estimates of OCI and a sample of
?rms from 16 European countries during 1991-2005, Goncharov and Hodgson (2008)
found that NI is better than TCI in terms of value relevance and ability to predict future
cash ?ows from operations. Further, TCI is less conservative than NI in that TCI
recognises good news on a timelier basis than NI. Barton et al. (2009) examined the
value relevance of eight measures of ?rm performance in 46 countries during 1996-2005
and report that value relevance peaks for measures above the line and no individual
measure of ?rm performance dominates other measures in all countries.
In contrast, a second set of studies found no evidence that TCI is more value
relevant than NI. Using estimates of comprehensive income, Dhaliwal et al. (1999) did
not ?nd any evidence that comprehensive income is more strongly associated with
contemporaneous returns, future cash ?ows and future NI than NI during 1994-1995.
They, however, found that marketable securities adjustments reported by ?nancial
companies are value relevant. Cahan et al. (2000) examined the value relevance of
comprehensive income in NZ during 1992-1997 and do not ?nd evidence that the
incremental value relevance of TCI relative to NI increased after the issuance of
Financial Reporting Standard 2 in 1994. They further report that there is no evidence
that individual components of OCI are incrementally value relevant over and above
TCI. Isidro et al. (2004) document signi?cant cross-country variation in dirty surplus
?ows during 1993-2001, but found little evidence that omission of dirty surplus ?ows
from residual income value estimates would have caused signi?cant valuation errors.
Properties of
NI and TCI
271
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A third set of studies found that TCI and OCI are value relevant. Kubota et al. (2009)
found that dirty surplus items have information content. Chambers et al. (2007) used
as-reportedmeasures of OCI andfound that OCI is pricedbythe market inthe post-SFAS
130 period. Their ?ndings also suggest that foreign currency translation adjustments
and unrealized gains and losses on available-for-sale securities were positively priced.
Usinga sample of cross-listed Canadian ?rms, Kanagaretnamet al. (2009) study ?ndings
suggest that components of other comprehensive income are associated with stock
returns, and that aggregate comprehensive income is more strongly associated with
price and stock returns than is net income. However, their results suggest that net
income is a better predictor of future net income, comprehensive income and cash ?ows
from operations than is comprehensive income. Overall, evidence on the relative value
relevance of NI and TCI is mixed.
A fourth set of studies examined whether the reporting location of TCI (i.e. in a
performance statement or in a statement of stockholders’ equity) matters to users.
Hirst and Hopkins (1998) employed an experimental research design and found that
the reporting of comprehensive income in the income statement helps analysts detect
earnings management. In an experiment involving MBA students, Maines and
McDaniel (2000) report that participants attached signi?cant weight to unrealised gains
when they appeared in the statement of comprehensive income rather than when they
appeared in a statement of stockholders’ equity. However, the ?ndings of Chambers et al.
(2007) suggest that investors pay greater attention to OCI when they are reported in the
statement of changes in equity rather than when they are reported in a performance
statement. Evidence on which format of reporting TCI is desirable to users is mixed is
therefore inconclusive.
Several studies have examined managers’ choice of reporting location of TCI. The
Lee et al. (2006) ?ndings indicated that insurers with a tendency to manage income
through realised securities’ gains and losses, and with a reputation for poor ?nancial
reporting quality are more likely to report comprehensive income in a statement of equity.
Similarly, Bamber et al. (2010) found that US managers with strong equity incentives and
less job security are less likely to report comprehensive income in a performance
statement.
While the empirical literature on comprehensive income is increasing, studies
examining properties of TCI under IFRS are scarce. Two studies (Cahan et al., 2000;
Barton et al., 2009) that examined comprehensive income in NZ used data from the
pre-IFRS era. The adoption of IFRS in NZ impacted on income and other ?nancial
statement numbers of listed companies in NZ (Kabir et al., 2010, Stent et al., 2010).
Further, in contrast to pre-IFRS NZ GAAP under which total recognised revenue and
expenses were reported in a statement of changes in equity, IAS 1 requires
comprehensive income to be reported in a performance statement. Prior research
suggests that the reporting location of TCI matters to users (Hirst and Hopkins, 1998;
Maines and McDaniel, 2000; Chambers et al., 2007). Further, the components of NI and
TCI are not the same under different sets of GAAP and hence the relative properties of
NI and TCI may vary depending on GAAP. Thus, the results of earlier studies on
properties of TCI may not be generalisable to the current IFRS setting in NZ. This
paper seeks to ?ll this gap by examining properties of NI and TCI in NZ as required by
the recently amended IAS 1.
ARJ
24,3
272
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
3. Hypotheses
3.1 Persistence of NI and TCI
Total comprehensive income (TCI) is the sum of bet income (NI) and other
comprehensive income (OCI). The components of OCI include:
.
changes in revaluation surplus;
.
actuarial gains and losses on de?ned bene?t plans;
.
gains and losses arising from translating ?nancial statements of a foreign
operation;
.
gains and losses on re-measuring available-for-sale ?nancial assets; and
.
the effective portion of gains and losses on hedging instruments in a cash ?ow
hedge (IASB, 2009, para 7).
These components arise from changes in interest rates and exchange rates and other
random walk processes (Smithson et al., 1995, cited in Chambers et al., 2007). Thus, OCI
is transitory[10] in nature (Chambers et al., 2007), and NI is likely to be more persistent
than TCI. Therefore, the ?rst hypothesis is:
H1. NI is more persistent than TCI.
3.2 Variability of NI and TCI
One argument against fair value accounting is that it increases the volatility in
?nancial statements[11]. Barth et al. (1995) found that fair value-based earnings are
more volatile than historical cost-based earnings. Similarly, Hodder et al. (2006) found
that the volatility of full fair value income is more than three times that of
comprehensive income and more than ?ve times that of net income. Since TCI
incorporates more value changes than does NI, based on the evidence, NI is likely to be
less volatile than TCI. Thus, the second hypothesis is:
H2. The cross-sectional variation of NI is less than that of TCI.
3.3 Predictive ability of NI and TCI
Net income under IFRS includes some gains and losses from changes in fair value, and
accruals from the application of the revenue recognition and matching principles. TCI
includes further accruals that re?ect transitory revaluations of assets and liabilities.
Accruals that arise from the application of the revenue recognition and matching
principles inthe NI better predict future cash?ows andnet income thando accruals arising
from transitory valuation changes in TCI (Barth et al. 2001a, b; Dechow and Schrand,
2004). Hence, NI is a better predictor of future cash ?ows andnet income thanTCI (Dechow
and Schrand, 2004; Kanagaretnam et al., 2009). The third and fourth hypotheses are:
H3. NI predicts one-year-ahead CFO better than TCI.
H4. NI predicts one-year-ahead NI better than TCI.
3.4 Value relevance of NI and TCI
If NI is more persistent than TCI and predicts one-year-ahead CFO and NI better than
does TCI, investors may view NI more favourably than TCI. Prior research suggests
that investors view the persistence of earnings components as desirable[12], and that
Properties of
NI and TCI
273
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the information content of earnings components is increasing with the components’
persistence (Lipe, 1986; Penman and Zhang, 2002). Further, since TCI includes more
transitory items than does NI, the earnings response coef?cient of TCI is likely to be
less than that of NI (Kothari and Zimmerman, 1995).
On the other hand, it can be argued that comprehensive income is consistent with
accounting-based valuation theory that expresses equity price as the sum of current
book value and present value of future comprehensive residual income (Linsmeier et al.,
1997). Linsmeier et al. (1997, p. 122) conclude: “Thus, for reported income to be most
useful in equity price valuation, it must be comprehensive”. However, the value
relevance of comprehensive income depends on investors’ ability to process
multi-component comprehensive income (Goncharov and Hodgson, 2008). Hence,
given the above discussion and the mixed empirical evidence (Cheng et al., 1993;
Dhaliwal et al., 1999; O’Hanlon and Pope, 1999; Chambers et al., 2007; Goncharov and
Hodgson, 2008; Kanagaretnamet al., 2009) on the relative value relevance of NI and TCI,
the study does not predict the relative value relevance of NI and TCI. This is expressed in
the following non-directional hypothesis:
H5. The value relevance of NI differs from TCI.
3.5 Reporting location of TCI and its value relevance
IAS 1 allows the reporting of TCI in a statement of comprehensive income or in two
statements (a separate income statement and a statement of comprehensive income)
(IASB, 2009). The ef?cient market hypothesis appears to suggest that reporting
location is irrelevant as long as the information is publicly disclosed. However,
the proposal of the IASB (2010) to eliminate the current option of reporting TCI in a
two-statement format suggests that the reporting location matters. Further,
prior research suggests that the reporting location of TCI is potentially important as
it affects the perceptions of the importance of OCI and TCI disclosed (Maines and
McDaniel, 2000).
The evidence regarding the preferred reporting format of TCI is mixed. Hirst and
Hopkins (1998) ?nd that reporting comprehensive income in the income statement
helps analysts detect earnings management. Maines and McDaniel (2000) ?nd that
participants in an experiment attach more weight to unrealised gains when they
appear in the statement of comprehensive income than when they are disclosed in a
statement of stockholders’ equity. However, Chambers et al. (2007) found that investors
pay greater attention to OCI when they are reported in the statement of changes in
equity rather than when they are reported in a performance statement. Thus, given the
mixed evidence, this study makes no prediction on the relative value relevance of TCI
when it is disclosed in a single statement rather than when it is reported in a
two-statement format. The hypothesis is as follows:
H6. The value relevance of TCI differs depending on its reporting location.
4. Methodology
4.1 Data and sample
The authors hand-collected ?nancial statement data from the 2010 annual reports of
listed companies from NZX Deep Archive[13] and returns data from Datastream.
They started with ?rms that have 2010 annual reports on NZX Deep Archive.
ARJ
24,3
274
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A total of 145 NZ listed ?rms had 2010 annual reports at the time of data collection.
From this, they excluded ?rms:
.
whose NI and TCI are the same (i.e. no OCI items);
.
whose ?nancial statements are presented in foreign currency;
.
whose ?nancial statement ?gures do not match those in the corresponding
notes[14];
.
whose ?nancial statements are not consolidated;
.
whose ?scal years in 2009 and 2010 are not equal in length;
.
which are overseas companies; and
.
which prepared their ?nancial statements using foreign GAAP.
This process yielded a sample of 86 ?rms. We use these 86 ?rms for analyses of
persistence, variability, and predictive ability of NI and TCI. Five of the sample ?rms
do not have returns data. Hence, the authors utilize 81 ?rms for tests of association of
NI and TCI with contemporaneous stock returns. Table I summarises the sample
selection and industry distribution.
Panel A of Table I shows that TCI and NI is the same for 39 out of 145 ?rms. Hence,
for 73.10 percent of the population TCI differs from NI. This percentage is similar to
the percentage (71 percent) reported by Dhaliwal et al. (1999). Panel B shows the
sample ?rms come from a wide variety of industries, with the largest number being
from the consumer industry.
4.2 Models
Persistence. The following two standard models are used to assess the persistence of NI
and TCI (Dechow and Schrand, 2004). The closer b is to 1, the more persistent the
variable is (Dechow and Schrand, 2004):
NI
tþ1
¼ a þ bNI
t
þ 1
t
ð1Þ
TCI
tþ1
¼ a þ bTCI
t
þ 1
t
ð2Þ
where NI is net income and TCI is total comprehensive income. Following Dechow
(1994) and Barth et al. (1995), both NI and TCI are de?ated by the weighted average
number of shares to mitigate the effects of heteroskedasticity. Since 2010 ?nancial
statements report comparative ?nancial statement amounts for 2009, this paper
employs 2009 NI and TCI to predict 2010 NI and TCI, respectively.
Variability. Following Barth et al. (1995), this study assesses variability by standard
deviation and compares the standard deviation of NI with that of TCI in 2010.
Predictive ability. Models (3) and (4) are used to assess which income metric – NI and
TCI – can predict one-year-ahead CFO better. Following prior literature (Dechow,
1994), the model having a higher adjusted R
2
will be considered to have better
predictive ability. Vuong’s (1989) test is used to ?nd the signi?cance of the difference in
adjusted R
2
of the two models[15]. Similarly, models (5) and (6) are used to evaluate the
relative ability of NI and TCI to predict one-year-ahead NI.
CFO is used because the recent conceptual framework jointly issued by the FASB
and the International Accounting Standards Board (IASB) states that ?nancial
reporting should assist users assess the prospects for future cash ?ows to the entity
Properties of
NI and TCI
275
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(FASB, 2010, para. OB 3 & 4). Further, the newly issued framework notes that
information about the ?rm’s ?nancial performance during a period is helpful in
assessing the entity’s ability to generate future cash ?ows (FASB, 2010, para. OB 18).
The authors also use NI as the variable to be predicted because income is widely used
by analysts in ?rm valuation (Dechow and Schrand, 2004):
CFO
tþ1
¼ a þ bNI
t
þ 1
t
ð3Þ
CFO
tþ1
¼ a þ bTCI
t
þ 1
t
ð4Þ
NI
tþ1
¼ a þ bNI
t
þ 1
t
ð5Þ
NI
tþ1
¼ a þ bTCI
t
þ 1
t
ð6Þ
Panel A: derivation of sample
Companies on NZX Deep Archive 145
Less: NI and TCI are the same 39
Financial statements are presented in
foreign currency
3
Figures in notes do not match corresponding
?gures in ?nancial statements
2
Non-consolidated statements 1
Different lengths of periods in 2009 and 2010 4
Overseas companies 9
Companies that prepared their ?nancial
statements using foreign GAAP
1
Total number of companies
excluded from the sample
59
Sample for tests of persistence,
variability, and predictive ability
86
Less: Missing returns data 5
Sample for tests of association with
stock returns
81
Panel B: industry distribution of sample ?rms
Industry No. of ?rms
Agriculture and ?shing 8
Building materials and construction 3
Consumer 12
Energy processing 7
Finance and other services 10
Food and beverages 6
Intermediate and durables 9
Investment 7
Leisure and tourism 4
Media and telecommunications 4
Mining 2
Ports 5
Property 4
Textiles and apparel 2
Transport 3
Total 86
Table I.
Sample
ARJ
24,3
276
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
where CFO is cash ?ows from operating activities, NI is net income and TCI is total
comprehensive income. All three variables are de?ated by the weighted average
number of shares. Since 2010 ?nancial statements report comparative ?nancial
statement amounts for 2009 also, this paper uses 2009 NI and TCI to predict 2010 CFO
and NI.
Value relevance of NI and TCI. Two models are used widely in the capital
market-based accounting research: the returns model and the price model (Kothari and
Zimmerman, 1995). Since the focus of this study is on assessing which income metric – NI
or TCI – is a better measure of ?rm performance, this paper follows Dechow (1994) and
Easton (1999), and employs the returns model. Kothari and Zimmerman (1995) argue and
document that the earnings response coef?cient inthe returns model is biased toward zero
because of the “price leads earnings” phenomenon and the consequent omitted variable
problem. However, as Dechow (1994) and Easton (1999) argue convincingly, when
the research objective is to assess earnings as a summary measure of ?rm performance,
the effect of this omitted variable (i.e. events not summarized in earnings but in price)
becomes the focus of investigation. Further, the price model suffers from
heteroskedasticity and a scaling problem (Kothari and Zimmerman, 1995; Easton,
1999). Thus, Easton (1999, p. 411) concludes: “[. . .] the inferences from returns models are
probablymore reliable andshould be used.” The following models are usedto evaluate the
relative ability of NI and TCI to summarise ?rmperformance as re?ected in stock returns:
R
t
¼ a þ bðNIC
t
_P
t21
Þ þ 1
t
ð7Þ
R
t
¼ a þ bðTCIC
t
_P
t21
Þ þ 1
t
ð8Þ
where R
t
is the stock returns over the ?scal year minus returns on the NZX market
portfolio, NIC
t
_P
t21
is net income to common per share de?ated by beginning-of-year
price, and TCIC
t
_P
t21
is total comprehensive income to common per share de?ated by
beginning-of-year stock price. Following Dechow (1994) and Goncharov and Hodgson
(2008), in models (7)-(8), this paper uses ?nancial statement variables on a per share basis,
de?ated by beginning-of-period price. Following Dechow (1994), market-wide returns are
deducted from stock returns because they have low association with realized cash ?ows
and earnings, and this improves the power of the tests. Following prior literature (Dechow,
1994), the model having a higher adjusted R
2
will be taken to explain stock returns better,
and Vuong’s (1989) test is used to test the signi?cance of the difference in the adjusted R
2
of the two models.
