You are on page 1of 23

Journal of Accounting,

Auditing & Finance


Is Other Comprehensive 1–23
ÓThe Author(s) 2016
Income Reported in the Reprints and permissions:
sagepub.com/journalsPermissions.nav
Income Statement More DOI: 10.1177/0148558X16670779
jaf.sagepub.com

Value Relevant? The Role


of Financial Statement
Presentation

Steve Lin1, Donel Martinez1, Changjiang Wang2,


and Ya-wen Yang3

Abstract
This study examines (a) whether other comprehensive income (OCI) is more value relevant
when reported in a more prominent and transparent location in financial statements, (b)
how economic condition affects value relevance of OCI reported in different locations, and
(c) whether value relevance of OCI is affected by the implementation of Accounting
Standard Update (ASU) 2011-05, which requires reporting OCI in either a combined state-
ment of income and comprehensive income or a separate statement of comprehensive
income (SCI), and prohibits reporting OCI only in the statement of changes in shareholders’
equity (SSE). Using manually collected data for the period 2000-2012, we find that investors
consistently priced OCI reported in the SSE during the entire test period. We also find that
investors priced OCI reported in the SCI only during recent financial crisis period (2007-
2009) when the magnitude and volatility of OCI significantly increased. Further evidence
shows that this finding is not driven by the investment bias of transient investors. Finally,
we find that value relevance of OCI decreases for firms that changed the reporting location
of OCI from SSE to SCI following the implementation of ASU 2011-05 compared with firms
that did not change the reporting location. As investors are expected to be more capable
of incorporating information reported in a more transparent location into price, this study
explores some plausible explanations for the puzzling finding and calls for future research.

Keywords
other comprehensive income, reporting location, financial statement presentation, value
relevance, comprehensive income

1
Florida International University, Miami, USA
2
University of Cincinnati, OH, USA
3
Wake Forest University, Winston-Salem, NC, USA

Corresponding Author:
Steve Lin, School of Accounting, College of Business, Florida International University, 11200 SW 8th Street, Miami,
FL 33199, USA.
Email: lins@fiu.edu

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


2 Journal of Accounting, Auditing & Finance

Introduction
Statement of Financial Accounting Standards No. 130 (SFAS 130) requires U.S. firms to
report other comprehensive income (OCI) and its components1 in either (a) a combined
statement of income and comprehensive income that reports both the components and
totals of net income and comprehensive income; (b) a separate statement of comprehensive
income that starts with net income, reports the components of OCI, and ends with compre-
hensive income (SCI)2; or (c) the statement of change in shareholders’ equity (SSE).
Accounting Standard Update No. 2011-05 (ASU 2011-05) eliminates the option to present
OCI in the SSE to increase the prominence of OCI and to converge the different practices
in reporting OCI (Financial Accounting Standards Board [FASB], 2011) between the U.S.
generally accepted accounting principles (GAAP) and the International Financial Reporting
Standard (IFRS). This study contributes to the literature by investigating three questions: Is
OCI reported in the SCI more value relevant than OCI reported in the SSE? Does economic
condition affect how investors value OCI reported in different locations? Has changing the
reporting location of OCI from SSE to SCI after the implementation of ASU 2011-05
increased value relevance of OCI?
Prior research on how financial statement presentation affects the way investors use the
information contained in OCI has yielded mixed results (Rees & Shane, 2012). On one
hand, Hirst and Hopkins (1998) find that analysts are better able to detect earnings manage-
ment and are less influenced by the presence of earnings management in their valuation
judgments when firms report OCI in a performance statement3 as opposed to in the SSE.
Maines and McDaniel (2000) find that investors consider the volatility of OCI when OCI is
reported in a performance statement. Moreover, recent studies (e.g., Bamber, Jiang,
Petroni, & Wang, 2010; Lee, Petroni, & Shen, 2006) find that firms with incentives to
manage earnings are more likely to avoid reporting OCI in a performance statement. Taken
together, these studies suggest that reporting OCI in the SCI provides more transparent
information to investors and could help deter managerial opportunistic activities, thus lend-
ing support to the FASB’s recent decision to eliminate the option of reporting OCI only in
the SSE. On the other hand using archival data, Chambers, Linsmeier, Shakespeare, and
Sougiannis (2007) find that investors price OCI when it is reported in the SSE. Given that
SSE is the predominant location for reporting OCI during the 1998-2003 period, they argue
that investors are better able to use the information contained in OCI when OCI is reported
in the SSE, a location that is more familiar to investors. The finding of Chambers et al.
(2007), however, is inconsistent with both FASB’s recent decision and the findings of prior
research that support reporting OCI in the SCI.
Rees and Shane (2012) suggest that the findings in Chambers et al. (2007) may be
driven by a relatively small number of firms reporting OCI in a performance statement
(around 20% of the sample firms) and self-selection bias. Our study addresses these issues
and provides further evidence on whether OCI is more value relevant when reported in the
SCI than in the SSE by using an extended dataset and a research design that controls for
self-selection bias.
As OCI largely contains changes in fair values of certain long-term financial assets and
liabilities, the magnitude and volatility of OCI are likely to increase when the fair values of
financial assets and liabilities become more volatile during economic turmoil.4 Presumably,
investors should pay more attention to OCI when its magnitude and volatility significantly
increase. However, little is known from prior literature about how investors value OCI
under volatile market condition. Our study, therefore, investigates the extent to which

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 3

financial statement presentation affects how investors value OCI during recent financial
crisis.
Moreover, many U.S. firms changed the reporting location of OCI from SSE to SCI
after the introduction of ASU 2011-05. Thus, ASU 2011-05 provides a natural setting to
investigate the effect of this change on value relevance of OCI. More specifically, our
study examines whether a switch of reporting OCI from SSE to SCI affects value relevance
of OCI.
To conduct our analyses, we manually collected OCI data for the period 2000-2012. In
addition to the test using the full sample, we divided the sample into three sub-periods:
2000-2006 (pre-financial crisis period), 2007-2009 (financial crisis period), and 2010-2012
(adoption period of ASU 2011-05). We use the 2000-2006 data to address the concerns of
Rees and Shane (2012) about self-selection issues in Chambers et al. (2007). We find that
about 81% of our sample firms reported OCI in the SSE during this period, similar to
Chambers et al. (2007). We also find that investors priced OCI when it is reported in the
SSE. In addition, OCI components, such as adjustments for foreign currency translation,
available-for-sale investments, and pension liabilities, are more value relevant when they
are reported in the SSE. These results are consistent with Chambers et al. (2007) that inves-
tors pay more attention to items reported in a predominant and familiar location after using
extended data and controlling for self-selection bias as suggested by Bamber et al. (2010)
and Rees and Shane (2012).
Next, we find that the absolute value and volatility of OCI significantly increased in the
financial crisis period (2007-2009), which indicates that OCI significantly fluctuated when
fair values of financial assets and liabilities were volatile. More importantly, we find that
investors priced OCI reported in both the SSE and SCI during this period, suggesting that
investors value OCI reported in the SCI when OCI is significant and volatile. This suggests
that significant and volatile OCI could increase investors’ attention toward OCI. In an addi-
tional analysis, we examine whether investor sophistication, investment horizon, and report-
ing location jointly affect value relevance of OCI. Prior research finds that transient
investors have a shorter-term horizon and are more likely to adjust their trading strategies
based on near-term factors (Bushee, 2001). Elliott, Krische, and Peecher (2010) find that
financial analysts expect transient investors to respond to OCI when it is reported more
transparently in the SCI. In other words, the finding that investors priced OCI reported in
the SCI during financial crisis period could be driven by transient investors’ overreaction to
this information. Our evidence, however, suggests that transient investors did not overprice
the information contained in OCI when OCI was reported in the SCI during recent financial
crisis.
Using the 2010-2012 data, we find that about 41% of our sample firms reported OCI in
the SCI, a 21% increase compared with the pre-adoption period. Further analysis shows
that investors only priced OCI reported in the SSE in the post-financial crisis period, sug-
gesting that investors no longer valued OCI reported in the SCI when OCI was not affected
by severe economic condition. Compared with the sample in earlier years, the 2010-2012
sample contains a more balanced representation of firms reporting OCI in each location,
which allows us to address the unbalanced sample issue (i.e., a small percentage of firms
reporting OCI in the SCI) in Chambers et al. (2007) as suggested by Rees and Shane
(2012).
We also use the 2010-2012 sample to investigate whether changing the reporting loca-
tion of OCI from SSE to SCI affects its value relevance. If reporting OCI in the SCI indeed
makes information more transparent, as suggested by prior experimental research and

