Does control-based approach to consolidated statements reflect market price
better than ownership-based approach?
Audrey Wen-hsin Hsu (Contact author)
Department of Accounting
National Taiwan University
No. 1, Sec. 4, Roosevelt Road
Taipei, Taiwan 106
Phone: +886-2-3366-1131
Email: Audrey.hsu@management.ntu.edu.tw
Rong-Ruey Duh
Department of Accounting
National Taiwan University
No. 1, Sec. 4, Roosevelt Road
Taipei, Taiwan 106
Email: rrduh@ntu.edu.tw
Kang Cheng
Morgan State University
1700 East Cold Spring Lane, Baltimore MD 21251
Email: cheng_professor@hotmail.com
This version: April 30, 2010
Acknowledgements: We appreciate the useful comments and suggestions of Victoria Wang, Ann
Chang and Yu Cong. We also thank Shupei Chang for assistance in collecting consolidation data.
The editing assistance by Malcolm R. Mayfield is gratefully acknowledged.
1
Does control-based approach to consolidated statements reflect market price
better than ownership-based approach?
Abstract
Motivated by the FASB and IASB’s recent Discussion Paper (DP) on how to define reporting
entities, this study investigates the value relevance of consolidated statements under US ARB
51’s ownership-based approach and IAS 27’s control-based approach. The results show that
consolidated financial statements based on a broader definition of control can provide more
useful accounting information than those based only on majority-ownership control. We also
address one concern raised in the DP, namely whether a reporting entity should use the common
control model to include entities that are under common control of an individual investor or
family. The results suggest that accounting standard boards should include the common control
model in defining the group reporting entity for firms with complex ownership structures.
Keywords: Consolidated statements, control-based approach, reporting entity, ownership-based
approach, ownership structure, ARB 51, IAS 27
2
Does control-based approach to consolidated statements reflect market price
better than ownership-based approach?
1. Introduction
The objective of this study is to evaluate whether consolidated financial statements under
control-based approach exhibit higher value relevance than statements under ownership-based
approach. On May 29, 2008, the International Accounting Standard Board (IASB) issued a
Discussion Paper (DP), Preliminary Views on an Improved Conceptual Framework for Financial
Reporting: The Reporting Entity. This DP was jointly developed by the Financial Accounting
Standard Board (FASB) and IASB as part of deliberations on the reporting entity concept for the
two boards’ common conceptual framework. In the DP, the two boards consider whether they
should define the composition of a group reporting entity based on a broader definition of control
than ownership.1 They propose defining control based on a “controlling entity model.”2
Specifically, if a group reporting entity comprises the controlling entity (the parent) and its
controlled entities (i.e. subsidiaries), majority ownership (i.e. over 50%) is not a necessary
condition to attain control. Instead, the controlling entity model defines control as the parent’s
power over another entity and the ability to obtain benefits (or to reduce the incidence of losses).
To obtain feedback on these proposals, the FASB and IASB have designed a series of questions
and called for comments.
As an alternative way to respond to the “call for comments”, our study attempts to provide
some direct and empirical evidence using field data on whether consolidated statements based on
1
The DP raises specific questions for respondents. Q5 in the DP: Do you agree that the composition of a group
reporting entity should be based on control? If not, why? For example, if you consider that another basis should be
used, which basis do you propose and why?
2
Q6. Assuming that control is used as the basis for determining the composition of a group reporting entity, do you
agree that the controlling entity model should be used as the primary basis for determining the composition of a
group entity? If not, why?
3
different reporting entity criteria affect investors’ decisions. We use Edwards-Bell-Ohlson
valuation framework to compare the market’s assessment of consolidated statements under
control-based approach with its assessment of statements under ownership-based approach. We
use a natural setting in Taiwan to examine the issue because all listed firms in Taiwan were
required to use ownership-based approach to define group reporting entities before 2005 and
control-based approach from 2005 on. Since Taiwan’s SFAS No. 7, Consolidated Financial
Statements (“TSFAS 7”), was issued in 1985, all listed firms in Taiwan have been required to
prepare consolidated statements. As issued, TSFAS 7 was equivalent to U.S. Accounting
Research Bulletin No. 51, Consolidated Financial Statements (“ARB 51”), which defines the
reporting entity based on ownership. Under Taiwan’s project to align its standards with IAS
(currently IFRS), however, TSFAS 7 was revised to follow IAS No. 27, Consolidated Financial
Statements (“IAS 27”), effective from 2005 on.3 IAS 27 defines the reporting entity based on
control.
Using all listed firms in Taiwan from 2000 to 2008, we compare the value relevance of
consolidated statements under the IAS control-based approach (2005-2008) with that under the
ARB 51 ownership-based approach (2000-2004). IAS 27 defines group reporting entities based
on the controlling entity model, in which the main criteria for being part of a reporting entity is
that the entity has decision-making power over both financial benefits and operating decisions. In
contrast, focusing only on majority ownership, ARB 51 only mandates the consolidation of
entities in which the parent owns, directly or indirectly, over 50% of the outstanding voting
shares (paragraph 2 of ARB 51). As the “over 50% ownership” criteria is at best only one of
3
The focus of the paper is on the IAS 27 (2003) revision. Although IAS 27 has been amended as results of
amendments in other IFRSs in 2004, 2006, 2007 and revised in 2008, no change is made to the criteria of controlbased approach. Throughout the paper, reference is made to IAS 27 (2003) as TSFAS 7 was amended following IAS
27 (2003).
4
several criteria evidencing the existence of the decision-making power for a group entity, the
ownership-based approach allows the parent to exclude important accounting information on
controlled entities from the consolidated financial statements by structuring ownership of the
controlled entities below 50%. Thus, we expect that consolidation under the IAS 27 controlbased approach should make consolidated statements more transparent than ownership-based
approach. Our results show that the value relevance of consolidated statements under the controlbased approach is much higher than that under the ownership-based approach.
In addition, the DP considers whether under some circumstances accounting standards
should require the reporting entity to include some entities for which the control rights do not
rest with the parent itself but with a common owner (i.e., an individual investor or family). The
DP refers to this approach as the common control model (“CCM”).4 As the IAS 27 controlling
entity model only encompasses entities under the parent’s control, the accounting information it
provides might be inadequate under some circumstances, such as when capital providers would
like to understand the full picture of all affiliates that are commonly controlled by a family.5
Thus, if the value relevance of consolidated financial statements under the IAS 27 control-based
approach is higher than the ownership-based approach, such a difference may be less pronounced
for firms to which CCM is more applicable.