Kothari and Zimmerman (1995) document that both the returns and price models
suffer from econometric and theoretical problems and hence suggest that researchers
use both models for de?nitive conclusions. Hence, following Dhaliwal et al. (1999), this
paper also assesses the relative value relevance of NI and TCI by running a price level
model. However, as Kothari and Zimmerman (1995) note, the price model does not
measure the information arrival during the period. Following Dhaliwal et al. (1999),
the paper runs the following two models:
P
t
¼ a þ b
1
BV
t
þ b
2
NIC
t
þ 1
t
ð9Þ
P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ 1
t
ð10Þ
where P
t
is price per share at the end of the ?scal year, BV
t
is book value of equity at
the end of the ?scal year, NIC
t
is net income available to the common shareholders,
Properties of
NI and TCI
277
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and TCIC
t
is total comprehensive income available to the common shareholders. The
?nancial statement variables are de?ated by the weighted average number of shares
during the ?scal year.
Reporting location of TCI and its value relevance. The following model is used to
assess whether the value relevance of TCI depends on where it is reported:
R
t
¼ a þ b
1
ðTCIC
t_
P
t21
Þ þ b
2
TCIC
t_
P
t21*
SINGLE þ 1
t
ð11Þ
where SINGLE is a binary variable that takes 1 if TCI is reported in the same statement
in which pro?t/loss is reported, 0 otherwise. All other variables are as de?ned in
model (8). A positive and signi?cant coef?cient of b
2
would be consistent with an
incremental value relevance of TCI when it is reported in the same statement in which
pro?t/loss is reported, rather than when a two-statement format is used for reporting
comprehensive income.
Given the limitations of the returns model, this paper also uses the following price
model to assess the impact of reporting location of TCI on its value relevance:
P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ b
3
TCIC
t *
SINGLE þ 1
t
ð12Þ
All variables are as de?ned in models (10) and (11). A positive and signi?cant
coef?cient of b
3
would be consistent with an incremental value relevance of TCI when
it is reported in a single statement of comprehensive income, rather than when it is
reported in a two-statement format.
5. Findings
5.1 Descriptive statistics and correlation matrices
Table II reports descriptive statistics of model variables and correlation coef?cients
between them. In Panel A the mean (median) NI per share (NI
tþ1
) and TCI per share
(TCI
tþ1
) in 2010 are $0.118 ($0.066) and $0.117 ($0.057), respectively, and are
statistically signi?cantly different from zero. The mean (median) OCI (not tabulated) in
2010 is 20.001 (0.000), which is not signi?cantly different from zero. During 2010, the
three most frequent items of OCI were cash ?ow hedge (50 ?rms), exchange difference
on translating foreign operations (45 ?rms), and changes in property, plant and
equipment revaluation reserve (18 ?rms). During 2010, 33 out of 86 ?rms reported TCI
in a single statement of comprehensive income.
Panels B and C of Table II report the correlation coef?cients between the model
variables. The correlation between net income and total comprehensive income is high
and positive. For example, the correlation coef?cients of NI
t
with TCI
t
is 0.894 and
NI
tþ1
with TCI
tþ1
is 0.991 in Panel B. Similarly, the correlation coef?cient of
NIC
t
_P
t21
with TCIC
t
_P
t21
is 0.963 and that of NIC
t
with TCIC
t
is 0.990. The high
correlations along with similar average magnitudes of net income and total
comprehensive income reported in Panel A may introduce bias against ?nding any
differences in the properties of net income and total comprehensive income.
The issue of multicollinearity does not arise in models (1)-(8) as these models employ
bivariate regressions. The highest correlation coef?cient between the variables in
models (9)-(12) is 0.439, which is between BV
t
and TCIC
t
. Thus, multicollinearity poses
no serious problem in the regressions in this study.
ARJ
24,3
278
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5.2 Properties of NI and TCI
Table III reports the results on the properties of NI and TCI. Panel A reports the results
on the persistence of NI and TCI. The adjusted R
2
is 0.166 and 0.150 for model (1) and (2),
respectively. Both models are signi?cant at less than 1 percent. The persistence
Variables N Mean Median SD
Panel A: descriptive statistics
NI
tþ1
86 0.118
* * *
0.066
* * *
0.412
NI
t
86 0.081
* * *
0.065
* * *
0.229
TCI
tþ1
86 0.117
* *
0.057
* * *
0.429
TCI
t
86 0.090
* * *
0.070
* * *
0.261
CFO
tþ1
86 0.239
* * *
0.143
* * *
0.281
R
t
81 20.30 20.017 0.381
NIC
t
_P
t21
81 0.001 0.047
* * *
0.203
TCIC
t
_
P
t21
81 20.004 0.049
* * *
0.217
P
t
81 2.030
* * *
1.520
* * *
1.928
BV
t
81 1.636
* * *
1.089
* * *
1.583
NIC
t
81 0.122
* * *
0.068
* * *
0.414
TCIC
t
81 0.121
* *
0.055
* * *
0.432
SINGLE 81 0.370
* * *
0.000
* * *
0.486
Panel B: correlation matrix for variables in models (1)-(6)
NI
tþ1
NI
t
TCI
tþ1
TCI
t
CFO
tþ1
NI
tþ1
1.000
NI
t
0.419
* * *
1.000
TCI
tþ1
0.991
* * *
0.437
* * *
1.000
TCI
t
0.376
* * *
0.894
* * *
0.399
* * *
1.000
CFO
tþ1
0.524
* * *
0.708
* * *
0.531
* * *
0.699
* * *
1.000
Panel C: correlation matrix for variables in models (7)-(12)
R
t
NIC
t
_P
t21
TCIC
t
_P
t21
P
t
BV
t
NIC
t
TCIC
t
SINGLE
R
t
1.000
NIC
t
_P
t21
0.313
* * *
1.000
TCIC
t
_
P
t21
0.264
* *
0.963
* * *
1.000
P
t
0.197
*
0.372
* * *
0.365
* * *
1.000
BV
t
0.146 0.300
* * *
0.286
* * *
0.789
* * *
1.000
NIC
t
0.104 0.510
* * *
0.495
* * *
0.541
* * *
0.437
* * *
1.000
TCIC
t
0.086 0.500
* * *
0.511
* * *
0.551
* * *
0.439
* * *
0.990
* * *
1.000
SINGLE 20.185
*
20.042 20.014 20.331
* * *
20.333
* * *
20.123 20.121 1.000
Notes: Statistically signi?cant at:
*
10,
* *
5 and
* * *
1 percent; NI
tþ1
¼ one-year-ahead net income per share;
NI
t
¼ current year net income per share; TCI
tþ1
¼ one-year-ahead total comprehensive income per share;
T CI
t
¼ current year total comprehensive income per share; CFO
tþ1
¼ one-year-ahead cash ?ows from
operating activities per share; R
t
¼ returns over the ?scal year minus retun on the NZX portfolio;
NIC
t
_P
t21
¼ current year net income to common per share de?ated by beginning-of-year share price;
TCIC
t
_P
t21
¼ current year total comprehensive income to common per share, de?ated by beginning-of-year
share price; P
t
¼ end-of-year share price; BV
t
¼ book value per share; NIC
t
¼ current year net income to
common per share; TCIC
t
¼ current year total comprehensive income to common per share; and SINGLE ¼ 1 if
the TCI is reported in a single statement of comprehensive income, 0 otherwise; the ?rst ?ve variables in this
table are those in models (1)-(6); the sample size for these models is 86. The last eight variables are those in
models (7)-(12); the sample size for these models is 81 as ?ve out of the 86 ?rms do not have returns data; all
?gures, except the ratios, are in dollars
Table II.
Descriptive statistics
and correlation matrices
Properties of
NI and TCI
279
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Panel A: persistence of NI and TCI
Independent
variables Model: NI
tþ1
¼ a þ bNI
t
þ e
t
(1) TCI
tþ1
¼ a þ bTCI
t
þ e
t
(2)
Constant 0.057 (2.338
* *
) 0.058 (1.901
*
)
NI
t
0.755 (2.340
* *
)
TCI
t
0.656 (3.057
* * *
)
N 86 86
Adjusted R
2
0.166 0.150
F-statistic 17.933
* * *
15.945
* * *
Panel B:
variability
of NI and TCI NI TCI
SD 0.412 0.429
N 86 86
F-statistic 1.087
Panel C: ability of NI and TCI to predict one-year-ahead CFO
Independent
variables Model: CFO
tþ1
¼ a þ bNI
t
þ e
t
(3) Model: CFO
tþ1
¼ a þ bTCI
t
þ e
t
(4)
Constant 0.169 (7.376
* * *
) 0.171 (7.403
* * *
)
NI
t
0.871 .(9.176
* * *
)
TCI
t
0.753 .(8.962
* * *
)
N 86 86
Adjusted R
2
0.495 0.483
F-statistic 84.200
* * *
80.318
* * *
Vuong z-statistic 1.008
Panel D: ability of NI and TCI to predict one-year-ahead NI
Independent
variables Model: NI
tþ1
¼ a þ bNI
t
þ e
t
(5) Model: NI
tþ1
¼ a þ bTCI
t
þ e
t
(6)
Constant 0.057 (2.338
* *
) 0.065 (2.255
* *
)
NI
t
0.755 (2.340
* *
)
TCI
t
0.593 (2.841
* * *
)
N 86 86
Adjusted R
2
0.166 0.131
F-statistic 17.933
* * *
13.840
* * *
Vuong z-statistic 1.341
Panel E: association with contemporaneous returns
Independent
variables Model:R
t
¼ a þ b(NIC
t
_P
t21
) þ e
t
(7) Model: R
t
¼ a þ b(TCIC
t
_P
t21
) þ e
t
(8)
Constant 20.031 (20.757) 20.028 (20.684)
NIC
t
_P
t21
0.586 (2.924
* * *
)
TCIC
t
_P
t21
0.464 (2.436
* *
)
N 81 81
Adjusted R
2
0.086 0.058
F-statistic 8.551
* * *
5.933
* *
Vuong z-statistic 1.117
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; NI
tþ1
¼ one-year-ahead net income per
share; NI
t
¼ current year net income per share; TCI
tþ1
¼ one-year-ahead total comprehensive income per
share; TCI
t
¼ current year total comprehensive income per share; CFO
tþ1
¼ one-year-ahead cash ?ows
from operating activities per share; R
t
¼ company stock returns over the ?scal year minus returns on the
NZX market portfolio; NIC
t
_P
t21
¼ current year net income to common per share de?ated by beginning-
of-year price; TCIC
t
_P
t21
¼ current year total comprehensive income to common per share de?ated by
beginning-of-year price; ?gures in parentheses are t-statistics; t-statistics in Panel A and D are White
(1980) heteroskedasticity-consistent
Table III.
Results on properties
of NI and TCI
ARJ
24,3
280
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
coef?cients of NI and TCI are 0.755 and 0.656, respectively. Thus, NI is potentially more
persistent than TCI. These results are consistent with Barton et al. (2009) who found TCI
is less persistent than NI. Un-tabulated results showthat the results remain similar when
NI and TCI are de?ated by alternative size measures. The persistence coef?cients are
1.312 and 0.801 for NI and TCI, respectively, when both NI and TCI are de?ated by total
assets. The corresponding coef?cients are 1.986 and 1.768, respectively, when de?ated
by market capitalisation at the end of the last ?scal year. These results are consistent
with H1.
Panel B shows the variability of NI and TCI. The standard deviation of NI and TCI is
0.412 and 0.429, respectively. Thus, consistent with H2, the variability of TCI is higher
than that of NI. However, the difference in variability is not signi?cant at the 10 percent
level. Further, untabulated results show that the results are sensitive to de?ators. When
NI and TCI are de?ated by total assets at the end of the year, the standard deviation of NI
(1.758) is higher than that of TCI (1.413) and the difference is signi?cant at less than
5 percent. When NI and TCI are de?ated by market capitalisation at the end of the
previous ?scal year, the standard deviation of NI and TCI is 0.677 and 0.654,
respectively. However, the difference in variability is not signi?cant at 10 percent. Thus,
overall the results do not support H2.
Panel Creports the results on the ability of NI and TCI to predict one-year-ahead CFO.
The models are signi?cant at less than 1 percent. The adjusted R
2
is 0.495 when NI is the
independent variable and 0.483 when TCI is the independent variable. The Vuong (1989)
z-statistic, however, indicates that the difference in adjusted R
2
is not signi?cant.
Further, the results in Panel C are sensitive to which de?ator is used. When CFO, NI and
TCI are de?ated by total assets, model (3) is not signi?cant at 10 percent but model (4) is
signi?cant at less than 10 percent, although TCI in model (4) is not signi?cant at 10
percent. Again, when CFO, NI and TCI are de?ated by market capitalisation, the
predictive ability of TCI is slightly higher than that of NI. Further, contrary to results of
prior studies (Barth et al. 2001a, b; Goncharov and Hodgson, 2008), the coef?cients of NI
and TCI are negative and signi?cant. Thus, overall the results do not support H3.
Panel D reports the results on the ability of NI and TCI to predict one-year-ahead NI.
The models are signi?cant at 1 percent. The adjusted R
2
is 0.166 when NI is the
independent variable and 0.131 when TCI is the independent variable. The Vuong
(1989) z-statistic is not, however, signi?cant at the 10 percent level. Further,
untabulated results show that the results are sensitive to de?ators. When models (5)
and (6) are re-estimated after de?ating the dependent and independent variables by
total assets, NI in model (5) is signi?cant at less than 1 percent and the adjusted R
2
is
0.908. TCI in model (6) is signi?cant at less than 1 percent and the adjusted R
2
is 0.962.
When market value is used as the de?ator, the predictive ability of TCI is slightly
higher than that of NI. Thus, overall the results do not support H4.
Panel E shows the association of NI and TCI with stock returns. Both models are
signi?cant at the less than 1 percent level. The earnings response coef?cients (0.586 for
NIC
t
_P
t21
and 0.464 for TCIC
t
_P
t21
) are signi?cant at less than 1 percent. The
coef?cient of TCIC
t
_P
t21
is lower than that of NIC
t
_P
t21
. This is consistent with TCI
incorporating more transitory items than NI (Kothari and Zimmerman, 1995)[16].
Further, the adjusted R
2
is 0.086 when NIC
t
_P
t21
is the independent variable and 0.058
when TCIC
t
_P
t21
is the independent variable. The explanatory power of NIC
t
_P
t21
is
about 48 percent higher than that of TCIC
t
_P
t21
though the Vuong (1989) z-statistic for
Properties of
NI and TCI
281
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the difference in adjusted R
2
is not signi?cant at the 10 percent level. The result of
Vuong’s test is to be interpreted with caution as the sample size here is small and the
Vuong statistic is sensitive to sample size[17]. The results in Panel E are consistent
with Barton et al. (2009) and Goncharov and Hodgson (2008). Barton et al. (2009) found
that TCI is less value relevant in 39 (including NZ) of their 46 sampled countries. The
results are similar when we use raw returns instead of market adjusted returns as the
dependent variable in models (7) and (8).