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


4 Journal of Accounting, Auditing & Finance

FASB and the International Accounting Standards Board (IASB), one should observe an
increase in value relevance of OCI after the introduction of ASU 2011-05. Using a differ-
ence-in-differences test, we find that firms whose reporting location of OCI changed from
SSE to SCI during 2010-2012 experienced a decrease in value relevance of OCI relative to
firms that did not change the reporting location. These results are consistent with our earlier
finding and those of Chambers et al. (2007), but generally inconsistent with the findings of
prior experimental research (Hirst & Hopkins, 1998), and the belief of FASB and IASB
that financial statement users can better perceive the information contained in OCI when it
is reported in a more transparent location such as the SCI.
Overall, we find that although reporting OCI in the SCI may increase information trans-
parency and reduce managerial opportunistic behavior, it does not appear to have enhanced
value relevance of OCI, which primarily consists of highly volatile and transitory unrea-
lized gains and losses caused by market fluctuations. Our results have several implications.
First, although ASU 2011-05 requires firms to report OCI in the SCI, investors consistently
valued OCI when it was only reported in the SSE, a location that was more familiar to
investors before the introduction of ASU 2011-05. Second, investors may not price OCI
reported in the SCI if they do not perceive OCI reported in the SCI to affect firms’ future
earnings. This is because OCI is generally small in amount, transitory and unpredictable in
nature, and could be reversed in future years (Jones & Smith, 2011). In contrast, investors
may value OCI reported in the SCI if they perceive OCI to significantly affect firms’ future
earnings when the magnitude and volatility of OCI are significant (Huang, Lin, &
Raghunandan, 2016). Finally, reporting OCI in a more prominent location of financial
statements such as the SCI could significantly increase the volatility and reduce the predic-
tive value of accounting income, which could adversely affect value relevance of OCI. As
a result, future research should investigate how investors value OCI when the SCI becomes
a prevailing reporting location of OCI following ASU 2011-05. Future research should also
investigate whether the change of OCI reporting location has any real effect on investors’
and other financial statement users’ decision making.
The remaining article is organized as follows: The next section reviews prior literature.
The section ‘‘Data and Sample’’ describes data and sample selection criteria. The
‘‘Research Design’’ section discusses the research design. The section ‘‘Results’’ reports
the empirical results, and the final section concludes.

Prior Literature
Previous archival studies have largely focused on value relevance and the predictive ability
of comprehensive income (CI) and OCI. Using 1994 and 1995 data, Dhaliwal,
Subramanyam, and Trezevant (1999) investigate the relative usefulness of CI and net
income (NI) for investors. They find that both NI and CI are value relevant, and OCI pro-
vides incremental explanatory power for returns beyond NI. They also find that NI is a
better performance indicator to predict stock returns, future NI, and future operating cash
flows (CFO) than CI. Dhaliwal et al. (1999) also find that the marketable securities adjust-
ment is the only component of OCI with incremental explanatory power for stock returns
beyond NI.
A few studies provide further evidence on value relevance of OCI. Using a larger set of
U.S. data for 1994-1998 (the pre-SFAS 130 period), Biddle and Choi (2006) examine value
relevance of various combinations of NI and components of OCI. Contrary to Dhaliwal
et al. (1999), they find that income measured by various combinations of NI and

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 5

components of OCI generally has higher explanatory power for stock returns than NI. Their
findings indicate that CI dominates NI in predicting stock returns. Biddle and Choi (2006)
also find that adjustments for foreign currency translation, available-for-sale securities, and
minimum pension liabilities provide significant incremental explanatory power for stock
returns beyond NI. Similarly, Kanagaretnam, Mathieu, and Shehata (2009) find that OCI is
value relevant, although the results appear to be stronger for the post- than for the pre-
SFAS 130 period. Consistent with Dhaliwal et al. (1999) but inconsistent with Biddle and
Choi (2006), Kanagaretnam et al. (2009) find that compared with CI, NI is a better predic-
tor of future NI, CI, and CFO.
As companies did not explicitly disclose OCI in their financial statements prior to SFAS
130 (effective from 1998), OCI could only be measured by estimation and may contain sig-
nificant measurement errors. Chambers et al. (2007) argue that the weak findings in
Dhaliwal et al. (1999) might be due to the measurement errors contained in the ‘‘as-if’’
estimates. After dividing their sample into the pre- (1994-1997) and post-SFAS 130 (1998-
2003) periods, Chambers et al. (2007) find that the ‘‘as-if’’ estimate of OCI is not value
relevant but the ‘‘as-reported’’ OCI is. They further find that OCI provides incremental
value-relevant information beyond NI in the post-SFAS 130 period. For the OCI compo-
nents, they find that adjustments for foreign currency translation and available-for-sale
investments are value relevant, and provide incremental price-relevant information beyond
NI in the post-SFAS 130 period.
More importantly, Chambers et al. (2007) is the first study to investigate whether the
reporting location of OCI affects value relevance of OCI. They find that investors appear
to price OCI if it is only reported in the SSE, a predominant location before ASU 2011-05
was issued. They also document that investors understand the transitory nature of OCI and
price OCI on a dollar-for-dollar basis.
As Rees and Shane (2012) point out, the findings of Chambers et al. (2007) contradict
those of previous experimental studies. Using buy-side analysts and manipulating whether
a company manages its earnings through available-for-sale marketable securities, Hirst and
Hopkins (1998) find that analysts can better detect earnings management when OCI is
reported in the SCI. Using non-professional investors and three presentation formats,
including reporting OCI in a footnote, the SSE, and a performance statement, Maines and
McDaniel (2000) find that investors consider the volatility of OCI when OCI is reported in
a performance statement. Maines and McDaniel (2000) conclude that the reporting format
affects how investors weigh OCI.
Presentation prominence of OCI also affects managers’ other financial reporting deci-
sions. Lee et al. (2006) find that firms in the property-liability insurance industry cherry
pick available-for-sale investments to sell, recognize disposal gains in the income state-
ment, and leave unrealized losses in the less-prominent SSE. Their results are generally
consistent with those of Bamber et al. (2010) that managers with strong incentives to
manage earnings are more likely to avoid reporting OCI in a performance statement.
Taken together, prior research examining how users of financial statements perceive and
value OCI information has yielded mixed results. To address the potential limitations in
Chambers et al. (2007) as pointed out by Rees and Shane (2012), we first extend Chambers
et al. (2007) by expanding the sample period to include the adoption period of ASU 2011-
05 and by controlling for self-selection bias. To provide additional evidence, we investigate
two major events that may affect value relevance of OCI. First, we examine whether finan-
cial market condition affects value relevance of OCI. OCI contains changes in fair values
of certain long-term financial assets and liabilities that are significant and more volatile

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


6 Journal of Accounting, Auditing & Finance

when financial market condition is severe. We therefore predict that investors may price
OCI reported in the SCI more if they perceive significant and volatile OCI to adversely
affect firms’ future earnings. Second, we examine whether value relevance of OCI changes
after firms shifted the reporting location of OCI from SSE to SCI as required by ASU
2011-05. We predict that if reporting OCI in the SCI increases information transparency,
then we should observe an increase in value relevance of OCI after the introduction of
ASU 2011-05.
A concurrent study by Schaberl and Victoravich (2015) examines the extent to which
ASU 2011-05 affected value relevance of OCI. Our study is different from that of Schaberl
and Victoravich (2015) as we also investigate (a) whether a relatively small number of firms
reporting OCI in the SCI in the pre-ASU 2011-05 period and self-selection bias, as suggested
by Rees and Shane (2012) and Bamber et al. (2010), affect empirical findings; (b) whether
financial market condition that increases the magnitude and volatility of OCI affects how
investors value OCI; (c) whether financial statement presentation affects how investors value
the components of OCI relative to aggregate OCI; and (d) the extent to which investor
sophistication and investment horizon affect how investors value OCI reported in the SCI.

Data and Sample


Our sample consists of S&P 500 firms in the period 2000-2012. We also form sub-samples
in the pre-financial crisis (2000-2006), financial crisis (2007-2009), and post-financial
crisis periods (2010-2012).5 We manually collected the reporting location of OCI for
sample firms, and reported OCI and its components for the period 1998-2012 from their
10-K filings.6 Sample firms’ financial data were obtained from the Compustat database,
and stock prices and returns were obtained from the Center for Research in Security Prices
(CRSP) database. We follow Chambers et al. (2007) and employ the diagnostics of
Belsley, Kuh, and Welsch (1980) to remove potential outliers.7
The sample for the period 2000-2012 has 4,669 firm-year observations, including 2,639
firm-year observations for the pre-financial crisis period, 964 firm-year observations for the
financial crisis period, and 1,066 firm-year observations for the post-financial crisis period.
In the pre-financial crisis period, 501 (19%) and 2,138 (81%) observations reported OCI
in the SCI and SSE, respectively. This finding is consistent with Chambers et al. (2007) that
less than 20% of U.S. firms reported their OCI in the SCI during 1998-2003. Among the
observations in the financial crisis period, 192 (20%) and 722 (80%) observations reported
OCI in the SCI and SSE, respectively; among the observations in the post-financial crisis
period, 442 (41.5%) and 624 (58.5%) observations reported OCI in the SCI and SSE, respec-
tively. Compared with the samples for the periods 2000-2006 and 2007-2009, our sample for
the period 2010-2012 is more balanced in terms of the number of observations in each report-
ing location, allowing us to address the unbalanced sample issue as suggested by Rees and
Shane (2012). We then use the post-financial crisis period sample (2010-2012) to test
whether change of the reporting location following ASU 2011-05 affects the pricing of OCI.