Using the existence of a pyramid structure or cross-holding structure as a proxy for the
applicability of CCM, we test whether the improvement of value relevance under the IAS 27
control-based approach is less pronounced for firms to which CCM is applicable than for other
firms. Claessens, Djankov, and Lang (2000) have claimed that ultimate owners (i.e. an individual
4
Q7. Do you agree that the common control model should be used in some circumstances only? If not, why? For
example, would you limit the composition of a group reporting entity to the controlling entity model only? Or would
you widen the use of the common control model?
5
This model can be implemented through additional disclosure supplementing the controlling entity model.
5
investor or family) of most companies in East Asia, including Taiwan, usually exercise their
control through pyramid structures and cross-holdings such that the controlling right to each
family-owned entity rests directly or indirectly with family members, and not with any other
family-owned entities. For example, a family may control seven companies, none of which
controls one another. The IAS 27 control-based approach would define seven reporting entities,
along with seven consolidated statements for this family empire. Our results show that
improvements in value relevance under the control-based approach relative to ownership-based
approach in Taiwan are less pronounced in firms with pyramid structures or cross-holding
structures, supporting the utility of the common control model for firms with complex ownership.
Our study not only provides important feedback on the inquiries raised in the DP but also
has some implications for the consolidation project between the FASB and IASB. The two
boards are in the process of jointly deliberating consolidation requirements to streamline the
application of consolidation accounting and to produce globally comparable consolidated
financial statements.6 Our study provides evidence that the controlling-entity model is a better
approach for consolidation and that the common control model is also essential for firms with
complex ownership structures. Finally, we also contribute to the literature on International
Financial Reporting Standards (IFRS). While many studies have investigated whether IFRS can
improve the quality of accounting information (Barth, Landsman, & Lang, 2008; Leuz &
Verrecchia, 2000; Leuz, 2003), their findings are based on the aggregate effect of all accounting
standards without exploring the pros/cons of a specific standard.7 This study, by focusing on a
6
According to each Board’s timetable, the FASB expects to publish an Exposure Draft on consolidation criteria by
the second quarter of 2010, while the IASB will make available a staff draft of its proposed final standard and will
also publish a request for views on the FASB proposal. The two Boards expect to jointly issue the final statement on
consolidation criteria by the third quarter of 2010.
7
Duh, Lee and Lin (2009) is one of the exceptions that explore the difference in a specific standard between the US
GAAP and IAS to evaluate possible economic consequences of allowing reversals of impairment losses.
6
specific standard – IAS 27 – may help standard-setting bodies more clearly understand the
implications of a specific policy choice.
The remainder of this paper is organized as follows. Section 2 discusses accounting
standards for consolidation requirements and develops our hypotheses. Research design and
sampling procedures are given in Section 3. Section 4 reports the empirical results, and Section 5
provides sensitivity tests. Section 6 gives the concluding remarks.
2. Prior literature and hypotheses development
2.1 Institutional background of consolidation standards
Consolidated financial statements are prepared for a group reporting entity that includes the
parent and its subsidiaries (Beams, Brozovsky, & Shoulders, 2009). While prior literature (Harris,
Lang & Möller, 1994; Niskanen, Kinnunen, & Kasanen, 1998; Abad, Laffarga, Garcia-Borbolla,
Larran, Pinero, & Garrod, 2000; Goncharov, Werner, & Zimmermann, 2009) has generally
agreed that consolidated statements are more value relevant than unconsolidated statements or
parent information, the FASB and IASB still have not reached a consensus on the criteria to
determine the entities that should be incorporated into a reporting entity or a consolidated entity.
FASB
Consolidation criteria for companies that issue voting shares (i.e. voting entities)
historically has not been a controversial issue in US GAAP. Under Accounting Research Bulletin
No. 51, Consolidated Financial Statements (ARB 51), issued in 1959, the consolidation
requirement is defined based on the ownership approach: 8
8
FASB refers to the ownership approach as the voting-interest approach. The voting-interest approach in turn refers
to the voting rights for the voting shares owned by shareholders. However, another type of voting interest can refer
to the voting-rights in the board of directors, which relates more to the control rights exercised by ultimate
7
“The usual condition for a controlling financial interest is ownership of a majority voting
interest, and, therefore, as a general rule ownership by one company, directly or indirectly,
of over fifty percent of the outstanding voting shares of another company is a condition
pointing toward consolidation.”
The dominance of the ownership-based approach is evidenced by the fact that it remained in
effect and basically unchanged for more than four decades.9
In ARB 51, the general rule of consolidation focuses on companies that issue voting shares
(i.e. voting entities), and consolidation is based on ownership above 50% of an entity's
outstanding voting shares. In practice, however, control may exist through means other than a
majority voting interest. For example, in 2001, the Enron scandal shocked the accounting
profession and forced the profession to reconsider consolidation criteria. By structuring related
entities below the 50% ownership threshold, Enron was able to use those “special purpose
entities” to take off liabilities and other “undesired” accounting information from their financial
statements and mislead the market. ARB 51 only applies to “voting entities” and does not require
consolidation of “special purpose entities” for which voting power is not meaningful in
determining control.10
IASB
shareholders as discussed in the literature (Claessens et al., 2000). Thus, to avoid confusion between the two types
of voting interests, we use “ownership-based approach” for the voting interests in the voting shares possessed.
9
ARB No. 51 was amended by SFAS No. 94, Consolidation of All Majority-Owned Subsidiaries, in 1987, and
SFAS No. 144, Accounting for the Impairment or Disposal of Long-Lived Assets, in 2001, to eliminate exceptions
to consolidation and temporary control (i.e., where the subsidiary is in legal reorganization or in bankruptcy).
However, the above-50% ownership rule was not changed.
10
In response to the scandal, in 2003 FASB issued an interpretation of ARB 51, Interpretation No. 46R (FIN 46R).