The results of models (9) and (10) are reported in Table IV. Both models are
signi?cant at less than 1 percent. The coef?cient of NIC
t
(1.130) is slightly less than that
of TCIC
t
(1.133). The adjusted R
2
are, however, similar (0.661 for model (9) vs 0.666 for
model (10))[18]. The Vuong (1989) z-statistic is not signi?cant at conventional levels.
5.3 Reporting location of TCI and its value relevance
Table Vreports the results of running models (11) and (12) testing the impact of reporting
location of TCI on its value relevance. Although model (11) is signi?cant, not one of the
Independent variables
Model: P
t
¼ a þ b
1
BV
t
þ b
2
NIC
t
þ e
t
(9)
Model: P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ e
t
(10)
Constant 0.531 (3.389
* * *
) 0.543 (3.471
* * *
)
BV
t
0.832 (6.513
* * *
) 0.825 (6.446
* * *
)
NIC
t
1.130 (2.169
* *
)
TCIC
t
1.133 (2.187
* *
)
N 81 81
Adjusted R
2
0.661 0.666
F-statistic 79.018
* * *
80.642
* * *
Vuong z-statistic 0.746
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; P
t
¼ stock price at the end of the
?scal year; BV
t
¼ book value of equity per share at the end of the ?scal year; NIC
t
¼ net income to
common per share; TCIC
t
¼ total comprehensive income to common per share; White (1980)
heteroskedasticity-consistent t-statistics are in parentheses
Table IV.
Results of the
price models
Independent
variables
Model: R
t
¼ a þ b
1
(TCIC
t
_P
t21
)
þ b
2
(TCIC
t
_P
t21*
SINGLE) þ e
t
(11)
Model: P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ b
3
(TCIC
t*
SINGLE) þ e
t
(12)
Constant 20.027 (20.668) 0.490 (3.227
* * *
)
TCIC
t
_P
t21
0.306 (0.903)
TCIC
t
_
P
t21*
SINGLE 0.434 (0.889)
BV
t
0.834 (6.518
* * *
)
TCIC
t
1.063 (2.274
* *
)
TCIC
t*
SINGLE 2.377 (1.764
*
)
N 81 81
Adjusted R
2
0.061 0.671
F-statistic 3.577
* *
55.422
* * *
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; single is 1 if TCI is reported in the
same statement in which pro?t/loss is reported, 0 otherwise; All are variables are as de?ned in
Tables III and IV; White (1980) heteroskedasticity-consistent t-statistics are in parentheses
Table V.
Effect of reporting
location of TCI on its
value relevance
ARJ
24,3
282
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
model variables is signi?cant at conventional levels. Model (12) is signi?cant at less than
1 percent and both BV
t
and TCIC
t
are positive and signi?cant. However, TCIC
t*
SINGLE,
the variable of interest, is positive and signi?cant at 10 percent. Thus, while the returns
model provides no evidence that the value relevance of total comprehensive income
depends on the reporting location, the price model provides only weak support for the
hypothesis that the value relevance of total comprehensive income is higher when it is
reported in a single statement. Overall, these results provide little support for the
hypothesis that the value relevance of TCI depends on the reporting location.
5.4 Correction for industry cluster effects
One potential limitation of the regression analyses above is that observations from
each industry cluster might be correlated as a result of an unobserved industry cluster
effect. To assess whether the results reported above are sensitive to industry cluster
effects, the ?xed effects transformations are used to eliminate the unobservable
industry effect (Wooldrige, 2002). In ?xed effects transformation, the industry average
of each model variable is deducted from the relevant variable of each ?rm belonging to
that industry and the regression models are re-estimated using the industry-demeaned
data. The regressions, however, do not contain any intercept as it is eliminated when
the industry average is deducted from the ?rm-level data (Wooldrige, 2002). The
standard errors are corrected for loss of degrees of freedom as one industry average is
calculated and deducted from the ?rm-level data for each industry (Wooldrige, 2002).
This section reports the (untabulated) results of regressions after correcting for any
unobserved clustering effect. When models (1) and (2) are re-estimated using the
industry-demeaned data, the persistence coef?cient of NI is 0.836, while that of TCI is
0.766. Both coef?cients are signi?cant at less than 1 percent. Thus, net income is
potentially more persistent than total comprehensive income.
In predictive ability analysis, the adjusted R
2
of model (3) is 0.491 and that of
model (4) is 0.450. The adjusted R
2
of model (5) is 0.211 and that of model (6) is 0.175.
Thus, net income has potentially greater predictive ability than total comprehensive
income.
In model (7), the coef?cient of NIC
t
_P
t21
is 0.556 and the adjusted R
2
is 0.087.
In model (8) the coef?cient of TCIC
t
_P
t21
is 0.458 and the adjusted R
2
is 0.065. Thus,
the earnings response coef?cient for net income is potentially higher than that for total
comprehensive income and net income is potentially more highly associated with
contemporaneous returns than total comprehensive income. In price models, the
adjusted R
2
s are similar (0.687 for model (9) and 0.698 for model (10)). In the re-estimated
model (11), the coef?cient of TCIC
t
_P
t21*
SINGLE is not signi?cant at 10 percent.
Similarly in model (12), the coef?cient of TCIC
t*
SINGLE is not signi?cant at 10 percent.
5.5 Marginal contribution of OCI
Up to this point, this study has compared NI with TCI as summary performance
measures. This section reports the un-tabulated results of the incremental predictive
ability and value relevance of OCI over and above those of NI. To assess the incremental
contribution of OCI, TCI is decomposed into NI and OCI. Following Biddle et al. (1995),
the incremental contribution of OCI is assessed by the statistical signi?cance of the
coef?cient of OCI.
Properties of
NI and TCI
283
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
When both NI and OCI are used as independent variables to predict one-year-ahead
CFO, NI is positive and signi?cant at less than 1 percent and OCI is positive and
signi?cant at less than 10 percent. The adjusted R
2
is 0.511. When both NI and OCI are
used to predict one-year-ahead NI, NI is positive and signi?cant at less than 1 percent
but OCI is not signi?cant. The adjusted R
2
is 0.156 and the model is signi?cant at less
than 1 percent.
Areturns model is estimatedbyincludingNIC
t
_P
t21
andOCIC
t
_P
t21
as independent
variables and market-adjusted returns as the dependent variable, to see whether
OCIC
t
_P
t21
has any incremental information content. OCIC
t
_P
t21
is other
comprehensive income to common per share de?ated by beginning-of-year share
price and NIC
t
_P
t21
is as de?ned in model (7). NIC
t
_P
t21
is positive and signi?cant at
less than 1 percent but OCIC
t
_P
t21
is not signi?cant at 10 percent. The overall model is
signi?cant at less than 1 percent and the adjusted R
2
is 0.093. A price model is also
estimated in which BV
t
, NIC
t
and OCIC
t
are used as independent variables and
end-of-period price is used as the dependent variable to assess whether OCIC
t
is
incrementally value relevant. OCIC
t
is other comprehensive income to common per
share. BV
t
and NIC
t
are as de?ned in model (9). BV
t
is signi?cant at less than 1 percent
and NIC
t
is signi?cant at less than 5 percent, but OCIC
t
is not signi?cant at 10 percent.
The adjusted R
2
is 0.664. Thus, the results suggest that other comprehensive income is
not incrementally value relevant.
6. Conclusions
This paper examines properties of NI and TCI of listed companies of NZ. Four
properties are examined: persistence, variability, predictive ability, and value
relevance. The authors also examine whether the value relevance of TCI depends on its
reporting location. This paper utilizes data on comprehensive income that NZ listed
companies reported in 2010 under the new requirement to report comprehensive
income under IAS 1. The sample in this study comprises 86 ?rms used to test the
persistence, variability, and predictive ability of NI and TCI, and 81 ?rms used to test
value relevance of NI and TCI.
This study ?nds that NI is potentially more persistent than TCI and potentially
explains contemporaneous stock returns better than TCI. These results are robust to
alternative de?ators and standard errors corrected for industry clusters. It, however,
?nds no signi?cant difference in the variability and predictive ability of NI and TCI.
The authors ?nd little evidence that the value relevance of TCI depends on its
reporting location. Further, the study ?nds that OCI has incremental ability to predict
one-year-ahead CFO, although the incremental ability of OCI to predict one-year-ahead
NI is not statistically signi?cant. The results also indicate that OCI is not incrementally
value relevant.
The literature on comprehensive income is growing (Dhaliwal et al., 1999; O’Hanlon
and Pope, 1999; Lee et al., 2006; Chambers et al., 2007; Goncharov and Hodgson, 2008).
While two studies (Cahan et al., 2000; Barton et al., 2009) have previously investigated
the properties of NI and comprehensive income in NZ, those studies were conducted
prior to the new IFRS regime in NZ. Adoption of IFRS affects ?nancial account
numbers in NZ (Kabir et al., 2010; Stent et al., 2010). Further, IAS 1 changes the
reporting location of comprehensive income in NZ from a statement of changes in
equity to a performance statement. Prior research (Maines and McDaniel, 2000)
ARJ
24,3
284
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
suggests that location of comprehensive income reporting may affect the weight users
attach to comprehensive income. Hence, results of prior research may not be
generalisable to the current IFRS setting in NZ. This study addresses this gap and
further investigates whether the value relevance of TCI depends on its reporting
location, because IAS 1 allows the option of reporting TCI either in a single statement
of comprehensive income or in a two-statement format (IASB, 2009), and the IASB
(2010) proposes the elimination of the option of the two-statement format. The results
suggest that the IASB (2010) proposal to require the reporting of TCI in a single
statement may not enhance the value relevance of TCI.
This study adds current evidence on the properties of NI, TCI and OCI under IFRS
to the international literature on comprehensive income. The results may be of
potential interest to securities analysts who use ?rm fundamentals to value a ?rm and
other users who use earnings in different contractual settings. The results may also be
of potential interest to the IASB.
The main limitation of this study is the small size of the sample. Thus, the results
may lack generalizability to other samples and statistical power. The small sample,
however, re?ects the small number of ?rms listed in NZ. It further re?ects that a the
time of the study, only one year of data are available for all the listed companies in NZ
after the implementation of the new requirement of comprehensive income reporting
under IAS 1. Further, due to the small sample size, the study could not investigate
whether the results vary across industries. Some industries (e.g. ?nancial institutions,
agriculture and ?shing) are markedly different from other industries, as ?rms
belonging to the former industries have to include some fair value gains and losses in
NI. Thus, the results documented here may not be equally applicable to all industries.
Future research may investigate this issue.
Notes
1. OCI incorporates items of income and expense that are not recognised in pro?t or loss under
IFRS (IASB, 2009, para 7). TCI is the change in owners’ equity from non-owner sources
(IASB, 2009, para 7). TCI is the sum of pro?t/loss and OCI (IASB, 2009, para 7).
2. During the pre-IFRS era, NZ entities reported a similar amount, called “total recognized
revenues and expenses” in a separate Statement of Movements in Equity in accordance with
Financial Reporting Standard 2 (Cahan et al., 2000).
3. Cahan et al. (2000) cover the 1992-1997 period, while Barton et al. (2009) cover 1996-2005.
4. NZ allowed adoption of IFRS from 2005 (ASRB, 2004, para 20). Following this option,
some NZ listed companies early adopted IFRS (Kabir et al., 2010).
5. For example, upward revaluation of property, plant and equipment (PPE) and intangibles is
an other comprehensive income item under IAS 1 (IASB, 2009). But US GAAP do not permit
upward revaluation of PPE and intangibles (Bellandi, 2009).
6. Dechow and Schrand (2004) note limitations of the above criteria. First, managed earnings
could be both persistent and predictable. Second, actual cash ?ows can be a noisy metric of
value. Third, fromthe perspective of faithful representation, variability of earnings is not bad
if it re?ects the underlying variation of the operation of the company. However, variability in
earnings may be induced by accounting also. Fourth, using contemporaneous association
between earnings and stock price assumes that the market is ef?cient and stock price quickly
re?ects all publicly available information. Despite these limitations, the accounting literature
supports the use of these attributes as indices of high earnings quality. Prior research shows
Properties of
NI and TCI
285
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that persistence of earnings components is a desirable attribute fromthe perspective of equity
valuation (Lipe, 1986). The conceptual framework for ?nancial reporting issued jointly by the
FASBandthe IASBargues that accountinginformationshouldassist users assess the amount,
timing anduncertainty of future cash ?ows to the entity (FASB, 2010, para OB3). Earnings are
used in a variety of contractual settings and widely used by analysts in ?rm valuation
(Dechow, 1994; Dechow and Schrand, 2004). These make the ability to predict future cash
?ows andnet income desirable attributes of accounting information. Further, the newly issued
conceptual framework focuses on the information needs of investors (FASB, 2010, para OB2
and BC1.16) and stipulates relevance as one of the two fundamental qualitative characteristics
of ?nancial information (FASB, 2010, para QC5). Association of the income metrics with
contemporaneous stock returns is one way to operationalise the relevance criterion to
investors (Barth et al., 2001).
7. In its exposure draft on presentation of items of other comprehensive income issued in
May 2010, the IASB (2010) proposes to change the title of the statement of comprehensive
income to “statement of pro?t or loss and other comprehensive income”. It further proposes,
inter alia, to eliminate the currently available option of reporting TCI in two statements.
Instead it proposes to require the reporting of TCI in a single statement with two sections –
pro?t/loss and OCI. It also proposes to require the disclosure of items of OCI that are recycled
to pro?t/loss separate from items of OCI that are not.
8. The sequence {Y
t
} follows a sub-martingale process if E(Y
tþ1
jY
0
, . . . , Y
t
) $Y
t
for all t. E is
an expectation operator. The random walk model is a sub-class of the sub-martingale
process (Ball and Watts, 1972).
9. Most studies on comprehensive income used estimates of TCI because accounting standards
did not require disclosure of TCI during the study periods.
10. Ohlson (1999) de?nes transitory earnings as possessing three characteristics:
(a) unpredictability, (b) forecast irrelevance, and (c) value irrelevance.
11. Barth et al. (1995, p. 579) note that American bankers made this argument at the time of the
FASB public hearing before promulgating SFAS 115.
12. However, if NI is managed to give it a more persistent look than TCI, investors may not view
persistence as desirable (Barton et al., 2009; Dechow and Schrand, 2004).
13. NZX Deep Archive is a subscription database service of the NZ Stock Exchange.
14. In case of one company, the net income ?gure reported in the income statement is different
from that reported in the earnings per share note. In case of another company, the number of
shares at the beginning of 2010 is not the same as that at the end of the 2009. Hence, these
two companies are excluded from the sample on the ground of lack of reliability of ?gures.
15. Vuong’s (1989) test is used to assess the statistical signi?cance of the difference in the
explanatory powers of two non-nested regression models. Dechow (1994) describes how
the Vuong (1989) statistic is derived. After Dechow (1994), this test has been widely used in
the accounting literature. The authors, however, note that Vuong’s (1989) test is sensitive to
sample size. This limitation is relevant to this study as the sample size is small.
16. Kothari and Zimmerman (1995, p. 178) argue that the earnings response coef?cient will be
smaller when earnings contains transitory items than when it does not. When earnings
contains both a randomwalk component and a zero-mean transitory component, the earnings
response coef?cient will be the weighted average of the coef?cient of the random walk
component of earnings and that of the transitory component.
17. Dhaliwal et al. (1999, Table 2) reported that the Vuong (1989) statistic for the difference in the
adjusted R
2
of 4.20 percent and 3.81 percent is signi?cant; their sample size is 11425.
ARJ
24,3
286
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
18. Dhaliwal et al. (1999) reported an adjusted R
2
of 0.528 and 0.521 for these models, and thus,
the explanatory powers of models (9) and (10) are comparable to those of Dhaliwal et al.
(1999).
References
Accounting Standards Review Board (2004), Release 8 The Role of the Accounting Standards
Review Board and the Nature of Approved Financial Reporting Standards, available at:
www.asrb.co.nz/default.asp?p¼3 (accessed 12 July 2008).