Research Design
The Effect of the Reporting Location of OCI on Its Value Relevance
We follow Chambers et al. (2007) to estimate the association between OCI and stock
returns conditional on the reporting location of OCI for the period 2000-2006 using the fol-
lowing model8:

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 7

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a4 DIS3OCIit + eit , ð1Þ

where RET is the buy-and-hold raw return for the window 8 months before to 4 months
after the fiscal year-end; NI is the net income after extraordinary items and discontinued
operations; LOSS is a dummy variable equal to 1 if a firm reports net loss, and 0 otherwise;
OCI is other comprehensive income as reported in 10-K; DSE is a dummy variable that
takes the value of 1 when OCI is reported in the SSE, and 0 otherwise; DIS is a dummy
variable that takes the value of 1 when OCI is reported in the SCI, and 0 otherwise; the
subscript it denotes firm i in year t. The variable definitions are summarized in the appen-
dix. Coefficients a3 and a4 in Model 1 capture the markets’ pricing of OCI when it is
reported in the SSE and SCI, respectively. A statistically significant coefficient a3 (a4) indi-
cates that OCI is value relevant when reported in the SSE (SCI).
Next, we examine whether the reporting location affects the pricing of each of the OCI
components. We estimate the following model, decomposing OCI into its components:

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit


+ a5 DSE3TDerivativesit + a6 DSE3TPensionit + a7 DSE3TOtherit
ð2Þ
+ a8 DIS3TCurrencyi t + a9 DIS3TAFSit + a10 DIS3TDerivativesit
+ a11 DIS3TPensionit + a12 DIS3TOtherit + eit :

The variables RET, NI, LOSS, DSE, and DIS are previously defined. TCurrency, TAFS,
TDerivatives, TPension, and TOther are unrealized gains and losses from foreign currency
translation, available-for-sale securities, derivative securities held as cash flow hedges,
defined benefit pension plan, and other items, respectively. Our variables of interests are
coefficients a3 through a12. Coefficients a3 through a7 capture the market pricing of OCI
components when reported in the SSE. Coefficients a8 through a12 capture the market pric-
ing of OCI components when reported in the SCI.

Controlling for Self-Selection of Reporting Location


Because firms had the choice of reporting OCI in the SSE or SCI prior to the implementa-
tion of ASU 2011-05, results from ordinary least squares (OLS) regressions may be subject
to potential selection bias. To address this concern, we extend Chambers et al. (2007) by
performing the Heckman (1979) two-stage procedure using the sample during 2000-2006.
In the first stage, we model the determinants of reporting location using the variables simi-
lar to those used by Bamber et al. (2010). Specifically, we estimate the following probit
regression:

DISit = b1 SIZEit + b3 LEVEit + b2 ROAit + b4 VOL OCIit + b5 Complexity OCIit


+ b6 Abs TAFSit + b7 Abs TCurrencyi t + b8 Abs TDerivativesit
ð3Þ
+ b9 Abs TPensionit + b10 Abs Tothert + Industry Fixed Effects
+ Year Fixed Effects + mit ,

where DIS is a dummy variable equal to 1 if the firm presents OCI in the SCI, and 0 other-
wise; SIZE is firm size, measured as the log of the market value of the firm’s common
shares outstanding at the fiscal year-end; ROA is the net income before extraordinary items,

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


8 Journal of Accounting, Auditing & Finance

scaled by total assets; LEVE is leverage, calculated as long-term debt divided by total
assets; VOL_OCI is the volatility of OCI relative to the volatility of NI, calculated as the
standard deviation of comprehensive income after it has been scaled by total assets, divided
by the standard deviation of net income after it has been scaled by total assets, over the ini-
tial comprehensive income reporting year and the prior 4 years; Complexity_OCI captures
the complexity of OCI, defined as the number of OCI components reported in 10-K;
Abs_TAFS (Abs_TCurrency, Abs_TDerivatives, Abs_TPension, and Abs_TOther) are the
ratio of the absolute value of TAFS (TCurrency, TDerivatives, TPension, and TOther) over
the total assets.
In the second stage, we add the inverse Mills ratio (IMR) obtained from the first-stage
regression to Models 1 and 2 to control for potential self-selection bias, and estimate the
following two models:

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a3 DIS3OCIit + a4 IMR + eit , ð4Þ

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit


+ a5 DSE3TDerivativesit + a6 DSE3TPensionit + a7 DSE3TOtherit
ð5Þ
+ a8 DIS3TCurrencyi t + a9 DIS3TAFSit + a10 DIS3TDerivativesit
+ a11 DIS3TPensionit + a12 DIS3TOtherit + a13 IMR + eit ,

where all variables in Models 4 and 5 are as previously defined.

The Effect of Reporting Location of OCI on Its Value Relevance During the
Financial Crisis
We investigate whether the association between the reporting location and value relevance
of OCI is driven by any sub-periods using the following regression model:

RETit = g0 + g1 NIit + g2 LOSS3NIit + g3 PRE CRISIS3DSE3OCIit


+ g4 PRE CRISIS3DIS3OCIit + g5 CRISIS3DSE3OCIit
ð6Þ
+ g6 CRISIS3DIS3OCIit + g7 POST CRISIS3DSE3OCIit
+ g8 POST CRISIS3DIS3OCIit + g9 CRISIS + g10 POST CRISIS + eit ,

where PRE_CRISIS is a dummy variable equal to 1 during the pre-financial crisis period
(2000-2006), and 0 otherwise; CRISIS is a dummy variable equal to 1 during the financial
crisis period (2007-2009), and 0 otherwise; POST_CRISIS is a dummy variable equal to 1
for the ASU 2011-05 adoption period (2010-2012), and 0 otherwise. All other variables are
previously defined.

The Effect of Changing the Reporting Location of OCI and Its Components on
Their Value Relevance
During 2010-2012, many firms moved the reporting location of OCI from SSE to SCI upon
the adoption of ASU 2011-05. We identify firms that switched the reporting location of
OCI after December 15, 2011 as the treatment firms, and the firms that continued to report
OCI in the SCI throughout the 3-year period as the control firms. The 2010-2012 sample

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 9

contains 1,014 observations, with 722 firm-year observations in the treatment group and
292 firm-year observations in the control group. We then perform a difference-in-differ-
ences regression to examine whether the change of reporting location affects the pricing of
OCI and its components using the following models:

RETit = g1 NIit + g2 LOSS3NIit + g3 TREAT + g4 POST + g5 TREAT3POST + g6 OCIit


+ g7 POST3OCIit + g8 TREAT3OCIit + g9 TREAT3POST3OCIit + eit :
ð7Þ

RETit = g1 NIit + g2 LOSS3NIit + g3 TREAT + g4 POST + g5 TREAT3POST + g6 TCurrencyi t


+ g7 TAFSit + g8 TDerivativesit + g9 TPensionit + g10 TOtherit
+ g11 POST3TCurrencyi t + g12 POST3TAFSit + g13 POST3TDerivativesit
+ g14 POST3TPensionit + g15 POST3TOtherit + g16 TREAT3POST3TCurrencyi t
+ g17 TREAT3POST3TAFSit + g18 TREAT3POST3TDerivativesit
+ g19 TREAT3POST3TPensionit + g20 TREAT3POST3TOtherit + eit ,
ð8Þ

where TREAT is an indicator variable that takes the value of 1 when the firm changed the
presentation of OCI from SSE to SCI after December 15, 2011, and 0 if the firm continued
to report OCI in the SCI. POST is an indicator variable which takes the value of 1 if the
fiscal year begins after December 15, 2011, and 0 if a firm’s fiscal year begins before
December 15, 2011.9 All other variables are previously defined.
In Model 7, the coefficient on OCI (g6) captures value relevance of OCI for the control
firms in the pre-ASU 2011-05 period; the coefficient on the interaction term POST 3 OCI
(g7) captures the incremental value relevance of OCI in the post-ASU 2011-05 period rela-
tive to the pre-ASU 2011-05 period for the control firms. The coefficient on the interaction
term TREAT 3 OCI (g8) captures the incremental value relevance of OCI for the treatment
firms relative to the control firms in the pre-ASU 2011-05 period. The coefficient on the
three-way interaction term TREAT 3 POST 3 OCI is our variable of interest; it captures
the incremental value relevance of OCI from the change of reporting location relative to
the time trend effect (as captured by POST 3 OCI) on value relevance of OCI. In Model 8,
the variables of interest are the interaction terms among TREAT, POST, and each of the
individual OCI components.
Unless noted otherwise, all the independent (financial) variables in Models 1 through 8
are scaled by the market value of common shares outstanding at 8 months before the fiscal
year-end. Following Chambers et al. (2007), we mean-difference all independent variables
within years to estimate fixed effects regressions.

Results
Main Results
Table 1 presents the descriptive statistics for the sample during the 2000-2006 period. We
use t statistics (Wilcoxon’s z statistics) to test the differences in means (medians) between
firms reporting OCI in the SSE and firms reporting OCI in the SCI. The mean (median)
RET for firms reporting OCI in the SCI (hereafter SCI firms) is 0.147 (0.136), significantly
greater than the mean (median) RET of 0.121 (0.117) for firms reporting OCI in the SSE

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


10 Journal of Accounting, Auditing & Finance

Table 1. Descriptive Statistics of the 2000-2006 Sample.