FIN 46R was issued to create a new model for consolidation and extend the boundaries of entities that must be
included for consolidation beyond what was covered in ARB 51. FIN 46R identifies a variable interest entity (VIE)
based on risks and rewards, to differentiate it from a voting entity for which majority-ownership continues to be
used for consolidation basis. FIN 46 was issued in January 2003 and revised in December 2003. Effectively, FIN
46R gives guidance on consolidation criteria for structures that are commonly referred to as special-purpose entities
or VIE. Determination of VIE can help determine who can quality as the primary beneficiary for a VIE. Based on
FIN46, only a company that qualifies as the primary beneficiary has to consolidate a VIE. To qualify as the primary
beneficiary, a firm must bear the risks and receives the rewards associated with a VIE. The fact that the primary
beneficiary does not own voting shares becomes inconsequential because control is effectively exercised through
operations and personnel decisions (Hoyle, Schaefer, & Doupnik, 2008).
8
In contrast to the US GAAP, which differentiates between voting entities (in ARB 51) and
variable interest entities (in FIN 46), the International Financial Reporting Standards (IFRS)
have a single model for consolidation: the controlling entity model. IAS No. 27, Consolidated
Financial Statements (“IAS 27”) sets group reporting criteria using control-based approach.
Control is clearly defined as “the power to govern the financial and operating policies of an
entity so as to obtain benefit from its activities”. Ownership of a majority of the voting interest is
not a necessary condition, as IAS 27 recognizes that control can be exercised through other
means.
Control is presumed to exist when the parent owns more than half of the voting power; to
this extent, it is similar to the ownership approach under US GAAP. However, in paragraph 13
of IAS 27, control also exists when the parent owns less than 50% of ownership but when there
is:
(a) power over more than half of the voting rights by virtue of an agreement with other investors;
(b) power to govern the financial and operating policies of the entity under a statute or an
agreement;
(c) power to appoint or remove the majority of the members of the board of directors or
equivalent governing body and control of the entity is by that board or body; or
(d) power to cast the majority of votes at meetings of the board of directors or equivalent
governing body and control of the entity is by that board or body.
Operating policies would include those policies that guide activities such as sales, marketing,
manufacturing, human resources and acquisitions.
9
Thus, IAS 27 considers facts or circumstances other than ownership of a majority voting
interests in determining the existence of control by a parent over its investee.11 The scope of
entities under the IAS 27 control-based approach is evidently broader than that under the ARB
51 ownership-based approach.
The unresolved issue between the two boards: control
As the two sets of consolidation standards from FASB and IASB are not entirely aligned,
the two boards are now working together to harmonize the consolidation requirements and
streamline the application of consolidation accounting. They expect to jointly issue the final
statement on the consolidation criteria by the third quarter of 2010.12
In May 2008, the two boards jointly issued a Discussion Paper (DP), “Preliminary Views on
an Improved Conceptual Framework for Financial Reporting: The Reporting Entity.” The core
issue is the definition of a reporting entity. The bottom line is how to circumscribe an area of
business activity of interest to existing and potential investors; and the key is control.
Specifically, in the DP, the two boards are seeking feedback on questions such as 1)
whether or not the composition of a group reporting entity should be based on control, 13 2)
whether or not the “controlling entity model” should be the primary basis for a group reporting
entity, 14 and 3) whether or not the “common control model” should be used in some
circumstances.15
11
For SPE, IFRS provides additional control indicators to consider when determining whether control over an SPE
exists in the Standing Interpretations Committee 12, Consolidation – Special Purpose Entities, an interpretation
relating to IAS No. 27 (SIC 12). These indicators include the concept of control described above as well as a
consideration of which party absorbs a majority of the risks or rewards. The indicators are considered in the context
of all relevant factors.
12
See note6
13
See note 1.
14
See note 2.
15
See note 4.
10
The “controlling entity model,” where the area of business activity is circumscribed by the
extent of an entity’s control over other entities, is not unfamiliar at all. In a sense, the current US
GAAP’s ownership-based approach and the IFRS’ control-based approach both fall under this
model. The main difference is the scope of control in voting entities. The concept of “common
control model” is more of a novelty. The idea is not to focus solely on the corporate controlling
party (i.e., the parent company); instead, it extends to situations where some entities are not
under control of the parent but under common control of an individual investor or family and
calls for filing combined financial statements. Regardless of model, it is made clear in the DP
that the objective of delineating reporting entities is to provide information that is useful to
existing and potential investors in making investment decisions.
2.2 Hypothesis Development
As the FASB and the IASB are considering the possibility that the reporting entity be based
on a broader definition of control than ownership alone, this study first provides evidence on
whether consolidated statements based on the IAS 27 control-based approach provide greater
value relevance than the ARB 51 ownership-based approach.
Conklin and Lecraw (1997) argue that ownership is not linked with control in a precise way,
and shareholdings may not reflect the decision-making structure. For example, control can be
exercised via control of operating decisions such as daily management decisions and information
flows within a firm. Control can also be achieved via technology such as licensing agreements
that can be revoked, or patents and distribution channels, and so on. IAS 27 extends the scope of
entities that must be included for consolidation beyond what was covered in ARB 51. In
particular, ARB 51 employs a more definitive ownership rule and IAS 27 employs a more
11
principle-based control approach in delineating the consolidation requirements. Under the
ownership-based approach, the parent can easily structure the ownership of investees below 50%
to exclude the investee from the group reporting entity.
Thus, if a specific equity-ownership percentage does not translate directly into control or the
lack thereof, this raises serious questions as to whether consolidated statements under ownershipbased approach can faithfully depict the full financial picture of an economic entity. While prior
literature (Harris et al., 1994; Niskanen et al., 1998; Abad et al., 2000; Goncharov et al., 2009)
has already suggested that consolidated statements are more value relevant than unconsolidated
statements, some studies have indicated that how one defines the boundary for consolidating
entities can also determine the value relevance of accounting information. For example, before
SFAS No. 94, Consolidation of All Majority-Owned Subsidiaries, ARB 51 allowed the
exceptions to consolidation of “non-homogeneous subsidiaries.” A manufacturing firm with
finance company subsidiaries, for instance, might exclude such subsidiaries from the
consolidation.16 Benis (1979) and Mohr (1988) find that this elimination of a significant
subsidiary from consolidation makes assessment of future cash flows difficult, as the financial
ratios tend to be biased and less transparent. In addition, Heian and Thies (1989) suggest that the
exclusion of non-homogeneous subsidiaries can lead to favourable effects upon capital structure
indicators, which can in turn reduce the probability of technical violation of debt covenant
agreements.