Ball, R. and Watts, R.L. (1972), “Some time series properties of accounting income”, Journal of
Finance, Vol. 27 No. 3, pp. 663-81.
Bamber, L.S., Jiang, J., Petroni, K. and Wang, I.Y. (2010), “Comprehensive income: who’s afraid of
performance reporting?”, The Accounting Review, Vol. 85 No. 1, pp. 97-126.
Barth, M.E. (2006), Research, Standard Setting, and Global Financial Reporting, Now Publishers
Inc., Hanover, MA.
Barth, M.E., Beaver, W.H. and Landsman, W.R. (2001), “The relevance of the value relevance
literature for ?nancial accounting standard setting: another view”, Journal of Accounting
& Economics, Vol. 31, pp. 77-104.
Barth, M.E., Cram, D.P. and Nelson, K.K. (2001), “Accruals and the prediction of future cash
?ows”, The Accounting Review, Vol. 76 No. 1, pp. 27-58.
Barth, M.E., Landsman, W.R. and Wahlen, J. (1995), “Fair value accounting: effects on banks’
earnings volatility, regulatory capital, and value of contractual cash ?ows”, Journal of
Banking & Finance, Vol. 19, pp. 577-605.
Barton, J., Hansen, B. and Pownall, G. (2009), “Which performance measures do investors value
the most – and why?”, available at: SSRN: http://ssrn.com/abstract¼1230563 (accessed
22 January 2011).
Bellandi, F. (2009), Accounting for Equity and Other Comprehensive Income, Casa Editrice
Universita` La Sapienza, Roma.
Biddle, G.C., Seow, G.M. and Siegel, A.F. (1995), “Relative versus incremental information
content”, Contemporary Accounting Research, Vol. 12, pp. 1-23.
Black, F. (1993), “Choosing accounting rules”, Accounting Horizons, Vol. 7 No. 4, pp. 1-17.
Bowen, R.M., Burgstahler, D. and Daley, L.A. (1986), “Evidence on the relationships between
earnings and various measures of cash ?ow”, The Accounting Review, Vol. LXI No. 4,
pp. 713-25.
Cahan, S.F., Courtenay, S.M., Gronewoller, P.L. and Upton, D.R. (2000), “Value relevance of
mandated comprehensive income disclosures”, Journal of Business Finance & Accounting,
Vol. 27 Nos 9/10, pp. 1273-301.
Chambers, D., Linsmeier, T.J., Shakespeare, C. and Sougiannis, T. (2007), “An evaluation of SFAS
No. 130 comprehensive income disclosures”, Review of Accounting Studies, Vol. 12,
pp. 557-93.
Cheng, C.S.A., Cheung, J.K. and Gopalakrishnan, V. (1993), “On the usefulness of operating
income, NI and comprehensive income in explaining security returns”, Accounting
& Business Research, Vol. 23, pp. 195-203.
Dechow, P.M. (1994), “Accounting earnings and cash ?ows as measures of ?rm performance: the
role accounting accruals”, Journal of Accounting & Economics, Vol. 18, pp. 3-42.
Dechow, P.M. and Schrand, C.M. (2004), Earnings Quality, Research Foundation for CFA
Institute, Charlottesville.
Properties of
NI and TCI
287
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dechow, P.M., Kothari, S.P. and Watts, R.L. (1998), “The relation between earnings and cash
?ows”, Journal of Accounting & Economics, Vol. 25, pp. 133-68.
Dhaliwal, D., Subramanyam, K.R. and Trezevant, R. (1999), “Is comprehensive income superior
to NI as a measure of ?rm performance?”, Journal of Accounting & Economics, Vol. 26,
pp. 43-67.
Easton, P.D. (1999), “Security returns and the value relevance of accounting data”, Accounting
Horizons, Vol. 13 No. 4, pp. 399-412.
Enria, A., Cappiello, L., Dierick, F., Grittini, S., Haralambous, A., Maddaloni, A., Molitor, P., Pires, F.
and Poloni, P. (2004), “Fair value accounting and ?nancial stability”, European Central Bank
Occasional Paper Series No. 13, available at: www.ecb.int/pub/pdf/scpops/ecbocp13.pdf
(accessed 12 June 2008).
Financial Accounting Standards Board (FASB) (2010), Statement of Financial Accounting
Concepts No. 8, Conceptual Framework for Financial Reporting Chapter 1, The Objective
of General Purpose Financial Reporting, and Chapter 3, Qualitative Characteristics of
Useful Financial information, available at: www.fasb.org/cs/BlobServer?
blobcol¼urldata&blobtable¼MungoBlobs&blobkey¼id&blobwhere¼1175821997186&
blobheader¼application%2Fpdf (accessed 30 January 2011).
Goncharov, I. and Hodgson, A. (2008), “The comprehensive income issue in Europe”, available at:
http://ssrn.com/abstract¼1313134 (accessed 16 November 2010).
Hirst, D.E. and Hopkins, P.E. (1998), “Comprehensive income reporting and analysts’ valuation
judgments”, Journal of Accounting Research, Vol. 36, pp. 47-75.
Hodder, L.D., Hopkins, P.E. and Wahlen, J.M. (2006), “Risk-relevance of fair value income
measures for commercial banks”, The Accounting Review, Vol. 81, pp. 337-75.
International Accounting Standards Board (IASB) (2009), International Accounting Standard 1,
Presentation of Financial Statements.
International Accounting Standards Board (IASB) (2010), Exposure Draft ED/2010/5,
Presentation of items of other comprehensive income – Proposed amendments to IAS 1,
available at: www.ifrs.org/NR/rdonlyres/58DEA5EA-8CFC-45A7-910E-A55F9B7EA04F/
0/ED_OCIMay10.pdf (accessed 12 December 2010).
Isidro, H., O’Hanlon, J. and Young, S. (2004), “Dirty surplus accounting ?ows: international
evidence”, Accounting & Business Research, Vol. 34 No. 4, pp. 383-410.
Kabir, M.H., Laswad, F. and Islam, A. (2010), “Impact of IFRS in New Zealand on accounts and
earnings quality”, Australian Accounting Review, Vol. 20, pp. 343-57.
Kanagaretnam, K., Mathieu, R. and Shehata, M. (2009), “Usefulness of comprehensive income
reporting in Canada”, Journal of Accounting & Public Policy, Vol. 28, pp. 349-65.
Kothari, S.P. and Zimmerman, J.L. (1995), “Price and return models”, Journal of Accounting
& Economics, Vol. 20, pp. 155-92.
Kubota, K., Suda, K. and Takehara, H. (2009), “Information content of other comprehensive
income and net income: evidence from Japanese ?rms”, available at: http://ssrn.com/
abstract¼1393746 (accessed 31 January 2011).
Lee, Y.-J., Petroni, K.R. and Shen, M. (2006), “Cherry picking, disclosure quality, and
comprehensive income reporting choices: the case of property-liability insurers”,
Contemporary Accounting Research, Vol. 23 No. 3, pp. 655-92.
Linsmeier, T.J., Gribble, J., Jennings, R.G., Lang, M.H., Penman, S.H., Petroni, K.R., Shores, D.,
Smith, J.H. and War?eld, T.D. (1997), “An issues paper on comprehensive income”,
Accounting Horizons, Vol. 11 No. 2, pp. 120-6.
ARJ
24,3
288
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Lipe, R.C. (1986), “The information contained in the components of earnings”, Journal of
Accounting Research, Vol. 24, pp. 37-68.
Maines, L.A. and McDaniel, L.S. (2000), “Effects of comprehensive-income characteristics on
nonprofessional investors’ judgments: the role of ?nancial-statement presentation format”,
The Accounting Review, Vol. 75 No. 2, pp. 179-207.
O’Hanlon, J.F. and Pope, P.F. (1999), “The value-relevance of UK dirty surplus accounting ?ows”,
British Accounting Review, Vol. 31, pp. 459-82.
Ohlson, J.A. (1999), “On transitory earnings”, Review of Accounting Studies, Vol. 4, pp. 145-62.
Penman, S.H. and Zhang, X.-J. (2002), “Accounting conservatism, the quality of earnings, and
stock returns”, Accounting Review, Vol. 77 No. 2, pp. 237-64.
Smithson, C., Smith, C. and Wilford, D. (1995), Managing Financial Risk: A Guide to Derivative
Products, Financial Engineering and Value Maximization, Irwin Professional Publishing,
New York, NY.
Stent, W., Bradbury, M. and Hooks, J. (2010), “IFRS in New Zealand: effects on ?nancial
statements and ratios”, Paci?c Accounting Review, Vol. 22 No. 2, pp. 92-107.
Vuong, Q.H. (1989), “Likelihood ratio tests for model selection and non-nested hypotheses”,
Econometrica, Vol. 57, pp. 307-33.
Watts, R.L. and Leftwich, R. (1977), “The time series of annual accounting earnings”, Journal of
Accounting Research, Vol. 15 No. 2, pp. 253-71.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and direct test for
heteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
Wooldrige, J.M. (2002), Econometric Analysis of Cross Section and Panel Data, The MIT Press,
Cambridge, MA.
About the authors
M. Humayun Kabir is currently a Senior Lecturer of Financial Accounting at Auckland
University of Technology. He has published in the Australian Accounting Review and the
Managerial Auditing Journal. His research interests are in ?nancial reporting and corporate
governance. M. Humayun Kabir is the corresponding author and can be contacted at:
[email protected]
Fawzi Laswad is a Professor and Head of School of Accountancy at Massey University.
He has wide research interests in ?nancial reporting and accounting education. He served as a
member of the Accounting Standards Review Board from 2002 to 2010, is a member of NZICA
Admissions Board and the Chair of the Academic Board of Advanced Business Education
Limited.
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
Properties of
NI and TCI
289
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Hong He, Zhijun Lin. 2015. Analyst Following, Information Environment and Value Relevance of
Comprehensive Income: Evidence from China. Asia-Pacific Journal of Financial Studies 44, 688-720.
[CrossRef]
2. Allan Hodgson, Mark Russell. 2014. Comprehending Comprehensive Income. Australian Accounting
Review 24:10.1111/auar.2014.24.issue-2, 100-110. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_990447411.pdf
The purpose of this paper is to investigate the properties of net income (NI) and total
comprehensive income (TCI) of listed companies in New Zealand (NZ). Four properties of TCI and NI
are examined: persistence, variability, predictive ability, and value relevance. Whether the value
relevance of TCI depends on its reporting location is also investigated.
Accounting Research Journal
Properties of net income and total comprehensive income: New Zealand evidence
M. Humayun Kabir Fawzi Laswad
Article information:
To cite this document:
M. Humayun Kabir Fawzi Laswad, (2011),"Properties of net income and total comprehensive income: New
Zealand evidence", Accounting Research J ournal, Vol. 24 Iss 3 pp. 268 - 289
Permanent link to this document:
http://dx.doi.org/10.1108/10309611111187000
Downloaded on: 24 January 2016, At: 21:14 (PT)
References: this document contains references to 45 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1099 times since 2011*
Users who downloaded this article also downloaded:
J acqueline Birt, Greg Shailer, (2011),"Forecasting confidence under segment reporting", Accounting
Research J ournal, Vol. 24 Iss 3 pp. 245-267 http://dx.doi.org/10.1108/10309611111186993
Kevin Clarke, J ack Flanagan, Sharron O'Neill, (2011),"Winning ARC grants: comparing accounting with
other commerce-related disciplines", Accounting Research J ournal, Vol. 24 Iss 3 pp. 213-244 http://
dx.doi.org/10.1108/10309611111186984
Victoria J . Clout, Larelle Chapple, Nilan Gandhi, (2013),"The impact of auditor independence regulations
on established and emerging firms", Accounting Research J ournal, Vol. 26 Iss 2 pp. 88-108 http://
dx.doi.org/10.1108/ARJ -Dec-2011-0045
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Properties of net income
and total comprehensive income:
New Zealand evidence
M. Humayun Kabir
Department of Accounting, Faculty of Business,
Auckland University of Technology, Auckland, New Zealand, and
Fawzi Laswad
School of Accountancy, Massey University, Palmerston North, New Zealand
Abstract
Purpose – The purpose of this paper is to investigate the properties of net income (NI) and total
comprehensive income (TCI) of listed companies in New Zealand (NZ). Four properties of TCI and NI
are examined: persistence, variability, predictive ability, and value relevance. Whether the value
relevance of TCI depends on its reporting location is also investigated.
Design/methodology/approach – A cross-sectional research design is used with data on TCI
reported by NZ listed companies in 2010 under the new disclosure requirement in IAS 1. Ordinary least
squares (OLS) regressions are used with a sample of 86 ?rms to test for persistence, variability, and
predictive ability, and 81 ?rms to test for value relevance of NI and TCI.
Findings – The study ?nds: NI is potentially more persistent than TCI and potentially explains
contemporaneous stock returns better than TCI; no signi?cant difference in the variability and
predictive ability of NI and TCI; little evidence that the value relevance of TCI depends on its reporting
location; other comprehensive income (OCI) has incremental ability to predict one-year-ahead CFO,
although the incremental ability of OCI to predict one-year-ahead NI is not statistically signi?cant;
and OCI is not incrementally value relevant.
Practical implications – The ?ndings would be of interest to securities analysts and other users in
valuing ?rms and when earnings are used in contractual settings (e.g. management compensation).
Further, the results would also be of potential interest to standard-setters.
Originality/value – The literature on comprehensive income is growing. However, the authors are
not aware of any study that investigates the properties of NI and TCI in accordance with the new
requirement to report comprehensive income in the amended IAS 1 which came into effect in NZ on
January 1, 2009. The paper adds current evidence on the properties of NI and TCI under IFRS to the
international literature.
Keywords New Zealand, Earnings, Income, Listed companies, Disclosure, Persistence, Variability,
Predictive ability, Value relevance, Comprehensive income
Paper type Research paper
Introduction
International Accounting Standard (IAS) 1 “Presentation of Financial Statements”
requires entities to report other comprehensive income (OCI) and total comprehensive
income (TCI) in a statement of comprehensive income for periods starting on or after
January 1, 2009 (IASB, 2009, para 10)[1]. IAS 1 allows two reporting formats:
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors wish to thank the two anonymous reviewers for their helpful comments on an earlier
version of this paper.
ARJ
24,3
268
Accounting Research Journal
Vol. 24 No. 3, 2011
pp. 268-289
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111187000
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(1) a single statement of comprehensive income including all revenues and
expenses; or
(2) two statements, a separate income statement and a statement of comprehensive
income (IASB, 2009, para 81).
Most New Zealand (NZ) entities reported comprehensive income during 2010 for the
?rst time in accordance with IAS 1[2].
This study examines the properties of net income (NI) and TCI of listed companies
in NZ. There is a growing body of literature on the properties of comprehensive income
(Dhaliwal et al., 1999; O’Hanlon and Pope, 1999; Cahan et al., 2000; Chambers et al.,
2007; Barton et al., 2009). Most of these studies employed estimates of TCI. Further,
studies investigating properties of NI and TCI under IFRS are sparse. Two studies
(Cahan et al., 2000; Barton et al., 2009) that examine the properties of measures of ?rm
performance in NZ relate to previous NZ GAAP before the adoption of International
Financial Reporting Standards (IFRS) in NZ[3].
There have been two signi?cant changes in NZ that are relevant to this study
subsequent to the period covered in Cahan et al. (2000) and Barton et al. (2009). First, NZ
?rms started preparing their ?nancial statements in accordance with IFRS in 2007[4],
and adoption of IFRS impacted on the ?nancial statement amounts of listed companies
in NZ (Kabir et al., 2010; Stent et al., 2010). Second, the requirement to report TCI in a
performance statement came into effect in NZ for the ?rst time for periods starting on
or after January 1, 2009. Under pre-IFRS NZ GAAP, components of comprehensive
income were reported in a statement of changes in equity (Cahan et al., 2000).