Full sample (n = 2,639) SSE = 0 (n = 501) SSE = 1 (n = 2,138)

Variable M Median SD M Median SD M Median SD t-stat z-stat

RET 0.126 0.121 0.269 0.147 0.136 0.266 0.121 0.117 0.269 1.95 2.01
NI 0.045 0.053 0.123 0.049 0.059 0.099 0.044 0.052 0.128 0.77 3.39
OCI 0.000 0.000 0.037 0.001 0.000 0.031 0.000 0.000 0.038 0.66 2.03
TCurrency 0.003 0.000 0.018 0.002 0.000 0.013 0.003 0.000 0.020 20.31 20.31
TAFS 0.000 0.000 0.011 0.002 0.000 0.017 0.000 0.000 0.009 2.85 2.68
TDerivatives 0.000 0.000 0.011 20.001 0.000 0.016 0.000 0.000 0.009 21.80 21.46
TPension 20.003 0.000 0.026 20.002 0.000 0.020 20.003 0.000 0.027 0.33 1.47
TOther 0.000 0.000 0.006 0.001 0.000 0.012 0.000 0.000 0.004 1.80 1.50
VOL_OCI 0.911 0.375 1.792 0.944 0.318 2.209 0.903 0.392 1.679 0.46 22.63
Complexity_OCI 2.916 3.000 1.101 3.036 3.000 1.181 2.887 3.000 1.080 2.72 2.77
Abs_TAFS 0.002 0.000 0.006 0.002 0.000 0.006 0.002 0.000 0.006 1.09 1.73
Abs_Tcurrency 0.006 0.002 0.009 0.005 0.001 0.008 0.006 0.002 0.009 21.89 22.16
Abs_Tderivatives 0.001 0.000 0.005 0.002 0.000 0.007 0.001 0.000 0.005 1.98 1.22
Abs_Tpension 0.005 0.000 0.016 0.004 0.000 0.015 0.005 0.000 0.016 20.63 20.63
Abs_Tother 0.001 0.000 0.004 0.001 0.000 0.007 0.001 0.000 0.003 4.04 3.95
SIZE 9.237 9.095 1.134 9.174 9.078 1.044 9.251 9.098 1.154 21.38 20.77
LEVE 0.242 0.228 0.170 0.243 0.216 0.185 0.242 0.231 0.167 0.17 20.58
ROA 0.055 0.048 0.096 0.051 0.043 0.063 0.056 0.050 0.103 21.03 22.87

Note. See the appendix for variable definitions. t-stat is the t statistic for the difference in means, and z-stat is the
z statistic from the Wilcoxon–Mann–Whitney rank test for the difference in medians. SSE = shareholders’ equity.

(hereafter SSE firms). SCI firms also have significantly higher median net income (0.059)
than SSE firms (0.052). Consistent with previous studies, both the mean and median of
OCI and OCI components for the full sample are close to 0, although SCI firms have a
higher median aggregate OCI than SSE firms (z statistics = 2.03). SCI firms have signifi-
cantly higher mean and median adjustments for available-for-sale securities (TAFS).
SCI firms have significantly lower median relative volatility of OCI versus NI
(VOL_OCI), measured by the relative standard deviation of OCI scaled by total assets
versus the standard deviation of NI scaled by total assets over the past 5 years, than do the
SSE firms. On average, sample firms report approximately three OCI components during
2000-2006 (Complexity_OCI). SCI firms have a higher magnitude of adjustments for deri-
vatives (Abs_TDerivatives) and other (Abs_TOther), but a lower magnitude of adjustments
for foreign currency translation (Abs_TCurrency) as a percentage of total assets. Finally,
we find that both groups of firms have similar market capitalization (SIZE) and leverage
(LEVE), and that SCI firms have significantly lower ROA compared with SSE firms.
Table 2 reports the regression results of Models 1 (the aggregate OCI model) and 2 (the
OCI component model), which examine the relative value relevance of OCI and its compo-
nents reported in the SSE and SCI. In the OCI model, we find that the coefficient on the
interaction term DSE 3 OCI is positive and significant at the 1% level (0.507, t statistic =
3.29), while the coefficient on the interaction term DIS 3 OCI is positive but insignificant
(0.179, t statistic = 0.74) after controlling for net income (positive and significant) and
losses (negative and significant). Turning to the component model, we also find that adjust-
ments for foreign currency translation (0.767, t statistic = 2.49), available-for-sale invest-
ments (1.519, t statistic = 1.78), and pension liabilities (0.461, t statistic = 2.06) reported in

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 11

Table 2. Value Relevance of OCI and Its Components Conditional on Reporting Location.
RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a4 DIS3OCIit + eit : ð1Þ
RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit + a5 DSE3TDerivativesit
+ a6 DSE3TPensionit + a7 DSE3TOtherit + a8 DIS3TCurrencyi t + a9 DIS3TAFSit
+ a10 DIS3TDerivativesit + a11 DIS3TPensionit + a12 DIS3TOtherit + eit : ð2Þ

Variable Coefficient t statistic Coefficient t statistic


Intercept 20.053 29.47 20.052 29.44
NI 1.628 8.39 1.601 8.51
LOSS 3 NI 21.544 27.60 21.517 27.70
DSE 3 OCI 0.507 3.29
DIS 3 OCI 0.179 0.74
DSE 3 TCurrency 0.760 2.49
DSE 3 TAFS 1.519 1.78
DSE 3 TDerivatives 21.090 21.54
DSE 3 TPension 0.461 2.06
DSE 3 TOther 20.026 20.02
DIS 3 TCurrency 1.727 1.78
DIS 3 TAFS 0.561 1.40
DIS 3 TDerivatives 21.932 23.37
DIS 3 TPension 0.325 0.95
DIS 3 TOther 20.536 20.95
Adjusted R2 .0938 .1009
Observations 2,639 2,639

Note. We use the sample for the period 1998-2006 to replicate Chambers, Linsmeier, Shakespeare, and Sougiannis
(2007) and statistically evaluate whether the reporting location affects investors’ pricing of OCI and its
components. Prior to ASU 2011-05, SFAS 130 required firms to disclose OCI and its components in (a) a
combined income statement that reports both the components and totals of net income and comprehensive
income; (b) a separate comprehensive income statement that starts with net income, reports the components of
OCI, and ends with comprehensive income (SCI); or (c) the statement of changes in equity (SSE). We estimate the
association between OCI (and its components) and stock returns conditional on the reporting location of OCI. See
the appendix for variable definitions. The t statistics are based on White (1980) heteroskedasticity-adjusted robust
variance estimates. OCI = other comprehensive income; SSE = shareholders’ equity.

the SSE are positively associated with stock returns, significant at the 10% level or better.
However, among the OCI components reported in the SCI, only adjustments for foreign
currency translation are positively associated with stock returns and are marginally signifi-
cant (1.727, t statistic = 1.78); adjustments for derivatives are negatively associated with
stock returns (21.932, t statistic = 23.37). Consistent with Chambers et al. (2007), our
results show that investors price aggregate OCI if it is only reported in the SSE. We find
similar results with OCI components. Overall, our results suggest that OCI information is
value relevant if OCI is only reported in the SSE during the period 2000-2006. Results in
Table 2 suggest that our extended data before recent financial crisis are comparable with
the data 1998-2003 used by Chambers et al. (2007).
After validating our data, we next examine whether the results are driven by self-selec-
tion bias as suggested by Rees and Shane (2012) and Bamber et al. (2010). Table 3 reports
the results of the first-stage regression of the Heckman (1979) two-stage procedure. The
results show that firms with smaller market capitalization (SIZE), higher complexity
(Complexity_OCI, that is, number of OCI components reported by firms), and higher

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


12 Journal of Accounting, Auditing & Finance

Table 3. First-Stage Regression Results of Reporting Location.