In addition, Duchac (2004) argue that firms have incentives to hide their liabilities off
their balance sheets to attain lower cost of capital, or higher ratings, to avoid violations of debt
covenants, and to increase financial flexibility and firm profitability. For instance, Enron has
structured transactions between itself and the investees in a way that allows profits to be shifted
16
See note 9.
12
from the investees to the parent firm or allows costs to be shifted from parent firm to the affiliate.
Some studies (Jian & Wong, 2009; Sherman and Young, 2001) also find that related party
transactions between a firm and its affiliates are associated with earnings management and
managers tend to use the transactions to attain higher compensation. Under IAS 27, all
unrealized gains and losses recognized in the parent’s accounts arising from parent-investee
intercompany transactions need to be 100% eliminated from the income statement and balance
sheet if investees are regarded as the parent’s subsidiaries (i.e., under the control of the parent).
However, if the investees are not qualified as the parent’s subsidiaries under ARB 51 rule (i.e.,
the parent’s ownership on the investees is below 50%), the unrealized gains and losses are only
partially eliminated up to the portion of the parent’s ownership. Whether an investee is defined
as a subsidiary or not can determine the extent to which accounting information aligns with
economic performance.
Thus, insofar as the IAS 27 control-based approach can include entities with less than 50%
ownership in the reporting entity, we would expect consolidations under that approach to be
more aligned with the economic entity concept and more transparent than those under the
ownership-based approach. To the extent that stock prices reflect all economic information,
consolidated financial statements based on the control approach will be more value relevant than
those based on the ownership approach. We develop our first hypothesis, in an alternative form,
as:
Hypothesis 1: The value relevance of consolidated statements is higher under the IAS 27
control-based approach than under the ARB 51 ownership-based approach.
Common control model
13
Next, this study tries to shed light on the “common control model” proposed in the DP.
Although IAS 27 uses control rights to delineate the boundaries of consolidation, “control” is
expected to rest directly or indirectly with the parent. IAS 27 defines a group reporting entity as a
group of corporations under the control of a common parent. It does not include those entities
(i.e., “sibling” entities) that are not under the control of the parent but are under the common
control of an individual investor or family. The concern is whether consolidated financial
statements under the IAS 27 control-based approach can depict the full financial picture of an
economic entity, particularly for one with a complex ownership structure. In the context of the
DP, some argue that it is crucial for creditors and equity investors to understand the combined
statements for those commonly controlled entities and use the “common control model” to set
the domain of a group reporting entity.
Prior literature (Claessens et al., 2000; Faccio & Lang, 2002; LaPorta et al., 1999; Shleifer
& Vishny, 1997) finds that ultimate owners in East Asia usually exercise their control through
pyramid structures and cross-holdings such that the control rights for each family-owned entity
do not rest directly or indirectly with any other family-owned entities, but directly or indirectly
with family members. Figure 1 illustrates this by showing the investment groups owned by one
family, the Hsu family. We only sketch a preliminary picture of the complex structure by
referring to the seven listed firms within the Hsu-owned empire. With the help of a pyramid
structure and cross-holdings, the Hsu family is the ultimate owner for the seven listed firms. The
control over the listed firms rests directly and indirectly with the ultimate owners, not with any
other listed firms. Thus, under the IAS 27 control-based approach, none of the listed firms needs
to be consolidated with the other listed firm(s), thereby creating seven group reporting entities
and seven consolidated statements for the Hsu family’s business empire.
14
[INSERT FIGURE 1 ABOUT HERE]
In this study, we argue that if the providers of capital to family-centered affiliates wish to
understand the resources of the whole group in order to estimate the risk of cash flows from the
affiliates’ operations, under some circumstances (as in the case of the seven listed firms
controlled by the Hsu family) it is crucial to combine financial statements for all commonly
controlled entities. If family members effectively manage those commonly controlled entities as
a single unit, interaction between commonly controlled entities exists such that the returns to
each commonly controlled entity’s capital providers depend on the whole commonly controlled
entities’ combined business operations. For example, loan guarantees among affiliated firms in
family groups have been extensive in East Asia. Loan guarantees among affiliated firms may
allow a family group to help family-affiliated firms reduce distress risk and exploit growth
opportunities as they arise (Shin & Park, 1999). Family members may also use loan guarantees
as a way of generating private benefits that are not shared with non-controlling shareholders, in
essence looting firms and expropriating wealth (Claessens et al., 2000; Faccio & Lang, 2002;
Haw, Hu, Hwang, & Wu, 2004; LaPorta et al., 1999). In either case, without combined financial
statements, providers of capital to family-centered affiliates would find it difficult to understand
the resources within the whole group in order to estimate the risk to cash flows from the group’s
operations.
Since IAS 27 only addresses the consolidation issue among controlling entities, but does not
address the consolidation issue between commonly controlled entities, we expect any
improvement in value relevance under the IAS control-based approach to be limited to firms
with simple ownership structures. Simple ownership structures are structures in which entities
commonly controlled by an individual investor or family (“sibling” entities) are less prevalent. If
15
an individual investor or family wishes to maintain control rights over such entities without
equivalent ownership, they would need to construct a complex structure (i.e., pyramid structure
or cross-holding structure). In those cases, consolidated financial statements under the IAS 27
control-based approach would overlook those commonly controlled entities. We therefore
develop our second hypothesis, in an alternative form, as:
Hypothesis 2: The value relevance of consolidated statements is higher under the IAS 27
control-based approach than under the ARB 51 ownership-based approach for firms with simple
ownership structures but not for firms with complex ownership structures.
3. Sample and Research Design
3.1 Sample Selection
We take advantage of a unique setting in Taiwan’s capital market where all public firms
have been required to prepare consolidated financial statements in accordance with Taiwan’s
SFAS No. 7, Consolidated Financial Statements (“TSFAS 7”), which was in the spirit of ARB
51 before 2005 and IAS 27 from 2005 on.