Research suggests that the disclosure of comprehensive income may matter to users
depending on the reporting location (Hirst and Hopkins, 1998; Maines and McDaniel,
2000; Chambers et al., 2007). Further, the relative properties of NI and TCI may vary
under different sets of GAAP (e.g. US GAAP, IFRS) as the components of NI and TCI
may vary under each set[5]. Hence, the results of prior studies on properties of
comprehensive income may not be generalized to the current IFRS setting in NZ.
This study utilizes data on comprehensive income that were reported in 2010 in
accordance with the requirements of IAS 1.
Dechow and Schrand (2004) de?ne earnings as of higher quality if they are:
.
more persistent and less volatile;
.
more strongly associated with future cash ?ows; and
.
more strongly associated with contemporaneous stock price performance or
market value[6].
Following Dechow and Schrand (2004), this paper investigates four properties of NI
and TCI:
(1) persistence;
(2) cross-sectional variation in the income metrics;
(3) ability to predict one-year-ahead cash ?ows from operating activities (CFO) and
NI; and
(4) associations with contemporaneous stock returns.
Properties of
NI and TCI
269
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This paper also examines whether the value relevance of TCI depends on where it is
disclosed – a single statement or two-statement format as permitted under IAS 1.
Investigating the properties of NI and comprehensive income is important because
investors may want to know which metric, NI or TCI, measures ?rm performance
better (Black, 1993). Prior research indicates that earnings are used in a variety of
situations such as bonus contracts, debt covenants, and ?rm valuation (Dechow, 1994).
Since earnings are widely used by analysts in ?rm valuation (Dechow and Schrand,
2004), knowledge of the properties of NI and TCI would help analysts place appropriate
emphasis on the summary income metrics when valuing ?rms.
Investigating the properties of NI and TCI is also important from another
perspective. Given the increasing use of fair value in accounting standards, there are
concerns about the properties of income under fair value accounting (Barth, 2006).
In particular, Enria et al. (2004) argued that income under fair value accounting is more
volatile than under the historical cost method. Since TCI incorporates all the realised
and unrealized gains and losses recognised under IFRS, comparing its properties with
those of NI would provide insights into the properties of income under fair value
accounting. Investigating whether the value relevance of TCI depends on its reporting
location is important as IAS 1 allows the option to report TCI in a single statement or
two-statement format, and the IASB (2010) recently issued an exposure draft proposing
the elimination of the two-statement format of reporting TCI[7].
The sample comprises 86 ?rms used to test the persistence, variability, and
predictive ability of NI and TCI, and 81 ?rms used to test value relevance of NI and
TCI, drawing on 2010 data. This study ?nds that NI is potentially more persistent than
is TCI and potentially explains contemporaneous stock returns better than TCI. These
results are robust to alternative de?ators and standard errors corrected for industry
clusters. However, the study does not ?nd any signi?cant difference in the variability
and predictive ability of NI and TCI and little evidence that the value relevance of TCI
depends on its reporting location. Further, OCI is found to have incremental ability
to predict one-year-ahead CFO, although the incremental ability of OCI in predicting
one-year-ahead NI is not statistically signi?cant. The results also indicate that OCI
is not incrementally value relevant.
This study makes two contributions to the literature on the properties of NI and
TCI. First, it provides evidence on the properties of NI and TCI under IFRS. Evidence
on the properties of NI and TCI under IFRS is sparse as the IFRS requirement to report
TCI is new. The earlier evidence on comprehensive income relates to pre-IFRS NZ
GAAP and other local GAAP. Thus, this study contributes current evidence on
comprehensive income under a new reporting regime (i.e. IAS 1). Second, the study
provides evidence on whether the value relevance of TCI depends on the reporting
location of TCI. This is of potential interest to standard-setters as the IASB has
proposed the removal of the option of reporting TCI in two statements. The remainder
of the paper is structured as follows. Section 2 provides a brief review of the literature,
Section 3 develops the hypotheses, and Section 4 discusses the research design and
methodology. Section 5 presents the ?ndings and Section 6 concludes the paper.
2. Literature review
There is an extensive literature on the properties of ?rm performance measures.
Early studies examined the time series properties, persistence, predictive ability
ARJ
24,3
270
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and value relevance of earnings and cash ?ows. Ball and Watts (1972) found that
annual accounting income follows a sub-martingale process, and Watts and Leftwich
(1977) provided evidence suggesting that annual income follows a random walk
process[8]. Lipe’s (1986) ?ndings suggest that different components of earnings vary in
persistence, while Bowen et al. (1986) found that current period earnings are not better
predictors of future cash ?ows than current period cash ?ows. On the other hand,
?ndings of Dechow et al. (1998) indicate that current earnings forecast future operating
cash ?ows better than current operating cash ?ows. Barth et al. (2001a, b) found that
cash ?ows and accrual components of current earnings have better predictive ability
for future cash ?ows than do several lags of earnings. Dechow’s (1994) results suggest
that the association between earnings and contemporaneous stock returns is stronger
than that between cash ?ows and contemporaneous stock returns when:
.
the performance measurement interval is short;
.
the absolute magnitude of accruals is large; and
.
the operating cycle is long.
The literature on comprehensive income is growing. However, while most studies have
used estimates of TCI, Chambers et al. (2007) document that estimates of TCI are
subject to measurement errors[9]. Further, the ?ndings of studies comparing the
properties of NI and TCI are not consistent. One set of studies ?nds that NI is superior to
TCI in terms of value relevance, predictive ability and conservatism. Cheng et al. (1993)
found that operating income and NI explain security returns better than comprehensive
income. O’Hanlon and Pope (1999) investigated the value relevance of dirty surplus
accounting ?ows under UK GAAP over the period 1972-92 and document that ordinary
pro?t and extraordinary items are associated with stock returns, but found no evidence
that dirty surplus ?ows are value relevant. Using estimates of OCI and a sample of
?rms from 16 European countries during 1991-2005, Goncharov and Hodgson (2008)
found that NI is better than TCI in terms of value relevance and ability to predict future
cash ?ows from operations. Further, TCI is less conservative than NI in that TCI
recognises good news on a timelier basis than NI. Barton et al. (2009) examined the
value relevance of eight measures of ?rm performance in 46 countries during 1996-2005
and report that value relevance peaks for measures above the line and no individual
measure of ?rm performance dominates other measures in all countries.
In contrast, a second set of studies found no evidence that TCI is more value
relevant than NI. Using estimates of comprehensive income, Dhaliwal et al. (1999) did
not ?nd any evidence that comprehensive income is more strongly associated with
contemporaneous returns, future cash ?ows and future NI than NI during 1994-1995.
They, however, found that marketable securities adjustments reported by ?nancial
companies are value relevant. Cahan et al. (2000) examined the value relevance of
comprehensive income in NZ during 1992-1997 and do not ?nd evidence that the
incremental value relevance of TCI relative to NI increased after the issuance of
Financial Reporting Standard 2 in 1994. They further report that there is no evidence
that individual components of OCI are incrementally value relevant over and above
TCI. Isidro et al. (2004) document signi?cant cross-country variation in dirty surplus
?ows during 1993-2001, but found little evidence that omission of dirty surplus ?ows
from residual income value estimates would have caused signi?cant valuation errors.
Properties of
NI and TCI
271
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A third set of studies found that TCI and OCI are value relevant. Kubota et al. (2009)
found that dirty surplus items have information content. Chambers et al. (2007) used
as-reportedmeasures of OCI andfound that OCI is pricedbythe market inthe post-SFAS
130 period. Their ?ndings also suggest that foreign currency translation adjustments
and unrealized gains and losses on available-for-sale securities were positively priced.
Usinga sample of cross-listed Canadian ?rms, Kanagaretnamet al. (2009) study ?ndings
suggest that components of other comprehensive income are associated with stock
returns, and that aggregate comprehensive income is more strongly associated with
price and stock returns than is net income. However, their results suggest that net
income is a better predictor of future net income, comprehensive income and cash ?ows
from operations than is comprehensive income. Overall, evidence on the relative value
relevance of NI and TCI is mixed.
A fourth set of studies examined whether the reporting location of TCI (i.e. in a
performance statement or in a statement of stockholders’ equity) matters to users.
Hirst and Hopkins (1998) employed an experimental research design and found that
the reporting of comprehensive income in the income statement helps analysts detect
earnings management. In an experiment involving MBA students, Maines and
McDaniel (2000) report that participants attached signi?cant weight to unrealised gains
when they appeared in the statement of comprehensive income rather than when they
appeared in a statement of stockholders’ equity. However, the ?ndings of Chambers et al.
(2007) suggest that investors pay greater attention to OCI when they are reported in the
statement of changes in equity rather than when they are reported in a performance
statement. Evidence on which format of reporting TCI is desirable to users is mixed is
therefore inconclusive.
Several studies have examined managers’ choice of reporting location of TCI. The
Lee et al. (2006) ?ndings indicated that insurers with a tendency to manage income
through realised securities’ gains and losses, and with a reputation for poor ?nancial
reporting quality are more likely to report comprehensive income in a statement of equity.
Similarly, Bamber et al. (2010) found that US managers with strong equity incentives and
less job security are less likely to report comprehensive income in a performance
statement.
While the empirical literature on comprehensive income is increasing, studies
examining properties of TCI under IFRS are scarce. Two studies (Cahan et al., 2000;
Barton et al., 2009) that examined comprehensive income in NZ used data from the
pre-IFRS era. The adoption of IFRS in NZ impacted on income and other ?nancial
statement numbers of listed companies in NZ (Kabir et al., 2010, Stent et al., 2010).
Further, in contrast to pre-IFRS NZ GAAP under which total recognised revenue and
expenses were reported in a statement of changes in equity, IAS 1 requires
comprehensive income to be reported in a performance statement. Prior research
suggests that the reporting location of TCI matters to users (Hirst and Hopkins, 1998;
Maines and McDaniel, 2000; Chambers et al., 2007). Further, the components of NI and
TCI are not the same under different sets of GAAP and hence the relative properties of
NI and TCI may vary depending on GAAP. Thus, the results of earlier studies on
properties of TCI may not be generalisable to the current IFRS setting in NZ. This
paper seeks to ?ll this gap by examining properties of NI and TCI in NZ as required by
the recently amended IAS 1.
ARJ
24,3
272
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
3. Hypotheses
3.1 Persistence of NI and TCI
Total comprehensive income (TCI) is the sum of bet income (NI) and other
comprehensive income (OCI). The components of OCI include:
.
changes in revaluation surplus;
.
actuarial gains and losses on de?ned bene?t plans;
.
gains and losses arising from translating ?nancial statements of a foreign
operation;
.
gains and losses on re-measuring available-for-sale ?nancial assets; and
.
the effective portion of gains and losses on hedging instruments in a cash ?ow
hedge (IASB, 2009, para 7).
These components arise from changes in interest rates and exchange rates and other
random walk processes (Smithson et al., 1995, cited in Chambers et al., 2007). Thus, OCI
is transitory[10] in nature (Chambers et al., 2007), and NI is likely to be more persistent
than TCI. Therefore, the ?rst hypothesis is:
H1. NI is more persistent than TCI.
3.2 Variability of NI and TCI
One argument against fair value accounting is that it increases the volatility in
?nancial statements[11]. Barth et al. (1995) found that fair value-based earnings are
more volatile than historical cost-based earnings. Similarly, Hodder et al. (2006) found
that the volatility of full fair value income is more than three times that of
comprehensive income and more than ?ve times that of net income. Since TCI
incorporates more value changes than does NI, based on the evidence, NI is likely to be
less volatile than TCI. Thus, the second hypothesis is:
H2. The cross-sectional variation of NI is less than that of TCI.
3.3 Predictive ability of NI and TCI
Net income under IFRS includes some gains and losses from changes in fair value, and
accruals from the application of the revenue recognition and matching principles. TCI
includes further accruals that re?ect transitory revaluations of assets and liabilities.
Accruals that arise from the application of the revenue recognition and matching
principles inthe NI better predict future cash?ows andnet income thando accruals arising
from transitory valuation changes in TCI (Barth et al. 2001a, b; Dechow and Schrand,
2004). Hence, NI is a better predictor of future cash ?ows andnet income thanTCI (Dechow
and Schrand, 2004; Kanagaretnam et al., 2009). The third and fourth hypotheses are:
H3. NI predicts one-year-ahead CFO better than TCI.
H4. NI predicts one-year-ahead NI better than TCI.
3.4 Value relevance of NI and TCI
If NI is more persistent than TCI and predicts one-year-ahead CFO and NI better than
does TCI, investors may view NI more favourably than TCI. Prior research suggests
that investors view the persistence of earnings components as desirable[12], and that
Properties of
NI and TCI
273
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the information content of earnings components is increasing with the components’
persistence (Lipe, 1986; Penman and Zhang, 2002). Further, since TCI includes more
transitory items than does NI, the earnings response coef?cient of TCI is likely to be
less than that of NI (Kothari and Zimmerman, 1995).
On the other hand, it can be argued that comprehensive income is consistent with
accounting-based valuation theory that expresses equity price as the sum of current
book value and present value of future comprehensive residual income (Linsmeier et al.,
1997). Linsmeier et al. (1997, p. 122) conclude: “Thus, for reported income to be most
useful in equity price valuation, it must be comprehensive”. However, the value
relevance of comprehensive income depends on investors’ ability to process
multi-component comprehensive income (Goncharov and Hodgson, 2008). Hence,
given the above discussion and the mixed empirical evidence (Cheng et al., 1993;
Dhaliwal et al., 1999; O’Hanlon and Pope, 1999; Chambers et al., 2007; Goncharov and
Hodgson, 2008; Kanagaretnamet al., 2009) on the relative value relevance of NI and TCI,
the study does not predict the relative value relevance of NI and TCI. This is expressed in
the following non-directional hypothesis:
H5. The value relevance of NI differs from TCI.
3.5 Reporting location of TCI and its value relevance
IAS 1 allows the reporting of TCI in a statement of comprehensive income or in two
statements (a separate income statement and a statement of comprehensive income)
(IASB, 2009). The ef?cient market hypothesis appears to suggest that reporting
location is irrelevant as long as the information is publicly disclosed. However,
the proposal of the IASB (2010) to eliminate the current option of reporting TCI in a
two-statement format suggests that the reporting location matters. Further,
prior research suggests that the reporting location of TCI is potentially important as
it affects the perceptions of the importance of OCI and TCI disclosed (Maines and
McDaniel, 2000).
The evidence regarding the preferred reporting format of TCI is mixed. Hirst and
Hopkins (1998) ?nd that reporting comprehensive income in the income statement
helps analysts detect earnings management. Maines and McDaniel (2000) ?nd that
participants in an experiment attach more weight to unrealised gains when they
appear in the statement of comprehensive income than when they are disclosed in a
statement of stockholders’ equity. However, Chambers et al. (2007) found that investors
pay greater attention to OCI when they are reported in the statement of changes in
equity rather than when they are reported in a performance statement. Thus, given the
mixed evidence, this study makes no prediction on the relative value relevance of TCI
when it is disclosed in a single statement rather than when it is reported in a
two-statement format. The hypothesis is as follows:
H6. The value relevance of TCI differs depending on its reporting location.
4. Methodology
4.1 Data and sample
The authors hand-collected ?nancial statement data from the 2010 annual reports of
listed companies from NZX Deep Archive[13] and returns data from Datastream.
They started with ?rms that have 2010 annual reports on NZX Deep Archive.
ARJ
24,3
274
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
A total of 145 NZ listed ?rms had 2010 annual reports at the time of data collection.
From this, they excluded ?rms:
.
whose NI and TCI are the same (i.e. no OCI items);
.
whose ?nancial statements are presented in foreign currency;
.
whose ?nancial statement ?gures do not match those in the corresponding
notes[14];
.
whose ?nancial statements are not consolidated;
.
whose ?scal years in 2009 and 2010 are not equal in length;
.
which are overseas companies; and
.
which prepared their ?nancial statements using foreign GAAP.