DISit = b1 SIZEit + b2 LEVEit + b3 ROAit + b4 VOL OCIit + b5 Complexity OCIit
+ b6 Abs TAFSit + b7 Abs TCurrencyi t + b8 Abs TDerivativesit + b9 Abs TPensionit
+ b10 Abs Totherit + Industry Fixed Effects + Year Fixed Effects + mit : ð3Þ

Variable Coefficient p value


Intercept 20.227 .6539
SIZE 20.128 .0094
LEVE 20.345 .2808
ROA 0.252 .694
VOL_OCI 20.009 .7676
Complexity_OCI 0.133 .0126
Abs_TAFS 5.667 .4851
Abs_TCurrency 20.778 .9067
Abs_TDerivatives 25.620 .586
Abs_TPension 20.553 .8795
Abs_Tother 32.806 .0227
Industry Fixed Effects Yes
Year Fixed Effects Yes
Adjusted R2 .0601

Note. In the first-stage regression, we examine the determinants of reporting location using the variables similar to
those used by Bamber, Jiang, Petroni, and Wang (2010). Specifically, we estimate a probit regression that models
the choice of reporting location on the firm characteristics and the types and magnitudes of OCI. DIS is a dummy
variable that equals 1 if the firm presents OCI in the SCI and 0 otherwise; SIZE is the log of the market value of the
firm’s common shares outstanding; LEVE is long-term debt divided by total assets; ROA is the net income before
extraordinary items, scaled by total assets; VOL_OCI is the volatility of OCI relative to the volatility of NI, calculated
as the standard deviation of comprehensive income scaled by total assets, divided by the standard deviation of net
income scaled by total assets, over the initial comprehensive income reporting year and the prior 4 years;
Complexity_OCI captures the complexity of OCI, defined as the number of OCI components the firm reports on its
10-K filing; Abs_TAFS (Abs_TCurrency, Abs_TDerivatives, Abs_TPension, and Abs_TOther) are the ratio of the absolute
value of TAFS (TCurrency, TDerivatives, TPension, and TOther) over the total asset.

magnitude of adjustments for other adjustments (Abs_TOther) are more likely to report
OCI in the SCI.
Table 4 reports the results of the second-stage regression with the IMR as in Models 4
and 5. The results presented in Table 4 support those in Table 2, in that OCI is value rele-
vant when it is only reported in the SSE. In fact, we find that both the coefficients and sta-
tistical significance of the interaction terms of DSE and OCI remain similar (0.509,
t statistic = 3.32) after controlling for IMR. This finding confirms that investors tend to
limit their attention to OCI when it is reported in the SSE, a predominant reporting location
prior to ASU 2011-05. The above finding, however, could be driven by the fact that only
about 19% of our sample firms reported OCI in the SCI during 2000-2006.
To further extend prior research and examine whether recent financial crisis and the pro-
hibition on reporting OCI in the SSE as required by the ASU 2011-05 affect how investors
value OCI, we add two sub-periods to include more recent data: the financial crisis period
(2007-2009) and the adoption of ASU 2011-05 period (2010-2012).
Table 5 reports the descriptive statistics of our key variables in three sub-periods. The
mean and median RET for both the pre- financial crisis and the ASU 2011-05 adoption
periods are generally above 0.12, but the mean and median RET for the financial crisis

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 13

Table 4. Second-Stage Regression Controlling for Self-Selection.


RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a3 DIS3OCIit + a4 IMR + eit : ð4Þ
RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit + a5 DSE3TDerivativesit
+ a6 DSE3TPensionit + a7 DSE3TOtherit + a8 DIS3TCurrencyi t + a9 DIS3TAFSit
+ a10 DIS3TDerivativesit + a11 DIS3TPensionit + a12 DIS3TOtherit + a13 IMR + eit : ð5Þ

Variable Coefficient t statistic Coefficient t statistic


Intercept 20.004 20.16 20.006 20.23
NI 1.602 8.34 1.576 8.45
Loss 3 NI 21.516 27.52 21.491 27.61
DSE 3 OCI 0.509 3.32
DIS 3 OCI 0.176 0.73
DSE 3 TCurrency 0.767 2.52
DSE 3 TAFS 1.495 1.75
DSE 3 TDerivatives 21.091 21.56
DSE 3 TPension 0.464 2.07
DSE 3 TOther 20.033 20.03
DIS 3 TCurrency 1.755 1.81
DIS 3 TAFS 0.515 1.27
DIS 3 TDerivatives 21.872 23.24
DIS 3 TPension 0.316 0.92
DIS 3 TOther 20.462 20.79
IMR 20.0326 21.95 20.031 21.87
Adjusted R2 .095 .102
Observations 2,639 2,639

Note. See the appendix for variable definitions. IMR is the inverse Mills ratio obtained from the first-stage
regression as in Model 3. The t statistics are based on White’s (1980) heteroskedasticity-adjusted robust variance
estimates.

period are both negative (20.006 and 20.059, respectively). The t test shows that RET is
significantly lower in the financial crisis period. Both the pre-financial crisis and adoption
periods have higher mean NI than the financial crisis period. The t test also suggests that
NI is significantly lower in the financial crisis period. Although all the means and medians
of aggregate OCI and its components are very close to 0 across three sub-periods, the mean
adjustments for foreign currency translation, available-for-sale investments, and pension
liabilities appear to be more significant than other adjustments. The mean adjustments for
pension liabilities are consistently negative across three sub-periods. More importantly, we
find that the financial crisis period has significantly higher mean and median relative 3- or
5-year volatility of OCI versus NI (VOL_OCI) than both the pre-financial crisis and the
adoption sub-periods. The t test shows that both volatility measures are significantly higher
in the financial crisis period compared with the pre-financial crisis. We also find that the
financial crisis period has higher absolute values of the adjustments for foreign currency
translation (Abs_TCurrency), available-for-sale investments (Abs_TAFS), derivatives
(Abs_TDerivatives), and pension liabilities (Abs_TPension) than both the pre-financial
crisis and adoption periods. The above findings suggest that firms experienced more signifi-
cant market fluctuations in fair values of certain long-term financial assets and liabilities
during recent financial crisis. Consistent with Table 1, sample firms report approximately
three OCI components (Complexity_OCI) across all the sub-periods.

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


14
Table 5. Descriptive Statistics of the 2000-2012 Sample and Sub-samples.

2000-2012 (n = 4,669) 2000-2006 (n = 2,639) 2007-2009 (n = 964) 2010-2012 (n = 1,066)

Variable M Median SD M Median SD M Median SD M Median SD t statistica t statisticb


RET 0.099 0.104 0.299 0.126 0.121 0.269 20.006 20.059 0.395 0.136 0.140 0.231 211.58 210.09
NI 0.046 0.056 0.114 0.045 0.053 0.123 0.035 0.056 0.117 0.057 0.063 0.082 22.32 25.11
OCI 20.001 0.000 0.051 0.000 0.000 0.037 20.001 0.001 0.089 20.002 20.001 0.026 20.52 0.32
TCurrency 0.002 0.000 0.019 0.003 0.000 0.018 0.002 0.000 0.025 20.001 0.000 0.010 21.48 2.39
TAFS 0.002 0.000 0.035 0.000 0.000 0.011 0.003 0.000 0.070 0.003 0.000 0.018 1.71 20.04
TDerivatives 0.000 0.000 0.010 0.000 0.000 0.011 0.000 0.000 0.013 0.000 0.000 0.005 0.64 0.20
TPension 20.004 0.000 0.024 20.003 0.000 0.026 20.005 0.000 0.028 20.005 20.001 0.014 22.72 20.93
TOther 0.000 0.000 0.015 0.000 0.000 0.006 0.000 0.000 0.029 0.000 0.000 0.005 20.02 20.18
VOL_OCI_3yr 1.747 0.524 5.464 1.362 0.414 3.428 2.457 0.873 5.679 2.003 0.595 8.452 7.15 1.45
VOL_OCI_5yr 1.096 0.445 2.108 0.911 0.375 1.792 1.279 0.587 2.245 1.378 0.627 2.595 5.22 20.94
Complexity_OCI 2.938 3.000 1.040 2.916 3.000 1.101 3.020 3.000 0.939 2.915 3.000 0.974 2.71 2.53
Abs_TAFS 0.002 0.000 0.006 0.002 0.000 0.006 0.002 0.000 0.007 0.001 0.000 0.004 1.92 5.03
Abs_Tcurrency 0.006 0.002 0.010 0.006 0.002 0.009 0.010 0.003 0.014 0.004 0.001 0.007 9.48 12.34
Abs_Tderivatives 0.001 0.000 0.005 0.001 0.000 0.005 0.002 0.000 0.006 0.001 0.000 0.004 1.66 2.80
Abs_Tpension 0.006 0.001 0.016 0.005 0.000 0.016 0.009 0.002 0.020 0.005 0.001 0.009 7.52 7.08

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Abs_Tother 0.001 0.000 0.004 0.001 0.000 0.004 0.001 0.000 0.004 0.000 0.000 0.002 0.34 2.94
SIZE 9.357 9.232 1.104 9.237 9.095 1.134 9.529 9.463 1.076 9.485 9.352 1.016 7.16 0.97
LEVE 0.241 0.226 0.169 0.242 0.228 0.170 0.243 0.228 0.170 0.239 0.222 0.165 0.12 0.47
ROA 0.057 0.051 0.087 0.055 0.048 0.096 0.052 0.052 0.083 0.066 0.056 0.063 20.80 24.19
a
For 2007-2009 versus 2000-2006 periods.
b
For 2007-2009 versus 2010-2012 periods.
Lin et al. 15

Table 6. Value Relevance of OCI Conditional on Reporting Location: Evidence From the 2000-2012
Sample.
RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a4 DIS3OCIit + eit : ð1Þ
RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit + a5 DSE3TDerivativesit
+ a6 DSE3TPensionit + a7 DSE3TOtherit + a8 DIS3TCurrencyi t + a9 DIS3TAFSit
+ a10 DIS3TDerivativesit + a11 DIS3TPensionit + a12 DIS3TOtherit + eit : ð2Þ

Variable Coefficient t statistic Coefficient t statistic


Intercept 20.045 211.56 20.044 211.37
NI 1.300 12.71 1.293 12.85
LOSS 3 NI 21.176 29.92 21.174 210.15
DSE 3 OCI 0.671 6.54
DIS 3 OCI 0.599 3.96
DSE 3 TCurrency 0.733 2.96
DSE 3 TAFS 1.113 6.07
DSE 3 TDerivatives 20.645 21.35
DSE 3 TPension 0.546 3.02
DSE 3 TOther 0.460 2.60
DIS 3 TCurrency 1.561 2.32
DIS 3 TAFS 0.608 6.04
DIS 3 TDerivatives 20.899 21.60
DIS 3 TPension 0.148 0.47
DIS 3 TOther 1.065 1.21
Adjusted R2 .0727 .0792
Observations 4,669 4,669

Note. We use the sample for the period 2000-2012 to statistically evaluate whether the reporting location affects
investors’ pricing of OCI and its components. See the appendix for variable definitions. The t statistics are based on
White’s (1980) heteroskedasticity-adjusted robust variance estimates. OCI = other comprehensive income.