Financial accounting standards in Taiwan are issued by the Accounting Research and
Development Foundation (ARDF). The ARDF followed US GAAP as its primary reference from
its inception in 1984 until the late 1990s. In 1985, the ARDF promulgated TSFAS 7 based on
ARB 51. Specifically, Paragraphs 8 and 11 of TSFAS 7, in the spirit of ARB 51, clearly stated
that consolidated financial statements are necessary for a fair presentation when one of the
companies in the group directly or indirectly has ownership of a majority (i.e. more than 50%) of
voting shares. In 2003, TSFAS 7 was amended as an attempt to converge with IAS 27. TSFAS 7
16
(revised), in the spirit of IAS 27, requires the inclusion of an entity if the parent has the power to
govern the financial and operating policies of an entity so as to obtain benefits from its activities
(Paragraph 16). The revised standard was effective for financial statements from 2005 on. The
change in the standard provides a setting to test for differences between the control-based
approach and the ownership-based approach in terms of the value relevance of consolidated
statements.
Our sample includes all the non-financial firms listed on the Taiwan Stock Exchange (TSE)
during the period 2000 to 2008. The sample period starts from 2000 to ensure that data is
available on pyramid structures and cross-holding structures. All the accounting and market data
used for this study are from the Taiwan Economic Journal (TEJ) database. The initial sample
includes 4,850 observations, from which we deleted 738 observations due to insufficient data to
calculate stock returns and 210 extreme observations (i.e., top and bottom 1% observations for
each variable) to avoid potential outliers. The final sample, therefore, contains 3,902
observations.
3.2 Research Models
To examine the value relevance differences proposed in our hypotheses, we use EdwardBell-Ohlson valuation model in which market value is modelled as a function of book value of
equity and net income. The model was originally proposed by Edwards and Bell (1961),
followed by Peasnell (1982), and was reinforced and popularized by Ohlson (1995) and Feltham
and Ohlson (1995).17 The model may be written as:
17
The dividend discount model acts as its starting point:
MVit
Et Dt
1
1 r
.
t
t 1
With the assumption of a “clean surplus accounting relationship” ( BVit
model can be recast in the Edward-bell-Ohlson framework: MV
it
BVit
Et NI t
t 1
17
BVit
1
1
CSEit
1 r
t
Dit ), the dividend discount
MVit
0
1
EQUITYit
2
NIit
(1)
it
where MVit denotes market value for firm i four months after the end of fiscal year t; EQUITYit
denotes the book value of stockholders’ equity for firm i at the end of year t; NIit denotes total
income for firm i for year t.
Following prior literature on consolidation (Abad et al., 2000), we use Edward-Bell-Ohlson
model to test value relevance because it conveniently incorporates both the balance sheet and the
income statement impacts in one model (Jennings, 1990). The model has also been frequently
used to test the value relevance of accounting information, as it provides a framework for
understanding the relation between prices and accounting information (Aboody & Lev, 1998;
Collins, Maydew, & Weiss, 1997; Collins, Pincus, & Xie, 1999). In Section 5, we also employ a
“return model” for robustness tests.
We decompose equity value into total assets, liabilities and non-controlling interest to
allow the coefficients on these variables to differ, and to estimate the incremental value
relevance for each component.18 Equation (1) can be reframed as equation (2).
4
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCI it
4
C _ NI it
2008
5,H
H 1
CONTROLH
6 ,Y
YEARY
(2)
it
Y 2000
where: C_ASSETit denotes consolidated assets for firm i at the end of year t; C_LIABILITYit
denotes consolidated liability for firm i at the end of year t; NCIit denotes non-controlling
interest for firm i at the end of year t; C_NIit denotes consolidated net income for firm i in year t.
We control for SIZEit, LOSSt, LEVit, and GRWOTHit, that may affect the value relevance of
accounting information (DeFond & Jiambalvo, 1994; Dechow & Dichev, 2002; Ghosh & Moon,
where D is total dividends; BV is the book value of shareholders’ equity; NI is total income; MV is total market
value of shareholders’ equity; and r is the discount rate.
18
For a group, the percentage of a subsidiary’s shares owned by the controlling parent company is not always 100 %.
Those shareholders of the subsidiary other than the parent firm are denoted as non-controlling interests, commonly
referred to as minority interests. 100% of the assets and liabilities of the parent company and subsidiary are shown
in the consolidated balance sheet, and any noncontrolling interest in the subsidiary’s net assets is reported separately
in the stockholders’ equity section of the consolidated balance sheet (Beams et al., 2009).
18
2005). SIZEit is defined as the natural logarithm of total market value for the current period t;
LOSSit is an indicator for firms having negative income in year t and zero otherwise; LEVit, is
defined as the ratio of total debts to total assets in year t; GROWTHit represents growth
opportunities, defined as the market-to-book ratio of shareholders’ equity. We also control for
the year effects such as period-specific economic effects, and firm-specific effects that are not
captured by the explanatory variables. YEARt is a time indicator that equals one if an
observation is from fiscal year t and zero otherwise. We use per share data in our cross-sectional
level model and check the sensitivity of our results to different deflators (i.e., book value of
equity or the market value of equity at the beginning of period t). The results are qualitatively
similar.
We estimate Equation (2) in a pooled regression and in a cross-sectional regression for each
of the years from 2000 to 2008. In Equation (2), significant positive values for
negative value for
2
1
and
4
, and
indicate the value relevance of consolidated assets, consolidated net income
and consolidated liabilities, respectively. To test whether control-based consolidated statements
have greater value relevance than ownership-based consolidated statements (H1), we insert an
indicator variable (POSTt) to interact with all variables in Equation (2). POSTt equals 1 when
observations are under the control-based approach after 2005 and zero when observations are
under the ownership-based approach before 2005.
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCIit
4
C _ NIit
5
POSTt
4
7
POSTt C _ LIABILITYit
8
POSTt NCIit
9
POSTt C _ NIit
6
,
7
10,H
and
POSTt C _ ASSETit
(3)
2008
H 1
The focus of regression (3) is on
6
CONTROLH
11,Y
YEARY
it
Y 2000
9
, which are incremental coefficients for value
relevance of consolidated assets, liabilities and net income, respectively during the period under
the control-based approach relative to the ownership-based approach. We expect significant
19
positive values for
6
and
9
, and negative value for
7
if Hypothesis 1 is supported.