This process yielded a sample of 86 ?rms. We use these 86 ?rms for analyses of
persistence, variability, and predictive ability of NI and TCI. Five of the sample ?rms
do not have returns data. Hence, the authors utilize 81 ?rms for tests of association of
NI and TCI with contemporaneous stock returns. Table I summarises the sample
selection and industry distribution.
Panel A of Table I shows that TCI and NI is the same for 39 out of 145 ?rms. Hence,
for 73.10 percent of the population TCI differs from NI. This percentage is similar to
the percentage (71 percent) reported by Dhaliwal et al. (1999). Panel B shows the
sample ?rms come from a wide variety of industries, with the largest number being
from the consumer industry.
4.2 Models
Persistence. The following two standard models are used to assess the persistence of NI
and TCI (Dechow and Schrand, 2004). The closer b is to 1, the more persistent the
variable is (Dechow and Schrand, 2004):
NI
tþ1
¼ a þ bNI
t
þ 1
t
ð1Þ
TCI
tþ1
¼ a þ bTCI
t
þ 1
t
ð2Þ
where NI is net income and TCI is total comprehensive income. Following Dechow
(1994) and Barth et al. (1995), both NI and TCI are de?ated by the weighted average
number of shares to mitigate the effects of heteroskedasticity. Since 2010 ?nancial
statements report comparative ?nancial statement amounts for 2009, this paper
employs 2009 NI and TCI to predict 2010 NI and TCI, respectively.
Variability. Following Barth et al. (1995), this study assesses variability by standard
deviation and compares the standard deviation of NI with that of TCI in 2010.
Predictive ability. Models (3) and (4) are used to assess which income metric – NI and
TCI – can predict one-year-ahead CFO better. Following prior literature (Dechow,
1994), the model having a higher adjusted R
2
will be considered to have better
predictive ability. Vuong’s (1989) test is used to ?nd the signi?cance of the difference in
adjusted R
2
of the two models[15]. Similarly, models (5) and (6) are used to evaluate the
relative ability of NI and TCI to predict one-year-ahead NI.
CFO is used because the recent conceptual framework jointly issued by the FASB
and the International Accounting Standards Board (IASB) states that ?nancial
reporting should assist users assess the prospects for future cash ?ows to the entity
Properties of
NI and TCI
275
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(FASB, 2010, para. OB 3 & 4). Further, the newly issued framework notes that
information about the ?rm’s ?nancial performance during a period is helpful in
assessing the entity’s ability to generate future cash ?ows (FASB, 2010, para. OB 18).
The authors also use NI as the variable to be predicted because income is widely used
by analysts in ?rm valuation (Dechow and Schrand, 2004):
CFO
tþ1
¼ a þ bNI
t
þ 1
t
ð3Þ
CFO
tþ1
¼ a þ bTCI
t
þ 1
t
ð4Þ
NI
tþ1
¼ a þ bNI
t
þ 1
t
ð5Þ
NI
tþ1
¼ a þ bTCI
t
þ 1
t
ð6Þ
Panel A: derivation of sample
Companies on NZX Deep Archive 145
Less: NI and TCI are the same 39
Financial statements are presented in
foreign currency
3
Figures in notes do not match corresponding
?gures in ?nancial statements
2
Non-consolidated statements 1
Different lengths of periods in 2009 and 2010 4
Overseas companies 9
Companies that prepared their ?nancial
statements using foreign GAAP
1
Total number of companies
excluded from the sample
59
Sample for tests of persistence,
variability, and predictive ability
86
Less: Missing returns data 5
Sample for tests of association with
stock returns
81
Panel B: industry distribution of sample ?rms
Industry No. of ?rms
Agriculture and ?shing 8
Building materials and construction 3
Consumer 12
Energy processing 7
Finance and other services 10
Food and beverages 6
Intermediate and durables 9
Investment 7
Leisure and tourism 4
Media and telecommunications 4
Mining 2
Ports 5
Property 4
Textiles and apparel 2
Transport 3
Total 86
Table I.
Sample
ARJ
24,3
276
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
where CFO is cash ?ows from operating activities, NI is net income and TCI is total
comprehensive income. All three variables are de?ated by the weighted average
number of shares. Since 2010 ?nancial statements report comparative ?nancial
statement amounts for 2009 also, this paper uses 2009 NI and TCI to predict 2010 CFO
and NI.
Value relevance of NI and TCI. Two models are used widely in the capital
market-based accounting research: the returns model and the price model (Kothari and
Zimmerman, 1995). Since the focus of this study is on assessing which income metric – NI
or TCI – is a better measure of ?rm performance, this paper follows Dechow (1994) and
Easton (1999), and employs the returns model. Kothari and Zimmerman (1995) argue and
document that the earnings response coef?cient inthe returns model is biased toward zero
because of the “price leads earnings” phenomenon and the consequent omitted variable
problem. However, as Dechow (1994) and Easton (1999) argue convincingly, when
the research objective is to assess earnings as a summary measure of ?rm performance,
the effect of this omitted variable (i.e. events not summarized in earnings but in price)
becomes the focus of investigation. Further, the price model suffers from
heteroskedasticity and a scaling problem (Kothari and Zimmerman, 1995; Easton,
1999). Thus, Easton (1999, p. 411) concludes: “[. . .] the inferences from returns models are
probablymore reliable andshould be used.” The following models are usedto evaluate the
relative ability of NI and TCI to summarise ?rmperformance as re?ected in stock returns:
R
t
¼ a þ bðNIC
t
_P
t21
Þ þ 1
t
ð7Þ
R
t
¼ a þ bðTCIC
t
_P
t21
Þ þ 1
t
ð8Þ
where R
t
is the stock returns over the ?scal year minus returns on the NZX market
portfolio, NIC
t
_P
t21
is net income to common per share de?ated by beginning-of-year
price, and TCIC
t
_P
t21
is total comprehensive income to common per share de?ated by
beginning-of-year stock price. Following Dechow (1994) and Goncharov and Hodgson
(2008), in models (7)-(8), this paper uses ?nancial statement variables on a per share basis,
de?ated by beginning-of-period price. Following Dechow (1994), market-wide returns are
deducted from stock returns because they have low association with realized cash ?ows
and earnings, and this improves the power of the tests. Following prior literature (Dechow,
1994), the model having a higher adjusted R
2
will be taken to explain stock returns better,
and Vuong’s (1989) test is used to test the signi?cance of the difference in the adjusted R
2
of the two models.
Kothari and Zimmerman (1995) document that both the returns and price models
suffer from econometric and theoretical problems and hence suggest that researchers
use both models for de?nitive conclusions. Hence, following Dhaliwal et al. (1999), this
paper also assesses the relative value relevance of NI and TCI by running a price level
model. However, as Kothari and Zimmerman (1995) note, the price model does not
measure the information arrival during the period. Following Dhaliwal et al. (1999),
the paper runs the following two models:
P
t
¼ a þ b
1
BV
t
þ b
2
NIC
t
þ 1
t
ð9Þ
P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ 1
t
ð10Þ
where P
t
is price per share at the end of the ?scal year, BV
t
is book value of equity at
the end of the ?scal year, NIC
t
is net income available to the common shareholders,
Properties of
NI and TCI
277
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and TCIC
t
is total comprehensive income available to the common shareholders. The
?nancial statement variables are de?ated by the weighted average number of shares
during the ?scal year.
Reporting location of TCI and its value relevance. The following model is used to
assess whether the value relevance of TCI depends on where it is reported:
R
t
¼ a þ b
1
ðTCIC
t_
P
t21
Þ þ b
2
TCIC
t_
P
t21*
SINGLE þ 1
t
ð11Þ
where SINGLE is a binary variable that takes 1 if TCI is reported in the same statement
in which pro?t/loss is reported, 0 otherwise. All other variables are as de?ned in
model (8). A positive and signi?cant coef?cient of b
2
would be consistent with an
incremental value relevance of TCI when it is reported in the same statement in which
pro?t/loss is reported, rather than when a two-statement format is used for reporting
comprehensive income.
Given the limitations of the returns model, this paper also uses the following price
model to assess the impact of reporting location of TCI on its value relevance:
P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ b
3
TCIC
t *
SINGLE þ 1
t
ð12Þ
All variables are as de?ned in models (10) and (11). A positive and signi?cant
coef?cient of b
3
would be consistent with an incremental value relevance of TCI when
it is reported in a single statement of comprehensive income, rather than when it is
reported in a two-statement format.
5. Findings
5.1 Descriptive statistics and correlation matrices
Table II reports descriptive statistics of model variables and correlation coef?cients
between them. In Panel A the mean (median) NI per share (NI
tþ1
) and TCI per share
(TCI
tþ1
) in 2010 are $0.118 ($0.066) and $0.117 ($0.057), respectively, and are
statistically signi?cantly different from zero. The mean (median) OCI (not tabulated) in
2010 is 20.001 (0.000), which is not signi?cantly different from zero. During 2010, the
three most frequent items of OCI were cash ?ow hedge (50 ?rms), exchange difference
on translating foreign operations (45 ?rms), and changes in property, plant and
equipment revaluation reserve (18 ?rms). During 2010, 33 out of 86 ?rms reported TCI
in a single statement of comprehensive income.
Panels B and C of Table II report the correlation coef?cients between the model
variables. The correlation between net income and total comprehensive income is high
and positive. For example, the correlation coef?cients of NI
t
with TCI
t
is 0.894 and
NI
tþ1
with TCI
tþ1
is 0.991 in Panel B. Similarly, the correlation coef?cient of
NIC
t
_P
t21
with TCIC
t
_P
t21
is 0.963 and that of NIC
t
with TCIC
t
is 0.990. The high
correlations along with similar average magnitudes of net income and total
comprehensive income reported in Panel A may introduce bias against ?nding any
differences in the properties of net income and total comprehensive income.
The issue of multicollinearity does not arise in models (1)-(8) as these models employ
bivariate regressions. The highest correlation coef?cient between the variables in
models (9)-(12) is 0.439, which is between BV
t
and TCIC
t
. Thus, multicollinearity poses
no serious problem in the regressions in this study.
ARJ
24,3
278
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5.2 Properties of NI and TCI
Table III reports the results on the properties of NI and TCI. Panel A reports the results
on the persistence of NI and TCI. The adjusted R
2
is 0.166 and 0.150 for model (1) and (2),
respectively. Both models are signi?cant at less than 1 percent. The persistence
Variables N Mean Median SD
Panel A: descriptive statistics
NI
tþ1
86 0.118
* * *
0.066
* * *
0.412
NI
t
86 0.081
* * *
0.065
* * *
0.229
TCI
tþ1
86 0.117
* *
0.057
* * *
0.429
TCI
t
86 0.090
* * *
0.070
* * *
0.261
CFO
tþ1
86 0.239
* * *
0.143
* * *
0.281
R
t
81 20.30 20.017 0.381
NIC
t
_P
t21
81 0.001 0.047
* * *
0.203
TCIC
t
_
P
t21
81 20.004 0.049
* * *
0.217
P
t
81 2.030
* * *
1.520
* * *
1.928
BV
t
81 1.636
* * *
1.089
* * *
1.583
NIC
t
81 0.122
* * *
0.068
* * *
0.414
TCIC
t
81 0.121
* *
0.055
* * *
0.432
SINGLE 81 0.370
* * *
0.000
* * *
0.486
Panel B: correlation matrix for variables in models (1)-(6)
NI
tþ1
NI
t
TCI
tþ1
TCI
t
CFO
tþ1
NI
tþ1
1.000
NI
t
0.419
* * *
1.000
TCI
tþ1
0.991
* * *
0.437
* * *
1.000
TCI
t
0.376
* * *
0.894
* * *
0.399
* * *
1.000
CFO
tþ1
0.524
* * *
0.708
* * *
0.531
* * *
0.699
* * *
1.000
Panel C: correlation matrix for variables in models (7)-(12)
R
t
NIC
t
_P
t21
TCIC
t
_P
t21
P
t
BV
t
NIC
t
TCIC
t
SINGLE
R
t
1.000
NIC
t
_P
t21
0.313
* * *
1.000
TCIC
t
_
P
t21
0.264
* *
0.963
* * *
1.000
P
t
0.197
*
0.372
* * *
0.365
* * *
1.000
BV
t
0.146 0.300
* * *
0.286
* * *
0.789
* * *
1.000
NIC
t
0.104 0.510
* * *
0.495
* * *
0.541
* * *
0.437
* * *
1.000
TCIC
t
0.086 0.500
* * *
0.511
* * *
0.551
* * *
0.439
* * *
0.990
* * *
1.000
SINGLE 20.185
*
20.042 20.014 20.331
* * *
20.333
* * *
20.123 20.121 1.000
Notes: Statistically signi?cant at:
*
10,
* *
5 and
* * *
1 percent; NI
tþ1
¼ one-year-ahead net income per share;
NI
t
¼ current year net income per share; TCI
tþ1
¼ one-year-ahead total comprehensive income per share;
T CI
t
¼ current year total comprehensive income per share; CFO
tþ1
¼ one-year-ahead cash ?ows from
operating activities per share; R
t
¼ returns over the ?scal year minus retun on the NZX portfolio;
NIC
t
_P
t21
¼ current year net income to common per share de?ated by beginning-of-year share price;
TCIC
t
_P
t21
¼ current year total comprehensive income to common per share, de?ated by beginning-of-year
share price; P
t
¼ end-of-year share price; BV
t
¼ book value per share; NIC
t
¼ current year net income to
common per share; TCIC
t
¼ current year total comprehensive income to common per share; and SINGLE ¼ 1 if
the TCI is reported in a single statement of comprehensive income, 0 otherwise; the ?rst ?ve variables in this
table are those in models (1)-(6); the sample size for these models is 86. The last eight variables are those in
models (7)-(12); the sample size for these models is 81 as ?ve out of the 86 ?rms do not have returns data; all
?gures, except the ratios, are in dollars
Table II.
Descriptive statistics
and correlation matrices
Properties of
NI and TCI
279
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Panel A: persistence of NI and TCI
Independent
variables Model: NI
tþ1
¼ a þ bNI
t
þ e
t
(1) TCI
tþ1
¼ a þ bTCI
t
þ e
t
(2)
Constant 0.057 (2.338
* *
) 0.058 (1.901
*
)
NI
t
0.755 (2.340
* *
)
TCI
t
0.656 (3.057
* * *
)
N 86 86
Adjusted R
2
0.166 0.150
F-statistic 17.933
* * *
15.945
* * *
Panel B:
variability
of NI and TCI NI TCI
SD 0.412 0.429
N 86 86
F-statistic 1.087
Panel C: ability of NI and TCI to predict one-year-ahead CFO
Independent
variables Model: CFO
tþ1
¼ a þ bNI
t
þ e
t
(3) Model: CFO
tþ1
¼ a þ bTCI
t
þ e
t
(4)
Constant 0.169 (7.376
* * *
) 0.171 (7.403
* * *
)
NI
t
0.871 .(9.176
* * *
)
TCI
t
0.753 .(8.962
* * *
)
N 86 86
Adjusted R
2
0.495 0.483
F-statistic 84.200
* * *
80.318
* * *
Vuong z-statistic 1.008
Panel D: ability of NI and TCI to predict one-year-ahead NI
Independent
variables Model: NI
tþ1
¼ a þ bNI
t
þ e
t
(5) Model: NI
tþ1
¼ a þ bTCI
t
þ e
t
(6)
Constant 0.057 (2.338
* *
) 0.065 (2.255
* *
)
NI
t
0.755 (2.340
* *
)
TCI
t
0.593 (2.841
* * *
)
N 86 86
Adjusted R
2
0.166 0.131
F-statistic 17.933
* * *
13.840
* * *
Vuong z-statistic 1.341
Panel E: association with contemporaneous returns
Independent
variables Model:R
t
¼ a þ b(NIC
t
_P
t21
) þ e
t
(7) Model: R
t
¼ a þ b(TCIC
t
_P
t21
) þ e
t
(8)
Constant 20.031 (20.757) 20.028 (20.684)
NIC
t
_P
t21
0.586 (2.924
* * *
)
TCIC
t
_P
t21
0.464 (2.436
* *
)
N 81 81
Adjusted R
2
0.086 0.058
F-statistic 8.551
* * *
5.933
* *
Vuong z-statistic 1.117
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; NI
tþ1
¼ one-year-ahead net income per
share; NI
t
¼ current year net income per share; TCI
tþ1
¼ one-year-ahead total comprehensive income per
share; TCI
t
¼ current year total comprehensive income per share; CFO
tþ1
¼ one-year-ahead cash ?ows
from operating activities per share; R
t
¼ company stock returns over the ?scal year minus returns on the
NZX market portfolio; NIC
t
_P
t21
¼ current year net income to common per share de?ated by beginning-
of-year price; TCIC
t
_P
t21
¼ current year total comprehensive income to common per share de?ated by
beginning-of-year price; ?gures in parentheses are t-statistics; t-statistics in Panel A and D are White
(1980) heteroskedasticity-consistent
Table III.