Table 6 reports the regression results for the period 2000-2012. In the aggregate OCI
model, we find that the coefficient on the interaction term DSE 3 OCI is positive and sig-
nificant at less than 1% level (0.671, t statistic = 6.54), and the coefficient on the interac-
tion term DIS 3 OCI is also positive and significant at less than 1% level (0.599, t statistic
= 3.96). The magnitude of the coefficients on both interaction terms significantly increases,
although both coefficients are not statistically different. Results for the OCI component
model show that adjustments for foreign currency translation (0.733, t statistic = 2.96),
available-for-sale investments (1.113, t statistic = 6.07), pension liabilities (0.546, t statistic
= 3.02), and others (0.460, t statistic = 2.60) reported in the SSE are positively associated
with returns, significant at the 5% level or better. Among the OCI components reported in
the SCI, adjustments for foreign currency translation (1.561, t statistic = 2.32) and avail-
able-for-sale investments (0.608, t statistic = 6.04) are also positively associated with
returns; adjustments for derivatives are negatively and marginally associated with returns
(20.899, t statistic = 1.60). Inconsistent with Table 2, we find that investors priced aggre-
gate OCI when reported in both the SSE and SCI during 2000-2012.
We further investigate whether the results reported in Table 6 are driven by any sub-
periods. Table 7 shows that investors consistently price OCI when it is reported in the SSE
in all three sub-periods. However, investors appear to price OCI reported in the SCI only in
the financial crisis period (0.745, t statistic = 6.47). Together with Table 5, we find that

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


16 Journal of Accounting, Auditing & Finance

Table 7. Intertemporal Test Using the 2000-2012 Sample.


RETit = g1 NIit + g2 LOSS3NIit + g3 PRE CRISIS3DSE3OCIit + g4 PRE CRISIS3DIS3OCIit
+ g5 CRISIS3DSE3OCIit + g6 CRISIS3DIS3OCIit + g7 POST CRISIS3DSE3OCIit
+ g8 POST CRISIS3DIS3OCIit + g9 CRISIS + g10 POST CRISIS + eit : ð6Þ

Variable Coefficient t statistic


Intercept 20.047 29.61
NI 1.353 12.46
LOSS 3 NI 21.236 29.97
PRE_CRISIS 3 DSE 3 OCI 0.521 3.40
PRE_CRISIS 3 DIS 3 OCI 0.165 0.67
CRISIS 3 DSE 3 OCI 0.737 4.97
CRISIS 3 DIS 3 OCI 0.745 6.47
POST_CRISIS 3 DSE 3 OCI 0.772 2.01
POST_CRISIS 3 DIS 3 OCI 0.100 0.31
CRISIS 20.019 22.13
POST_CRISIS 0.023 2.86
Adjusted R2 .0772
Observations 4,669

Note. PRE_CRISIS is a dummy variable equal to 1 for the pre-financial crisis period (2000-2006) and 0 otherwise.
CRISIS is a dummy variable equal to 1 for the financial crisis period (2007-2009) and 0 otherwise. POST_CRISIS is a
dummy variable equal to 1 for the ASU 2011-05 adoption period (2010-2012) and 0 otherwise. See the appendix
for other variable definitions. The t statistics are based on White’s (1980) heteroskedasticity-adjusted robust
variance estimates.

investors priced OCI reported in the SCI only when the magnitude and volatility of OCI
significantly increased during recent financial crisis. This suggests that investors perceive
significant and more volatile OCI during financial crisis to adversely affect firms’ future
earnings.
Some firms reported OCI in both the SSE and SCI. Further analysis finds that most
firms reported aggregate OCI in the SSE and the components of OCI in the SCI if these
firms chose to report OCI in the SCI. As a result, a significant coefficient on DIS 3 OCI
indicates that OCI reported in both the SCI and SSE is valued by investors (relative to
firms that only report OCI in the SSE). Although it is impossible to completely separate the
effect of the components of OCI reported in the SCI from the effect of aggregate OCI
reported in the SSE using our research design, we believe that the coefficient on DIS 3
OCI should capture the valuation effect of reporting OCI in the SCI. This is because these
firms voluntarily chose to report detailed OCI in the SCI, while aggregated OCI is reported
in the SSE as part of change in shareholders’ equity. As the coefficient on DSI 3 OCI is
not significant during 2000-2006 and 2010-2012, which implies that detailed OCI reported
in the SCI does not contain important information compared with detailed OCI reported in
the SSE during these two periods.
Because of the above limitation, we believe that the valuation effect of change in report-
ing location of OCI from SSE to SCI following ASU 2011-05 provides an important insight
on how investors value OCI when reported in the SCI. Our analysis shows that 330 U.S.
firms changed the location of reporting OCI from SSE to SCI due to the introduction of
ASU 2011-05: 62 firms in 2011 and 268 firms in 2012. We removed firms that voluntarily
adopted ASU 2011-05 in 201110 to focus on the effect of mandatory adoption. We then

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 17

Table 8. A Difference-in-Differences Test Using the 2010-2012 Sample.

Variable Coefficient t statistic Coefficient t statistic


Intercept 20.004 20.26 20.005 20.39
NI 0.975 4.31 1.067 4.58
LOSS 3 NI 20.447 21.76 20.559 22.06
TREAT 20.026 21.59 20.025 21.54
POST 20.022 21.05 20.029 21.41
TREAT 3 POST 0.035 1.33 0.042 1.60
OCI 20.806 21.82
POST 3 OCI 1.860 3.47
TREAT 3 OCI 0.925 1.60
TREAT 3 POST 3 OCI 22.061 22.91
TCurrency 0.332 0.58
TAFS 21.243 23.55
TDerivatives 0.996 0.52
TPension 1.124 1.79
TOther 20.091 20.14
POST 3 TCurrency 4.369 1.59
POST 3 TAFS 2.404 4.83
POST 3 TDerivatives 25.967 21.54
POST 3 TPension 20.483 20.52
POST 3 TOther 0.990 1.35
TREAT 3 POST 3 TCurrency 26.545 22.03
TREAT 3 POST 3 TAFS 21.200 22.35
TREAT 3 POST 3 TDerivatives 6.143 0.89
TREAT 3 POST 3 TPension 20.812 20.96
TREAT 3 POST 3 TOther 21.210 22.70
Adjusted R2 .0823 .0936
Observations 1,014 1,014

Note. This table reports the results of whether changing the reporting location of OCI affects the value relevance of
OCI and its components. TREAT is a dummy variable equal to 1 when the firm changed the presentation of OCI
from SSE to SCI (treatment firms), and 0 if the firm continued to report OCI in the SCI throughout the test period
(control firms). For the treatment firms, POST is a dummy variable that takes the value of 1 if the fiscal year begins
after December 15, 2011, and 0 otherwise. See the appendix for other variable definitions. The t statistics are
based on White’s (1980) heteroskedasticity-adjusted robust variance estimates.

perform a difference-in-differences regression to investigate change in value relevance of


OCI for those firms that switched the reporting location of OCI to SCI from SSE as com-
pared with those firms that continued to report OCI in the SCI. As shown in Model 7, our
variable of interest is the three-way interaction term TREAT 3 POST 3 OCI.
Table 8 shows that POST 3 OCI is positive and significant at the 1% level (1.860, t sta-
tistic = 3.47), indicating that overall investors value OCI more in the adoption period. Our
variable of interest TREAT 3 POST 3 OCI, however, is negative and significant at the 1%
level (t statistic = 22.91). This indicates that investors actually value OCI less when firms
switched OCI from SSE to SCI compared with those firms that continued to report OCI in
the SCI. We also find that the results for OCI components are interesting. First, compared
with Table 5, adjustments for foreign currency translation (4.369, t statistic = 1.59) and
derivatives (25.967, t statistic = 21.54) are only weakly associated with returns in the
adoption period. Second, the interaction terms of TREAT 3 POST 3 TCurrency (26.545,
t statistic = 22.03), TREAT 3 POST 3 TAFS (21.200, t statistic = 22.35), and TREAT 3