Specifically, if consolidation incorporates all investees that the parent can exert control over
rather than only the investees in which the parent has at least 50% ownership, we expect that
consolidated balance sheets and income statements under the control-based approach can capture
an economic entity more fully. To the extent that stock prices can fully reflect information on an
economic entity, we expect the value relevance of consolidated assets, consolidated liabilities
and consolidated income to be higher under the control-based approach than under the
ownership-based approach.
To control for cross-sectional and time-series correlation, for all regression models we
adjust t-statistics in panel regressions for heteroskedasticity and within-firm correlation using
clustered standard errors.
Common control tests: Pyramid structure and cross-holding structure
Claessens et al. (2000) find that corporate control in East Asia is typically enhanced by
pyramid structures and cross-holdings such that the control rights for each family-owned entity
do not rest directly or indirectly on any other family-owned entities, but in family members.
Thus, we use pyramid structure or cross-holding structure as a proxy for commonly controlled
entities. We decompose our sample into four subgroups based on whether firms are associated
with pyramid structures and cross-holding structures set up by ultimate owners: (1) firms
associated with neither structure; (2) firms only associated with pyramid structure; (3) firms only
associated with cross-holding structure; and (4) firms associated with both structures. As the
control-based approach under IAS 27 does not account for common control, we would expect its
improvement over the ownership-based approach in value relevance, if any, will be less
20
pronounced for firms that are within pyramid structures and/or cross-holding structures because
these firms have more commonly controlled siblings than those in simple structures.
4. Empirical Results
4.1 Descriptive statistics
The descriptive statistics of all the variables examined in this study are reported in Table 1.
Accounting variables are reported on a per share basis, and are deflated by share price at the end
of the prior year. Panel A shows that the mean value of shareholders’ equity per share (EQUITYit)
is NT16.973, which is equal to the sum of the mean value of consolidated assets per share
(C_ASSETit: NT 33.504) and consolidated liability per share (C_LIABILITYit: NT 15.698) less
noncontrolling interest (NCIit : NT 0.833).19 Panel B separately reports the mean (median) value
for observations during 2000-2004 applying ARB 51-equivalent TSFAS 7 and observations
during 2005-2008 applying IAS27-equivalent TSFAS 7(revised). When comparing values across
the two groups, we find that the mean (median) values of MV and RET during the ARB 51
periods are 20.061 (14.205) and 0.093 (0.001), similar to the corresponding mean (median)
values during the IAS 27 periods. While there is no much difference between the mean value of
EQUITYit during 2000-2004 and 2005-2008, the mean value of C_ASSETit, C_LIABILITYit,
and NCIit are 32.433, 15.357 and 0.674, respectively during 2000-2004. These values are smaller
than the corresponding mean values of C_ASSETit, C_LIABILITYit, and NCIit during 2005-2008.
This suggests that more investees have been consolidated under the control-based approach after
companies switch from ARB 51-equivalent TSFAS 7 to IAS 27-equivalent TSFAS 7 (revised).
19
The conversion rate between US dollars and New Taiwan dollars (NT) is about 1:33.
21
Panel C exhibits the Pearson and Spearman correlation coefficients for all the variables
under analysis. It shows that market value and stock returns are correlated with consolidated
assets, consolidated liabilities and consolidated income.
[INSERT TABLE 1 ABOUT HERE]
4.2 Regression Models
Test of H1
Columns (1) and (2) in Table 2 report estimates of equations (2) and (3) using panel
regression with fixed effects for industry and year. In Column (1), the reported results are based
on regressing market value on consolidated assets, consolidated liabilities, non-controlling
interest and consolidated net income for the whole sample. The coefficients on consolidated
assets, liabilities, non-controlling interest and net income for total sample are 0.822, -0.988, 0.766, and 1.055, with t-statistics of 7.19, -5.80, -8.13 and 2.14, respectively. These results are
consistent with prior literature finding that book value of total assets and liabilities, along with
net income, can help explain stock price. Consistent with So and Smith (2009), our results also
show that the market perceives non-controlling interest (NCI) as a liability. This supports parent
company theory that the stock price only reflects positively the equity owned by equity holders
of the parent firm (Scofield, 1996).
[INSERT TABLE 2 ABOUT HERE]
Column (2) shows estimates of equation (3) on the value relevance of consolidated
statements before and after 2005. To test H1, whether consolidated statements under the controlbased approach are more value relevant than those under the ownership-based approach, we
divide the whole sample into two sub-periods: 2000-2004 and 2005-2008. The former
22
corresponds to the period applying ARB 51-equivalent TSFAS 7, whereas the latter corresponds
to the period applying IAS27-equivalent TSFAS 7(revised). The main parameters of interest,
POSTt
C_ASSETit, POSTt
C_LIABILITYit, and POSTt
C_NIit measure the difference
between IAS 27 control-based and ARB 51 ownership-based consolidation rules in terms of the
relation between market value and consolidated assets, consolidated liabilities and consolidated
net income, respectively. All variables interacting with POSTt are consistently significant at the
1% level with the predicted signs. Specifically, the coefficients on POSTt
C_LIABILITYit, and POSTt
C_ASSETit, POSTt
C_NIit are 0.382 (t=3.81), -0.518 (t=-4.62), and 2.724 (t=9.70),
respectively. In line with H1, the results show that consolidated accounting information based on
the control approach is more closely associated with market value than consolidated statements
based on the ownership approach. Overall, these results indicate that value relevance of
consolidated statements under the IAS 27 control-based approach is higher than that under the
ownership-based approach.
As for the coefficient on POSTt
NCIit, we do not find any significant effects, which
indicates that investors do not change their perception of non-controlling interests and that the
parent’s shareholders view NCI as liability as before.
Test of H2
In Table 3 we partition the full sample into four groups based on whether a pyramid
structure is used and/or a cross-holding structure is used. Panel A refers to the results for the
firms that do not establish either pyramid or crossholdings; Panel B refers to the results for the
firms that only establish pyramid; Panel C refers to the results for the firms that only establish
crossholdings; Panel D refers to the results for the firms that establish both pyramid and
crossholdings. The overall regression results indicate that the improvement in value relevance
23
after 2005 is primarily driven by the sample firms without pyramid structures or crossholdings.