Results on properties
of NI and TCI
ARJ
24,3
280
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
coef?cients of NI and TCI are 0.755 and 0.656, respectively. Thus, NI is potentially more
persistent than TCI. These results are consistent with Barton et al. (2009) who found TCI
is less persistent than NI. Un-tabulated results showthat the results remain similar when
NI and TCI are de?ated by alternative size measures. The persistence coef?cients are
1.312 and 0.801 for NI and TCI, respectively, when both NI and TCI are de?ated by total
assets. The corresponding coef?cients are 1.986 and 1.768, respectively, when de?ated
by market capitalisation at the end of the last ?scal year. These results are consistent
with H1.
Panel B shows the variability of NI and TCI. The standard deviation of NI and TCI is
0.412 and 0.429, respectively. Thus, consistent with H2, the variability of TCI is higher
than that of NI. However, the difference in variability is not signi?cant at the 10 percent
level. Further, untabulated results show that the results are sensitive to de?ators. When
NI and TCI are de?ated by total assets at the end of the year, the standard deviation of NI
(1.758) is higher than that of TCI (1.413) and the difference is signi?cant at less than
5 percent. When NI and TCI are de?ated by market capitalisation at the end of the
previous ?scal year, the standard deviation of NI and TCI is 0.677 and 0.654,
respectively. However, the difference in variability is not signi?cant at 10 percent. Thus,
overall the results do not support H2.
Panel Creports the results on the ability of NI and TCI to predict one-year-ahead CFO.
The models are signi?cant at less than 1 percent. The adjusted R
2
is 0.495 when NI is the
independent variable and 0.483 when TCI is the independent variable. The Vuong (1989)
z-statistic, however, indicates that the difference in adjusted R
2
is not signi?cant.
Further, the results in Panel C are sensitive to which de?ator is used. When CFO, NI and
TCI are de?ated by total assets, model (3) is not signi?cant at 10 percent but model (4) is
signi?cant at less than 10 percent, although TCI in model (4) is not signi?cant at 10
percent. Again, when CFO, NI and TCI are de?ated by market capitalisation, the
predictive ability of TCI is slightly higher than that of NI. Further, contrary to results of
prior studies (Barth et al. 2001a, b; Goncharov and Hodgson, 2008), the coef?cients of NI
and TCI are negative and signi?cant. Thus, overall the results do not support H3.
Panel D reports the results on the ability of NI and TCI to predict one-year-ahead NI.
The models are signi?cant at 1 percent. The adjusted R
2
is 0.166 when NI is the
independent variable and 0.131 when TCI is the independent variable. The Vuong
(1989) z-statistic is not, however, signi?cant at the 10 percent level. Further,
untabulated results show that the results are sensitive to de?ators. When models (5)
and (6) are re-estimated after de?ating the dependent and independent variables by
total assets, NI in model (5) is signi?cant at less than 1 percent and the adjusted R
2
is
0.908. TCI in model (6) is signi?cant at less than 1 percent and the adjusted R
2
is 0.962.
When market value is used as the de?ator, the predictive ability of TCI is slightly
higher than that of NI. Thus, overall the results do not support H4.
Panel E shows the association of NI and TCI with stock returns. Both models are
signi?cant at the less than 1 percent level. The earnings response coef?cients (0.586 for
NIC
t
_P
t21
and 0.464 for TCIC
t
_P
t21
) are signi?cant at less than 1 percent. The
coef?cient of TCIC
t
_P
t21
is lower than that of NIC
t
_P
t21
. This is consistent with TCI
incorporating more transitory items than NI (Kothari and Zimmerman, 1995)[16].
Further, the adjusted R
2
is 0.086 when NIC
t
_P
t21
is the independent variable and 0.058
when TCIC
t
_P
t21
is the independent variable. The explanatory power of NIC
t
_P
t21
is
about 48 percent higher than that of TCIC
t
_P
t21
though the Vuong (1989) z-statistic for
Properties of
NI and TCI
281
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the difference in adjusted R
2
is not signi?cant at the 10 percent level. The result of
Vuong’s test is to be interpreted with caution as the sample size here is small and the
Vuong statistic is sensitive to sample size[17]. The results in Panel E are consistent
with Barton et al. (2009) and Goncharov and Hodgson (2008). Barton et al. (2009) found
that TCI is less value relevant in 39 (including NZ) of their 46 sampled countries. The
results are similar when we use raw returns instead of market adjusted returns as the
dependent variable in models (7) and (8).
The results of models (9) and (10) are reported in Table IV. Both models are
signi?cant at less than 1 percent. The coef?cient of NIC
t
(1.130) is slightly less than that
of TCIC
t
(1.133). The adjusted R
2
are, however, similar (0.661 for model (9) vs 0.666 for
model (10))[18]. The Vuong (1989) z-statistic is not signi?cant at conventional levels.
5.3 Reporting location of TCI and its value relevance
Table Vreports the results of running models (11) and (12) testing the impact of reporting
location of TCI on its value relevance. Although model (11) is signi?cant, not one of the
Independent variables
Model: P
t
¼ a þ b
1
BV
t
þ b
2
NIC
t
þ e
t
(9)
Model: P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ e
t
(10)
Constant 0.531 (3.389
* * *
) 0.543 (3.471
* * *
)
BV
t
0.832 (6.513
* * *
) 0.825 (6.446
* * *
)
NIC
t
1.130 (2.169
* *
)
TCIC
t
1.133 (2.187
* *
)
N 81 81
Adjusted R
2
0.661 0.666
F-statistic 79.018
* * *
80.642
* * *
Vuong z-statistic 0.746
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; P
t
¼ stock price at the end of the
?scal year; BV
t
¼ book value of equity per share at the end of the ?scal year; NIC
t
¼ net income to
common per share; TCIC
t
¼ total comprehensive income to common per share; White (1980)
heteroskedasticity-consistent t-statistics are in parentheses
Table IV.
Results of the
price models
Independent
variables
Model: R
t
¼ a þ b
1
(TCIC
t
_P
t21
)
þ b
2
(TCIC
t
_P
t21*
SINGLE) þ e
t
(11)
Model: P
t
¼ a þ b
1
BV
t
þ b
2
TCIC
t
þ b
3
(TCIC
t*
SINGLE) þ e
t
(12)
Constant 20.027 (20.668) 0.490 (3.227
* * *
)
TCIC
t
_P
t21
0.306 (0.903)
TCIC
t
_
P
t21*
SINGLE 0.434 (0.889)
BV
t
0.834 (6.518
* * *
)
TCIC
t
1.063 (2.274
* *
)
TCIC
t*
SINGLE 2.377 (1.764
*
)
N 81 81
Adjusted R
2
0.061 0.671
F-statistic 3.577
* *
55.422
* * *
Notes: Statistically signi?cant at:
*
10,
* *
5, and
* * *
1 percent; single is 1 if TCI is reported in the
same statement in which pro?t/loss is reported, 0 otherwise; All are variables are as de?ned in
Tables III and IV; White (1980) heteroskedasticity-consistent t-statistics are in parentheses
Table V.
Effect of reporting
location of TCI on its
value relevance
ARJ
24,3
282
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
model variables is signi?cant at conventional levels. Model (12) is signi?cant at less than
1 percent and both BV
t
and TCIC
t
are positive and signi?cant. However, TCIC
t*
SINGLE,
the variable of interest, is positive and signi?cant at 10 percent. Thus, while the returns
model provides no evidence that the value relevance of total comprehensive income
depends on the reporting location, the price model provides only weak support for the
hypothesis that the value relevance of total comprehensive income is higher when it is
reported in a single statement. Overall, these results provide little support for the
hypothesis that the value relevance of TCI depends on the reporting location.
5.4 Correction for industry cluster effects
One potential limitation of the regression analyses above is that observations from
each industry cluster might be correlated as a result of an unobserved industry cluster
effect. To assess whether the results reported above are sensitive to industry cluster
effects, the ?xed effects transformations are used to eliminate the unobservable
industry effect (Wooldrige, 2002). In ?xed effects transformation, the industry average
of each model variable is deducted from the relevant variable of each ?rm belonging to
that industry and the regression models are re-estimated using the industry-demeaned
data. The regressions, however, do not contain any intercept as it is eliminated when
the industry average is deducted from the ?rm-level data (Wooldrige, 2002). The
standard errors are corrected for loss of degrees of freedom as one industry average is
calculated and deducted from the ?rm-level data for each industry (Wooldrige, 2002).
This section reports the (untabulated) results of regressions after correcting for any
unobserved clustering effect. When models (1) and (2) are re-estimated using the
industry-demeaned data, the persistence coef?cient of NI is 0.836, while that of TCI is
0.766. Both coef?cients are signi?cant at less than 1 percent. Thus, net income is
potentially more persistent than total comprehensive income.
In predictive ability analysis, the adjusted R
2
of model (3) is 0.491 and that of
model (4) is 0.450. The adjusted R
2
of model (5) is 0.211 and that of model (6) is 0.175.
Thus, net income has potentially greater predictive ability than total comprehensive
income.
In model (7), the coef?cient of NIC
t
_P
t21
is 0.556 and the adjusted R
2
is 0.087.
In model (8) the coef?cient of TCIC
t
_P
t21
is 0.458 and the adjusted R
2
is 0.065. Thus,
the earnings response coef?cient for net income is potentially higher than that for total
comprehensive income and net income is potentially more highly associated with
contemporaneous returns than total comprehensive income. In price models, the
adjusted R
2
s are similar (0.687 for model (9) and 0.698 for model (10)). In the re-estimated
model (11), the coef?cient of TCIC
t
_P
t21*
SINGLE is not signi?cant at 10 percent.
Similarly in model (12), the coef?cient of TCIC
t*
SINGLE is not signi?cant at 10 percent.
5.5 Marginal contribution of OCI
Up to this point, this study has compared NI with TCI as summary performance
measures. This section reports the un-tabulated results of the incremental predictive
ability and value relevance of OCI over and above those of NI. To assess the incremental
contribution of OCI, TCI is decomposed into NI and OCI. Following Biddle et al. (1995),
the incremental contribution of OCI is assessed by the statistical signi?cance of the
coef?cient of OCI.
Properties of
NI and TCI
283
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
When both NI and OCI are used as independent variables to predict one-year-ahead
CFO, NI is positive and signi?cant at less than 1 percent and OCI is positive and
signi?cant at less than 10 percent. The adjusted R
2
is 0.511. When both NI and OCI are
used to predict one-year-ahead NI, NI is positive and signi?cant at less than 1 percent
but OCI is not signi?cant. The adjusted R
2
is 0.156 and the model is signi?cant at less
than 1 percent.
Areturns model is estimatedbyincludingNIC
t
_P
t21
andOCIC
t
_P
t21
as independent
variables and market-adjusted returns as the dependent variable, to see whether
OCIC
t
_P
t21
has any incremental information content. OCIC
t
_P
t21
is other
comprehensive income to common per share de?ated by beginning-of-year share
price and NIC
t
_P
t21
is as de?ned in model (7). NIC
t
_P
t21
is positive and signi?cant at
less than 1 percent but OCIC
t
_P
t21
is not signi?cant at 10 percent. The overall model is
signi?cant at less than 1 percent and the adjusted R
2
is 0.093. A price model is also
estimated in which BV
t
, NIC
t
and OCIC
t
are used as independent variables and
end-of-period price is used as the dependent variable to assess whether OCIC
t
is
incrementally value relevant. OCIC
t
is other comprehensive income to common per
share. BV
t
and NIC
t
are as de?ned in model (9). BV
t
is signi?cant at less than 1 percent
and NIC
t
is signi?cant at less than 5 percent, but OCIC
t
is not signi?cant at 10 percent.
The adjusted R
2
is 0.664. Thus, the results suggest that other comprehensive income is
not incrementally value relevant.
6. Conclusions
This paper examines properties of NI and TCI of listed companies of NZ. Four
properties are examined: persistence, variability, predictive ability, and value
relevance. The authors also examine whether the value relevance of TCI depends on its
reporting location. This paper utilizes data on comprehensive income that NZ listed
companies reported in 2010 under the new requirement to report comprehensive
income under IAS 1. The sample in this study comprises 86 ?rms used to test the
persistence, variability, and predictive ability of NI and TCI, and 81 ?rms used to test
value relevance of NI and TCI.
This study ?nds that NI is potentially more persistent than TCI and potentially
explains contemporaneous stock returns better than TCI. These results are robust to
alternative de?ators and standard errors corrected for industry clusters. It, however,
?nds no signi?cant difference in the variability and predictive ability of NI and TCI.
The authors ?nd little evidence that the value relevance of TCI depends on its
reporting location. Further, the study ?nds that OCI has incremental ability to predict
one-year-ahead CFO, although the incremental ability of OCI to predict one-year-ahead
NI is not statistically signi?cant. The results also indicate that OCI is not incrementally
value relevant.
The literature on comprehensive income is growing (Dhaliwal et al., 1999; O’Hanlon
and Pope, 1999; Lee et al., 2006; Chambers et al., 2007; Goncharov and Hodgson, 2008).
While two studies (Cahan et al., 2000; Barton et al., 2009) have previously investigated
the properties of NI and comprehensive income in NZ, those studies were conducted
prior to the new IFRS regime in NZ. Adoption of IFRS affects ?nancial account
numbers in NZ (Kabir et al., 2010; Stent et al., 2010). Further, IAS 1 changes the
reporting location of comprehensive income in NZ from a statement of changes in
equity to a performance statement. Prior research (Maines and McDaniel, 2000)
ARJ
24,3
284
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
suggests that location of comprehensive income reporting may affect the weight users
attach to comprehensive income. Hence, results of prior research may not be
generalisable to the current IFRS setting in NZ. This study addresses this gap and
further investigates whether the value relevance of TCI depends on its reporting
location, because IAS 1 allows the option of reporting TCI either in a single statement
of comprehensive income or in a two-statement format (IASB, 2009), and the IASB
(2010) proposes the elimination of the option of the two-statement format. The results
suggest that the IASB (2010) proposal to require the reporting of TCI in a single
statement may not enhance the value relevance of TCI.
This study adds current evidence on the properties of NI, TCI and OCI under IFRS
to the international literature on comprehensive income. The results may be of
potential interest to securities analysts who use ?rm fundamentals to value a ?rm and
other users who use earnings in different contractual settings. The results may also be
of potential interest to the IASB.
The main limitation of this study is the small size of the sample. Thus, the results
may lack generalizability to other samples and statistical power. The small sample,
however, re?ects the small number of ?rms listed in NZ. It further re?ects that a the
time of the study, only one year of data are available for all the listed companies in NZ
after the implementation of the new requirement of comprehensive income reporting
under IAS 1. Further, due to the small sample size, the study could not investigate
whether the results vary across industries. Some industries (e.g. ?nancial institutions,
agriculture and ?shing) are markedly different from other industries, as ?rms
belonging to the former industries have to include some fair value gains and losses in
NI. Thus, the results documented here may not be equally applicable to all industries.