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


18 Journal of Accounting, Auditing & Finance

POST 3 TOther (21.210, t statistic = 22.70) are all negative and significant at the 5%
level or higher. Together, we find that value relevance of OCI significantly reduces after
firms changed the reporting location of OCI to SCI from SSE. Our finding is generally
inconsistent with the expectations of FASB and IASB that reporting OCI in the SCI should
provide investors more transparent information and increase value relevance of OCI. This
finding also contradicts the findings of prior experimental research that investors consider
OCI in their decision making when it is reported in a more transparent performance state-
ment. A potential reason for this finding is that reporting OCI in the SCI may significantly
increase volatility and reduces the predictive value of accounting income, which adversely
affects value relevance of OCI. This result is, however, consistent with our previous find-
ings and those of Chambers et al. (2007) that investors consistently price OCI when it is
reported in the SSE.11

Additional Analyses
The investor sophistication, investment horizon, and the relation between reporting loca-
tion and value relevance of OCI. A recent experimental study by Elliott et al. (2010) exam-
ines the joint influence of accounting transparency and investor base on analysts’
expectation of mispricing of OCI. They find that analysts expect a larger overpricing for
firms reporting OCI in the SCI when the firm’s most important investors are transient
rather than dedicated investors.
Results from Elliott et al. (2010) indicate that analysts expect transient investors to over-
look the transitory nature of OCI and to value OCI when it is reported in a more transpar-
ent location. As a sensitivity test, we investigate whether our finding that OCI reported in
the SCI became value relevant during recent financial crisis was due to transient investors’
overreaction to the information. Compared with dedicated investors that have a long-term
horizon and are geared up for long-term capital, transient investors more likely adjust their
trading strategies based on near-term factors (Bushee, 2001; Elliott et al., 2010). We test
whether OCI reported in the SCI was more value relevant when a firm’s most important
investors were transient institutions during 2007-2009.
We perform our test using the following models:

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3OCIit + a4 DIS3OCIit + a5 TIHit


ð9Þ
+ a6 TIH3DSE3OCIit + a7 TIH3DIS3OCIit + eit ,

RETit = a1 NIit + a2 LOSS3NIit + a3 DSE3TCurrencyi t + a4 DSE3TAFSit


+ a5 DSE3TDerivativesit + a6 DSE3TPensionit + a7 DSE3TOtherit
+ a8 DIS3TCurrencyi t + a9 DIS3TAFSit + a10 DIS3TDerivativesit
+ a11 DIS3TPensionit + a12 DIS3TOtherit + a13 TIH
+ a14 TIH3DSE3TCurrencyi t + a15 TIH3DSE3TAFSit ð10Þ
+ a16 TIH3DSE3TDerivativesit + a17 TIH3DSE3TPensionit
+ a18 TIH3DSE3TOtherit + a19 TIH3DIS3TCurrencyi t
+ a20 TIH3DIS3TAFSit + a21 TIH3DIS3TDerivativesit
+ a22 TIH3DIS3TPensionit + a23 TIH3DIS3TOtherit + eit ,

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 19

Table 9. Value Relevance of OCI Conditional on Reporting Location and Investor Base: 2007 to
2009.

Variable Coefficient t statistic Coefficient t statistic


Intercept 20.044 23.37 20.043 23.30
NI 0.531 2.36 0.606 2.79
LOSS 3 NI 20.443 21.55 20.461 21.72
DSE 3 OCI 0.701 3.80
DIS 3 OCI 0.644 1.73
DSE 3 TCurrency 0.494 0.85
DSE 3 TAFS 1.152 4.81
DSE 3 TDerivatives 0.568 0.95
DSE 3 TPension 20.407 20.57
DSE 3 TOther 0.578 3.82
DIS 3 TCurrency 2.419 1.69
DIS 3 TAFS 0.770 2.10
DIS 3 TDerivatives 20.341 20.08
DIS 3 TPension 21.740 21.32
DIS 3 TOther 5.539 1.09
TIH 0.012 0.78 0.016 1.02
TIH 3 DSE 3 OCI 0.182 0.63
TIH 3 DIS 3 OCI 0.137 0.36
TIH 3 DSE 3 TCurrency 0.481 0.60
TIH 3 DSE 3 TAFS 0.042 0.08
TIH 3 DSE 3 TDerivatives 21.483 20.74
TIH 3 DSE 3 TPension 1.008 1.25
TIH 3 DSE 3 TOther 20.361 20.16
TIH 3 DIS 3 TCurrency 24.425 21.82
TIH 3 DIS 3 TAFS 20.420 20.99
TIH 3 DIS 3 TDerivatives 25.635 21.08
TIH 3 DIS 3 TPension 2.051 0.80
TIH 3 DIS 3 TOther 23.184 1.76
Adjusted R2 .0764 .1044
Observations 926 926

Note. This table presents the results of whether investor base and reporting location of OCI affect the value
relevance of OCI and its components. TIH is a dummy variable that takes the value of 1 if the percentage of
transient institutional ownership is above the sample median, 0 otherwise. The t statistics are based on White’s
(1980) heteroskedasticity-adjusted robust variance estimates. OCI = other comprehensive income.

where TIH is a dummy variable that takes the value of 1 if the percentage of transient insti-
tutional ownership is above the sample median, 0 otherwise. The classification of transient
investors is based on the one used in Bushee (2001).12 All the other variables are previ-
ously defined and summarized in the appendix.
Results reported in Table 9 show that the coefficients on TIH 3 DSE 3 OCI and TIH 3
DIS 3 OCI are insignificant, suggesting that transient investors do not affect value relevance
of OCI, no matter where they are reported. It appears that the investment bias of transient
investors does not drive our finding that investors value OCI reported in the SCI during the
financial crisis period when the magnitude and volatility of OCI are more significant.

The magnitude of OCI and the relation between reporting location and value relevance of
OCI. We find that investors valued OCI reported in the SCI during the recent financial

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


20 Journal of Accounting, Auditing & Finance

crisis period when the magnitude and volatility of OCI significantly increased, which implies
that the importance of OCI in a firm’s operation could matter. To further investigate this issue,
we partition our sample by the magnitude of OCI, measured by both reported OCI and the
absolute value of OCI. The results (untabulated) indicate that OCI reported in the SCI is not
value relevant for firms with relatively large magnitude of OCI during both the pre- and post-
financial crisis periods. However, we find that OCI reported in the SCI is value relevant during
the financial crisis period. This finding is consistent with the notion that investors only price
OCI reported in the SCI when it is significant and more volatile during poor economic condi-
tion. One potential reason for this finding is that investors do not perceive OCI reported in the
SCI to affect firms’ future earnings because OCI is generally transitory and unpredictable, and
can be reversed in future years (Jones & Smith, 2011). However, investors may value OCI
reported in the SCI when they perceive OCI to significantly affect firms’ future earnings when
the magnitude and volatility of OCI become significant during financial crisis (Huang et al.,
2016). In contrast, we find that OCI reported in the SSE is consistently value relevant for firms
with relatively large magnitude of OCI during each of the sub-periods.

Conclusion
This study examines whether the reporting location of OCI affects value relevance of OCI.
Using manually collected data during 2000-2012, we find that investors consistently value
OCI when reported in the SSE. Investors appear to value OCI reported in the SCI only in
the financial crisis period when the magnitude and volatility of OCI are more significant.
We also examine whether transient investors affect the association between reporting loca-
tion and value relevance of OCI during recent financial crisis, and find that the potential
investment bias of transient investors does not affect this result.
Using the unique setting of mandatory adoption of ASU 2011-05, we examine whether
investors value OCI more after firms switched the reporting location of OCI from SSE to SCI.
We find that firms that switched experience a significant reduction in value relevance of OCI
compared with firms that continued to report OCI in the SSE. This is consistent with the finding
that the market prices OCI when only reported in the SSE, but is inconsistent with the notion
that investors should be more capable of incorporating information reported in a more transpar-
ent location into prices. Our results are robust after controlling for sample selection bias.
Our results suggest that although reporting OCI in the SCI increases information transpar-
ency, investors do not perceive OCI to affect firms’ future earnings and therefore do not value
OCI when reported in the SCI. Our results also suggest that investors value OCI reported in
the SCI only when the magnitude and volatility of OCI are important. Finally, reporting OCI
in the SCI may adversely affect value relevance of OCI because it increases volatility and
reduces the predictive value of accounting income. Our findings are timely and should be of
interest to regulators, researchers, practitioners, and users of accounting information.
Our study is not without limitations. First, we do not rule out the possibility that inves-
tors value OCI only when it is reported in the SSE simply because the SSE has been its pre-
dominant reporting location. Investors are likely to grow accustomed to new reporting
practices. Thus, our results should be viewed as preliminary because our sample period
ends in the first adoption year of ASU 2011-05 (i.e., 2012). Future research may investigate
how investors value OCI following ASU 2011-05 when SCI becomes a prevailing reporting
location of OCI. Second, it is also possible that the models used in this study are not pow-
erful enough to capture the effect of the change in OCI reporting location on value rele-
vance of OCI. For example, our research design cannot sufficiently separate the effect of

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 21

reporting detailed OCI in the SCI from the effect of reporting aggregate OCI in the SSE
when firms report OCI both in the SCI and in the SSE. Finally, this study does not examine
whether change in OCI reporting location has any real effect on investors’ and other finan-
cial statement users’ decision making.