Specifically, for firms without pyramid structures, the coefficients on POSTt
C_ASSETit (0.421,
t=2.59), and POSTt
C_LIABILITYit
C_NIit (2.852, t=6.73) are significantly positive; POSTt
(-0.597, t=-3.34) is significantly negative. However, for firms associated with the pyramid
structure and/or crossholdings (Panel B-Panel D), we find no similar patterns except in the case
of net income. Specifically, in Panel B, we find that consolidated statements under the controlbased approach are not more relevant for predicting market price than those under the
ownership-based approach for firms with pyramid. Our findings imply that while the controlbased approach encompasses more entities that can be controlled by the parent firm without
acquiring 50% ownership, the approach does not address the control that rests in an individual
investor or family through pyramid and cross-holding structures. This supports H2.
[INSERT TABLE 3 ABOUT HERE]
Panel C reports results of replicating the analysis in Panel B for firms with cross-holding
structures. Our results support the view that the loss of credibility in consolidated statements for
firms affiliated with many sibling entities cannot be reduced by adoption of the IAS 27 controlbased approach. This supports the concern raised in the DP (Paragraph 83) that, without
combined statements for entities commonly controlled by the same owners, capital providers of
the family-centered affiliates might have difficulty understanding the resources within the whole
group and fail to estimate the amounts, timing and uncertainty of cash flows from those
companies.
5. Sensitivity Analysis
5.1. Difference-in-difference tests
24
We conduct a difference-in-difference test to mitigate some concerns on the effect of
macro and economic changes over time and firm characteristics that could have on the pricing of
consolidated statements. We separate our firms into the group A that is affected by the new
consolidation rule and the group B that is not affected by the rule, and compare any differential
pricing of consolidated statements in the IAS 27-equivalent period between group A and group B.
Group A refers to the firms who did not consolidate some investees before 2005 under
ownership-based approach, but started to consolidate those investees from 2005 on under
control-based approach. Group B refers to those firms who are not affected by the revisedTSFAS 7. We expect that the improvement in the value-relevance after 2005 can only be
observed in group A, but not in group B.
We construct one measure to estimate whether the firm is affected by the IAS27equivalent TSFAS 7, and separate our sample into group A and group B. The measure is an
indicator that is equal to one (group A) when the parent has an investee with ownership less than
50% that exists in both 2004 and 2005, but the parent only consolidates the investee in 2005; and
is equal to zero when the parent consolidates the investee in both years or in neither years (group
B). Table 4 reports the results. Our results show that POSTt
POSTt
C_ASSETit (0.520, t=5.40) and
C_NIit (2.232, t=7.74) are significantly positive; POSTt
C_LIABILITYit (-0.626, t=-
5.54) is significantly negative in group A. For group B, we do not find any improvement in the
value relevance of consolidated assets and liabilities after 2005. 20 Our results reconfirm our
hypotheses that control-based approach can improve market pricing better than ownership-based
approach. 21
[INSERT TABLE 4 ABOUT HERE]
20
We rerun our analysis of Table 3 and 4 using only the sample of group A, and the results are qualitatively similar.
Finally, to ensure our coefficients are stable over time, we also re-run the tests using Fama-Macbeth regressions to
compare the value relevance between the period 2000-2004 and 2005-2008, and the results are qualitatively similar.
21
25
5.2. Return Model
While Edward-Bell-Ohlson valuation model is frequently used in empirical work, it suffers
from several shortcomings. In particular, it has been suggested that omitted variables are likely to
affect the level (i.e. market value) regression more than the change (i.e. returns) regression
(Kothari & Zimmerman, 1995). Thus, following previous studies (Easton, 1999), we also employ
a returns model for robustness tests. Following Easton and Harris (1991), and Ohslon and Shroff
(1992), we incorporate both the level of and annual changes in consolidated net income to test
their informativeness with respect to stock returns.
4
RETit
0
1
C _ NI it
2
C _ NI it
3
POSTt
4
POSTt C _ NI it
5
POSTt
C _ NI it
2008
6 ,H
CONTROLH
H 1
7 ,Y
YEARY
it
(4)
Y 2000
where: RETit denotes the firm’s annual stock returns, cumulated from eight months before the
end of fiscal year t through four months after the end of fiscal year t; ∆C_NIit is the change in
consolidated net income per share during each accounting period t.
If consolidated statements under the IAS control-based approach are more value relevant
than under the ARB ownership-based approach, we expect the differential slope coefficients for
the interaction variables (POSTt
C_NIit and/or POSTt
△C_NI it) to be positive and
significant.
Column one of Table 5 reports results for equation (4), which tests the incremental
informativeness of the level of, and changes in, consolidated net income. The results indicate that
the annual changes in consolidated net income after 2005 are more incrementally informative
than before 2005, as the coefficient on POSTit
△C_NIit is 0.021, significant at the 1% level.
The results reconfirm H1 that the control-based approach provides information that is more value
relevant than the ownership-based approach.
26
Columns two to five of Table 5 report the results of replicating the analysis of Table 3 by
substituting the market value model with the returns model to see if our results are sensitive to
the potential omitted variables in the price model. Again, we only observe significant
improvements after 2005 among firms that are not controlled by ultimate owners through a
pyramid structure or cross-holding structure. For firms that adopt a pyramid structure and/or
cross-holding structure, we do not observe any difference between IAS-control based approach
and ARB ownership-based approach in terms of the incremental informativeness of the change
in net income, as POSTt △P_NIit and POSTt △S_NIit are insignificant.
[INSERT TABLE 5 ABOUT HERE]
5.3. Value relevance of parent/subsidiary accounting information
As consolidated information reflects information that has a direct influence on the
financial well-being of the parent company’s shareholders, we further explore the sources of the
improvement in value relevance under the IAS 27 control-based approach. In so doing, we take
advantage of the dual reporting system in Taiwan that requires listed companies to file with the
regulatory agency consolidated statements as well as the parent’s unconsolidated financial
statements. We thus decompose consolidated information into information on the parent’s main
operations (denoted as P) and information on subsidiaries’ main operations (denoted as S). This
allows us to test whether the improvement in value relevance of consolidated statements is
driven by an improvement in the value relevance of both the parent’s information and
subsidiaries’ information.