Future research may investigate this issue.
Notes
1. OCI incorporates items of income and expense that are not recognised in pro?t or loss under
IFRS (IASB, 2009, para 7). TCI is the change in owners’ equity from non-owner sources
(IASB, 2009, para 7). TCI is the sum of pro?t/loss and OCI (IASB, 2009, para 7).
2. During the pre-IFRS era, NZ entities reported a similar amount, called “total recognized
revenues and expenses” in a separate Statement of Movements in Equity in accordance with
Financial Reporting Standard 2 (Cahan et al., 2000).
3. Cahan et al. (2000) cover the 1992-1997 period, while Barton et al. (2009) cover 1996-2005.
4. NZ allowed adoption of IFRS from 2005 (ASRB, 2004, para 20). Following this option,
some NZ listed companies early adopted IFRS (Kabir et al., 2010).
5. For example, upward revaluation of property, plant and equipment (PPE) and intangibles is
an other comprehensive income item under IAS 1 (IASB, 2009). But US GAAP do not permit
upward revaluation of PPE and intangibles (Bellandi, 2009).
6. Dechow and Schrand (2004) note limitations of the above criteria. First, managed earnings
could be both persistent and predictable. Second, actual cash ?ows can be a noisy metric of
value. Third, fromthe perspective of faithful representation, variability of earnings is not bad
if it re?ects the underlying variation of the operation of the company. However, variability in
earnings may be induced by accounting also. Fourth, using contemporaneous association
between earnings and stock price assumes that the market is ef?cient and stock price quickly
re?ects all publicly available information. Despite these limitations, the accounting literature
supports the use of these attributes as indices of high earnings quality. Prior research shows
Properties of
NI and TCI
285
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that persistence of earnings components is a desirable attribute fromthe perspective of equity
valuation (Lipe, 1986). The conceptual framework for ?nancial reporting issued jointly by the
FASBandthe IASBargues that accountinginformationshouldassist users assess the amount,
timing anduncertainty of future cash ?ows to the entity (FASB, 2010, para OB3). Earnings are
used in a variety of contractual settings and widely used by analysts in ?rm valuation
(Dechow, 1994; Dechow and Schrand, 2004). These make the ability to predict future cash
?ows andnet income desirable attributes of accounting information. Further, the newly issued
conceptual framework focuses on the information needs of investors (FASB, 2010, para OB2
and BC1.16) and stipulates relevance as one of the two fundamental qualitative characteristics
of ?nancial information (FASB, 2010, para QC5). Association of the income metrics with
contemporaneous stock returns is one way to operationalise the relevance criterion to
investors (Barth et al., 2001).
7. In its exposure draft on presentation of items of other comprehensive income issued in
May 2010, the IASB (2010) proposes to change the title of the statement of comprehensive
income to “statement of pro?t or loss and other comprehensive income”. It further proposes,
inter alia, to eliminate the currently available option of reporting TCI in two statements.
Instead it proposes to require the reporting of TCI in a single statement with two sections –
pro?t/loss and OCI. It also proposes to require the disclosure of items of OCI that are recycled
to pro?t/loss separate from items of OCI that are not.
8. The sequence {Y
t
} follows a sub-martingale process if E(Y
tþ1
jY
0
, . . . , Y
t
) $Y
t
for all t. E is
an expectation operator. The random walk model is a sub-class of the sub-martingale
process (Ball and Watts, 1972).
9. Most studies on comprehensive income used estimates of TCI because accounting standards
did not require disclosure of TCI during the study periods.
10. Ohlson (1999) de?nes transitory earnings as possessing three characteristics:
(a) unpredictability, (b) forecast irrelevance, and (c) value irrelevance.
11. Barth et al. (1995, p. 579) note that American bankers made this argument at the time of the
FASB public hearing before promulgating SFAS 115.
12. However, if NI is managed to give it a more persistent look than TCI, investors may not view
persistence as desirable (Barton et al., 2009; Dechow and Schrand, 2004).
13. NZX Deep Archive is a subscription database service of the NZ Stock Exchange.
14. In case of one company, the net income ?gure reported in the income statement is different
from that reported in the earnings per share note. In case of another company, the number of
shares at the beginning of 2010 is not the same as that at the end of the 2009. Hence, these
two companies are excluded from the sample on the ground of lack of reliability of ?gures.
15. Vuong’s (1989) test is used to assess the statistical signi?cance of the difference in the
explanatory powers of two non-nested regression models. Dechow (1994) describes how
the Vuong (1989) statistic is derived. After Dechow (1994), this test has been widely used in
the accounting literature. The authors, however, note that Vuong’s (1989) test is sensitive to
sample size. This limitation is relevant to this study as the sample size is small.
16. Kothari and Zimmerman (1995, p. 178) argue that the earnings response coef?cient will be
smaller when earnings contains transitory items than when it does not. When earnings
contains both a randomwalk component and a zero-mean transitory component, the earnings
response coef?cient will be the weighted average of the coef?cient of the random walk
component of earnings and that of the transitory component.
17. Dhaliwal et al. (1999, Table 2) reported that the Vuong (1989) statistic for the difference in the
adjusted R
2
of 4.20 percent and 3.81 percent is signi?cant; their sample size is 11425.
ARJ
24,3
286
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
18. Dhaliwal et al. (1999) reported an adjusted R
2
of 0.528 and 0.521 for these models, and thus,
the explanatory powers of models (9) and (10) are comparable to those of Dhaliwal et al.
(1999).
References
Accounting Standards Review Board (2004), Release 8 The Role of the Accounting Standards
Review Board and the Nature of Approved Financial Reporting Standards, available at:
www.asrb.co.nz/default.asp?p¼3 (accessed 12 July 2008).
Ball, R. and Watts, R.L. (1972), “Some time series properties of accounting income”, Journal of
Finance, Vol. 27 No. 3, pp. 663-81.
Bamber, L.S., Jiang, J., Petroni, K. and Wang, I.Y. (2010), “Comprehensive income: who’s afraid of
performance reporting?”, The Accounting Review, Vol. 85 No. 1, pp. 97-126.
Barth, M.E. (2006), Research, Standard Setting, and Global Financial Reporting, Now Publishers
Inc., Hanover, MA.
Barth, M.E., Beaver, W.H. and Landsman, W.R. (2001), “The relevance of the value relevance
literature for ?nancial accounting standard setting: another view”, Journal of Accounting
& Economics, Vol. 31, pp. 77-104.
Barth, M.E., Cram, D.P. and Nelson, K.K. (2001), “Accruals and the prediction of future cash
?ows”, The Accounting Review, Vol. 76 No. 1, pp. 27-58.
Barth, M.E., Landsman, W.R. and Wahlen, J. (1995), “Fair value accounting: effects on banks’
earnings volatility, regulatory capital, and value of contractual cash ?ows”, Journal of
Banking & Finance, Vol. 19, pp. 577-605.
Barton, J., Hansen, B. and Pownall, G. (2009), “Which performance measures do investors value
the most – and why?”, available at: SSRN: http://ssrn.com/abstract¼1230563 (accessed
22 January 2011).
Bellandi, F. (2009), Accounting for Equity and Other Comprehensive Income, Casa Editrice
Universita` La Sapienza, Roma.
Biddle, G.C., Seow, G.M. and Siegel, A.F. (1995), “Relative versus incremental information
content”, Contemporary Accounting Research, Vol. 12, pp. 1-23.
Black, F. (1993), “Choosing accounting rules”, Accounting Horizons, Vol. 7 No. 4, pp. 1-17.
Bowen, R.M., Burgstahler, D. and Daley, L.A. (1986), “Evidence on the relationships between
earnings and various measures of cash ?ow”, The Accounting Review, Vol. LXI No. 4,
pp. 713-25.
Cahan, S.F., Courtenay, S.M., Gronewoller, P.L. and Upton, D.R. (2000), “Value relevance of
mandated comprehensive income disclosures”, Journal of Business Finance & Accounting,
Vol. 27 Nos 9/10, pp. 1273-301.
Chambers, D., Linsmeier, T.J., Shakespeare, C. and Sougiannis, T. (2007), “An evaluation of SFAS
No. 130 comprehensive income disclosures”, Review of Accounting Studies, Vol. 12,
pp. 557-93.
Cheng, C.S.A., Cheung, J.K. and Gopalakrishnan, V. (1993), “On the usefulness of operating
income, NI and comprehensive income in explaining security returns”, Accounting
& Business Research, Vol. 23, pp. 195-203.
Dechow, P.M. (1994), “Accounting earnings and cash ?ows as measures of ?rm performance: the
role accounting accruals”, Journal of Accounting & Economics, Vol. 18, pp. 3-42.
Dechow, P.M. and Schrand, C.M. (2004), Earnings Quality, Research Foundation for CFA
Institute, Charlottesville.
Properties of
NI and TCI
287
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dechow, P.M., Kothari, S.P. and Watts, R.L. (1998), “The relation between earnings and cash
?ows”, Journal of Accounting & Economics, Vol. 25, pp. 133-68.
Dhaliwal, D., Subramanyam, K.R. and Trezevant, R. (1999), “Is comprehensive income superior
to NI as a measure of ?rm performance?”, Journal of Accounting & Economics, Vol. 26,
pp. 43-67.
Easton, P.D. (1999), “Security returns and the value relevance of accounting data”, Accounting
Horizons, Vol. 13 No. 4, pp. 399-412.
Enria, A., Cappiello, L., Dierick, F., Grittini, S., Haralambous, A., Maddaloni, A., Molitor, P., Pires, F.
and Poloni, P. (2004), “Fair value accounting and ?nancial stability”, European Central Bank
Occasional Paper Series No. 13, available at: www.ecb.int/pub/pdf/scpops/ecbocp13.pdf
(accessed 12 June 2008).
Financial Accounting Standards Board (FASB) (2010), Statement of Financial Accounting
Concepts No. 8, Conceptual Framework for Financial Reporting Chapter 1, The Objective
of General Purpose Financial Reporting, and Chapter 3, Qualitative Characteristics of
Useful Financial information, available at: www.fasb.org/cs/BlobServer?
blobcol¼urldata&blobtable¼MungoBlobs&blobkey¼id&blobwhere¼1175821997186&
blobheader¼application%2Fpdf (accessed 30 January 2011).
Goncharov, I. and Hodgson, A. (2008), “The comprehensive income issue in Europe”, available at:
http://ssrn.com/abstract¼1313134 (accessed 16 November 2010).
Hirst, D.E. and Hopkins, P.E. (1998), “Comprehensive income reporting and analysts’ valuation
judgments”, Journal of Accounting Research, Vol. 36, pp. 47-75.
Hodder, L.D., Hopkins, P.E. and Wahlen, J.M. (2006), “Risk-relevance of fair value income
measures for commercial banks”, The Accounting Review, Vol. 81, pp. 337-75.
International Accounting Standards Board (IASB) (2009), International Accounting Standard 1,
Presentation of Financial Statements.
International Accounting Standards Board (IASB) (2010), Exposure Draft ED/2010/5,
Presentation of items of other comprehensive income – Proposed amendments to IAS 1,
available at: www.ifrs.org/NR/rdonlyres/58DEA5EA-8CFC-45A7-910E-A55F9B7EA04F/
0/ED_OCIMay10.pdf (accessed 12 December 2010).
Isidro, H., O’Hanlon, J. and Young, S. (2004), “Dirty surplus accounting ?ows: international
evidence”, Accounting & Business Research, Vol. 34 No. 4, pp. 383-410.
Kabir, M.H., Laswad, F. and Islam, A. (2010), “Impact of IFRS in New Zealand on accounts and
earnings quality”, Australian Accounting Review, Vol. 20, pp. 343-57.
Kanagaretnam, K., Mathieu, R. and Shehata, M. (2009), “Usefulness of comprehensive income
reporting in Canada”, Journal of Accounting & Public Policy, Vol. 28, pp. 349-65.
Kothari, S.P. and Zimmerman, J.L. (1995), “Price and return models”, Journal of Accounting
& Economics, Vol. 20, pp. 155-92.
Kubota, K., Suda, K. and Takehara, H. (2009), “Information content of other comprehensive
income and net income: evidence from Japanese ?rms”, available at: http://ssrn.com/
abstract¼1393746 (accessed 31 January 2011).
Lee, Y.-J., Petroni, K.R. and Shen, M. (2006), “Cherry picking, disclosure quality, and
comprehensive income reporting choices: the case of property-liability insurers”,
Contemporary Accounting Research, Vol. 23 No. 3, pp. 655-92.
Linsmeier, T.J., Gribble, J., Jennings, R.G., Lang, M.H., Penman, S.H., Petroni, K.R., Shores, D.,
Smith, J.H. and War?eld, T.D. (1997), “An issues paper on comprehensive income”,
Accounting Horizons, Vol. 11 No. 2, pp. 120-6.
ARJ
24,3
288
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Lipe, R.C. (1986), “The information contained in the components of earnings”, Journal of
Accounting Research, Vol. 24, pp. 37-68.
Maines, L.A. and McDaniel, L.S. (2000), “Effects of comprehensive-income characteristics on
nonprofessional investors’ judgments: the role of ?nancial-statement presentation format”,
The Accounting Review, Vol. 75 No. 2, pp. 179-207.
O’Hanlon, J.F. and Pope, P.F. (1999), “The value-relevance of UK dirty surplus accounting ?ows”,
British Accounting Review, Vol. 31, pp. 459-82.
Ohlson, J.A. (1999), “On transitory earnings”, Review of Accounting Studies, Vol. 4, pp. 145-62.
Penman, S.H. and Zhang, X.-J. (2002), “Accounting conservatism, the quality of earnings, and
stock returns”, Accounting Review, Vol. 77 No. 2, pp. 237-64.
Smithson, C., Smith, C. and Wilford, D. (1995), Managing Financial Risk: A Guide to Derivative
Products, Financial Engineering and Value Maximization, Irwin Professional Publishing,
New York, NY.
Stent, W., Bradbury, M. and Hooks, J. (2010), “IFRS in New Zealand: effects on ?nancial
statements and ratios”, Paci?c Accounting Review, Vol. 22 No. 2, pp. 92-107.
Vuong, Q.H. (1989), “Likelihood ratio tests for model selection and non-nested hypotheses”,
Econometrica, Vol. 57, pp. 307-33.
Watts, R.L. and Leftwich, R. (1977), “The time series of annual accounting earnings”, Journal of
Accounting Research, Vol. 15 No. 2, pp. 253-71.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and direct test for
heteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
Wooldrige, J.M. (2002), Econometric Analysis of Cross Section and Panel Data, The MIT Press,
Cambridge, MA.
About the authors
M. Humayun Kabir is currently a Senior Lecturer of Financial Accounting at Auckland
University of Technology. He has published in the Australian Accounting Review and the
Managerial Auditing Journal. His research interests are in ?nancial reporting and corporate
governance. M. Humayun Kabir is the corresponding author and can be contacted at:
[email protected]
Fawzi Laswad is a Professor and Head of School of Accountancy at Massey University.
He has wide research interests in ?nancial reporting and accounting education. He served as a
member of the Accounting Standards Review Board from 2002 to 2010, is a member of NZICA
Admissions Board and the Chair of the Academic Board of Advanced Business Education
Limited.
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
Properties of
NI and TCI
289
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
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Hong He, Zhijun Lin. 2015. Analyst Following, Information Environment and Value Relevance of
Comprehensive Income: Evidence from China. Asia-Pacific Journal of Financial Studies 44, 688-720.
[CrossRef]
2. Allan Hodgson, Mark Russell. 2014. Comprehending Comprehensive Income. Australian Accounting
Review 24:10.1111/auar.2014.24.issue-2, 100-110. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
4
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_990447411.pdf