Appendix Variable Definitions.

Variable Definitions
RET Buy-and-hold raw return for the window 8 months before to 4 months after the
fiscal year-end
NI The net income after extraordinary items and discontinued operations, scaled by
the market value 8 months before fiscal year-end
LOSS A dummy variable that equals 1 if a firm reports net loss and 0 otherwise;
OCI The value reported as ‘‘other comprehensive income,’’ scaled by the market value
8 months before fiscal year-end
DIS A dummy variable that takes the value of 1 when OCI is reported in the SCI, and
0 otherwise
DSE A dummy variable that takes the value of 1 when OCI is reported in the SSE, and
0 otherwise
TCurrency The unrealized gains and losses from foreign currency translation, scaled by the
market value 8 months before fiscal year-end
TAFS The unrealized gains and losses from available-for-sale securities, scaled by the
market value 8 months before fiscal year-end
TDerivatives The unrealized gains and losses from derivative securities and hedging activities,
scaled by the market value 8 months before fiscal year-end
TPension The unrealized gains and losses from defined benefit plans, scaled by the market
value 8 months before fiscal year-end
TOther The unrealized gains and losses resulting from other items, scaled by the market
value 8 months before fiscal year-end
SIZE The log of the market value of the firm’s common shares outstanding
LEVE Long-term debt scaled by total assets
ROA Net income before extraordinary items, scaled by total assets
VOL_OCI The standard deviation of comprehensive income scaled by total assets, divided
by the standard deviation of net income scaled by total assets, measured over
the initial comprehensive income reporting year and the prior 4 years
Complexity_OCI The number of OCI components the firm reported on its10-K filing
Abs_TAFS The ratio of absolute value of TAFS over the total assets
Abs_TCurrency The ratio of absolute value of TCurrency over the total assets
Abs_TDerivatives The ratio of absolute value of TDerivatives over the total assets
Abs_TPension The ratio of absolute value of TPension over the total assets
Abs_TOther The ratio of absolute value of TOther over the total assets
PRE_CRISIS An indicator variable equal to 1 during the pre-financial crisis period (2000-2006)
and 0 otherwise
CRISIS An indicator variable equal to 1 during the financial crisis period (2007-2009) and
0 otherwise
POST_CRISIS An indicator variable equal to 1 for the ASU 2011-05 adoption period (2010-
2012) and 0 otherwise
TREAT An indicator variable that takes the value of 1 if the firm switched the reporting
location of OCI from SSE to SCI, 0 if the firm did not switch the reporting
location of OCI
POST An indicator variable that takes value of 1 if a firm’s fiscal year begins after
December 15, 2011, and 0 if a firm’s fiscal year begins before December 15,
2011

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


22 Journal of Accounting, Auditing & Finance

Acknowledgments
The authors appreciate Agnes Chen, Xiaochuan (Kelly) Huang, Elio Alfonso, Andrea Kelton, Brent
Liao, Tom Linsmeier, Greg McPhee, Phil Shane, Clark Wheatley, and all the workshop participants
at Hong Kong Polytechnic University, Florida International University, and Wake Forest University
for their valuable comments. Finally, the authors thank both anonymous reviewers of Journal of
Accounting, Auditing, and Finance for their numerous detailed and constructive comments.

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article.
Notes
1. The main components include unrealized gains or losses from foreign currency translation, avail-
able-for-sale securities, derivative securities and hedging activities, and accrued pension
liabilities.
2. This study uses statement of comprehensive income (SCI) throughout the article to refer to firms
reporting other comprehensive income (OCI) in a combined statement of income and comprehen-
sive income or a separate statement of comprehensive income, although approximately 88% of
these firms reported OCI in a separate statement of comprehensive income during our test period.
3. A combined statement of comprehensive income is equivalent to a statement of financial perfor-
mance, which is the term used in International Accounting Standard (IAS) 1 and prior research.
4. For example, during recent financial crisis (2007-2009) many firms recognized large unrealized
losses from their available-for-sale investments in OCI. Economist Brian Wesbury believes that
‘‘Mark-to-market accounting rules have turned a large problem into a humongous one. A vast
majority of mortgages, corporate bonds, and structured debts are still performing. But because
the market is frozen, the prices of these assets have fallen below their true value.’’ He further
states that many banks were pushed toward insolvency and forced to sell their assets at fire-sale
prices, which then caused values to fall even further during recent financial crisis (Pozen, 2009).
5. The post-financial crisis period is also the adoption period of ASU 2011-05 in this study.
6. We collected the data back to 1998. The requirement of calculating the volatility of OCI with at
least 3 years’ data restricts the sample to 2000-2012.
7. We deleted observations with the absolute value of RSTUDENT larger than 2 or DFFITS static
larger than 2. Large values of DFFITS indicate influential observations.
8. We also replicated Chambers, Linsmeier, Shakespeare, and Sougiannis (2007) using data for the
1998-2003 period. Results are consistent with those for the 2000-2006 period.
9. We also change the definition of treatment firms to firms that switched the reporting location of
OCI during the 3-year period. As a result, we change the definition of both TREAT and POST.
TREAT is an indicator variable that takes the value of 1 when the firm changed the presentation
of OCI from SSE to SCI during 2010-2012, and 0 if the firm continued to report OCI in the SCI.
POST is an indicator variable: For the treatment firms, it takes the value of 1 after the firm chan-
ged the reporting location, and 0 for the period before the switch; for the control firms, it takes
the value of 1 if the fiscal year begins after December 15, 2011, and 0 if a firm’s fiscal year
begins before December 15, 2011. There are 1,052 observations, with 871 firm-year observations
in the treatment group and 181 firm-year observations in the control group. We find that the
results are generally consistent.

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016


Lin et al. 23

10. We also perform the robustness test, including the firms that adopted ASU 2011-05 in 2011. The
results are qualitatively similar.
11. Untabulated results show that our findings and conclusion apply to both financial and non finan-
cial firms.
12. We obtained the institutional investor classification data from Bushee’s website and merged the
classifications with the Thomson Reuters institutional holdings (13F) data. We then calculated
the average percentage of shares owned by transient investors during the year for each company
in our sample.

References
Bamber, L. S., Jiang, J., Petroni, K. R., & Wang, I. Y. (2010). Comprehensive income: Who’s afraid
of performance reporting? The Accounting Review, 85, 97-126.
Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: Identifying influential data
and sources of collinearity. New York, NY: John Wiley.
Biddle, G., & Choi, J. H. (2006). Is comprehensive income useful? Journal of Contemporary
Accounting & Economics, 2, 1-32.
Bushee, B. (2001). Do institutional investors prefer near-term earnings over long-run value?
Contemporary Accounting Research, 18, 207-246.
Chambers, D., Linsmeier, T. J., Shakespeare, C., & Sougiannis, T. (2007). An evaluation of SFAS
No. 130 comprehensive income disclosures. Review of Accounting Studies, 12, 557-593.
Dhaliwal, D., Subramanyam, K., & Trezevant, R. (1999). Is comprehensive income superior to net
income as a measure of firm performance? Journal of Accounting & Economics, 26, 43-67.
Elliott, W. B., Krische, S., & Peecher, M. (2010). Expected mispricing: The joint influence of
accounting transparency and investor base. Journal of Accounting Research, 48, 343-381.
Financial Accounting Standards Board. (2011). Accounting standards update: Comprehensive income
(Topic 220). Norwalk, CT: Author.
Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153-161.
Hirst, D., & Hopkins, P. (1998). Comprehensive income reporting and analysts’ valuation judgments.
Journal of Accounting Research, 36, 47-74.
Huang, W., Lin, S., & Raghunandan, K. (2016). The volatility of other comprehensive income and
audit fees. Accounting Horizons, 30, 195-210.
Jones, D. A., & Smith, K. J. (2011). Comparing the value relevance, predictive value, and persistence
of other comprehensive income and special items. The Accounting Review, 86, 2047-2074.
Kanagaretnam, K., Mathieu, R., & Shehata, M. (2009). Usefulness of comprehensive income report-
ing in Canada. Journal of Accounting and Public Policy, 28, 349-365.
Lee, Y., Petroni, K., & Shen, M. (2006). Cherry picking, financial reporting quality, and comprehen-
sive income reporting choices: The case of property-liability insurers. Contemporary Accounting
Research, 23, 665-700.
Maines, L., & McDaniel, L. (2000). Effects of comprehensive-income characteristics on nonprofes-
sional investors’ judgments: The role of financial statement presentation format. The Accounting
Review, 75, 177-204.
Pozen, R. C. (2009). Is it fair to blame fair value accounting for the financial crisis? Harvard
Business Review, 87(11), 84-92.
Rees, L. L., & Shane, P. B. (2012). Academic research and standard-setting: The case of other com-
prehensive income. Accounting Horizons, 26, 789-815.
Schaberl, P., & Victoravich, L. M. (2015). Reporting location and the value relevance of accounting
information: The case of other comprehensive income. Advances in Accounting, 31, 239-246.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroscedasticity. Econometrica, 48, 817-838.

Downloaded from jaf.sagepub.com at CORNELL UNIV on October 8, 2016