MVit
0
1
P _ ASSETit
9 POST P _ ASSETit
2
S _ ASSETit
4
15
3
10 POST S _ ASSETit
POSTit S _ NI it
4
S _ LIABILITYit
11 POST P _ LIABILITYit
2008
16 ,H
H 1
P _ LIABILITYit
CONTROLH
17 ,Y
YEAR
it
Y 2000
27
5
NCIit
6
12 POST S _ LIABILITYit
P _ NIit
7
S _ NIit
13 POST NCI it
8
POSTit
14 POSTit P _ NI it
(5)
where: P_ASSETit denotes parent firm i’s total operating assets and non-controlling financial
assets at the end of year t; S_ASSETit denotes subsidiaries’ assets for firm i at the end of year t;
P_LIABILITYit denotes parent firm i’s liability at the end of year t; S_LIABILITYit denotes
subsidiaries’ liability for firm i at the end of year t; P_NIit denotes parent firm i’s total income
from its own operations and income from non-controlling financial assets; S_NIit denotes the
proportionate share of subsidiaries’ total income for firm i.
First, we expect that the value relevance of the subsidiaries’ information can improve as
the consolidation procedure can decompose the aggregate information (i.e., investment accounts
in the parent’s accounts) into the amount of subsidiaries’ assets, liabilities and non-controlling
interest of the subsidiaries (Beams et al., 2009). Furthermore, we also expect that the value
relevance of the parent’s information can improve when companies switch from ARB 51
approach to IAS 27 approach. Under IAS 27, all unrealized gains and losses recognized in the
parent’s accounts arising from parent-investee intercompany transactions need to be 100%
eliminated from the income statement and balance sheet if investees are regarded as the parent’s
subsidiaries (i.e., under the control of the parent). However, if the investees are not qualified as
the parent’s subsidiaries under ARB 51 rule (i.e., the parent’s ownership on the investees is
below 50%), the unrealized gains and losses are only partially eliminated up to the portion of the
parent’s ownership. Some studies (Jian & Wong, 2009; Sherman and Young, 2001) find that
related party transactions between a firm and its affiliates are associated with earnings
management and managers tend to use the transactions to attain higher compensation. For
instance, parent firm may engage in channel stuffing by forcing an entity to purchase higher than
normal inventory levels, and increase the sales and profitability of the parent firm. If the
affiliates are part of the consolidated entity, the channel-stuffing transactions would be
28
eliminated. However, if the entity is not qualified as a subsidiary, the consolidated earnings
would be overstated without removing the unrealized gains from the intercompany transactions.
Table 6 reports the regression results. Column 1 of Table 6 shows that the coefficients on
the interaction variables, POSTt
t=6.12), POSTt
P_ASSETit (0.372, t=5.74), POSTt
P_NIit, (2.936, t=14.30) and POSTt
positive and the coefficients on POSTt
S_ASSETit (0.464,
S_NIit (1.878, t=5.71), are all significantly
P_LIABILITYit (-0.554, t=-6.69) and POSTt
S_LIABILITYit (-0.406 t=-4.56) are significantly negative. When the ARB 51 ownership-based
approach is replaced with the IAS 27 control-based approach, investors increase their confidence
in the consolidated reports and unconsolidated reports. Thus, the control-based approach to
consolidation has significant incremental value relevance in providing more credible accounting
information to investors.
Column 2 (column 3) provides regression results for firms without (with) a complex
structure. The results show that the improved value relevance after 2005 is primarily driven by
the sample firms not associated with a complex structure. For example, for firms without a
pyramid structure or crossholding, the coefficients on POSTt
POSTt
S_ASSET it (0.337, t=2.90), POSTt
t=4.95) are significantly positive; POSTt
P_ASSETit (0.385, t=3.94) ,
P_NIit (3.131, t=10.41), POSTt
S_NI it (2.368,
P_LIABILITYit (-0.597, t=-4.95), and POSTt
S_LIABILITYit (-0.344, t=-2.52) are significantly negative. However, for firms with both
pyramid structure and cross-holding, we find no similar patterns, except for net income. The
results reconfirm H2 that the improvement in value relevance of the IAS 27 control-based
approach over the ARB 51 ownership-based approach does not apply to firms associated with
complex ownership structures.
[INSERT TABLE 6 ABOUT HERE]
29
We also employ return model for robustness tests. Untabulated results also support our
findings that both income from parent’s operations and subsidiaries operations can improve
under control-based approach of IAS27 as compared to ownership-based approach of ARB 51.
6. Conclusion
As part of a joint project to develop a common conceptual framework for financial reporting,
a discussion paper (DP) jointly issued by the IASB and the FASB sought feedback on the
definition of group reporting entity. At the core of the DP is how to delineate the scope of a
group reporting entity for the purpose of compiling consolidated financial statements.
Specifically, the DP requests comments on three inquiries: (1) whether composition of a group
reporting entity should be based on control; (2) whether the controlling entity model should be
used as the primary basis to define a group entity; and (3) whether in some circumstances we
should use a common control model that is more comprehensive than the controlling entity
model.
To provide direct feedback on these inquiries, this study compares the value relevance of
consolidated financial statements reported under the IAS 27 control-based approach and the
ARB 51 ownership-based approach. Using the regulatory switch from the ownership-based
approach to the control-based approach in Taiwan, we find that consolidation under the controlbased approach can better depict the financial picture of a group entity than consolidation under
the ownership-based approach. This result supports the effectiveness of the control-based
approach in extending the scope of consolidated entities beyond those under the ownership-based
approach. Prior studies (Duchac, 2004; Jian & Wong, 2009; Sherman and Young, 2001) argue
that firms have incentives to hide their liabilities off their balance sheets or to increase financial
profitability to attain lower cost of capital. Our findings imply that the control-based approach
30
can reduce the flexibility for the parent to structure transactions between itself and the investees
in a way that allows the liabilities or costs to be shifted from the parent to investees, or allow
profits to shift from the parent to the investees.
Also, to examine the common control model, this study uses complicated ownership
structures (pyramid structures or cross-holding) to proxy for situations that might call for use of
the common control model. Since consolidated financial statements of commonly controlled
entities are not generally available under current accounting standards, we take advantage of the
ownership-structure feature in East Asia where family members tend to construct complex
structures to attain control over all family-centered affiliates without equivalent ownership. As
current accounting standards do not require inclusion of entities under the common control of the
members of a single family, the complex structure for each family group usually ends up with
many reporting entities and many sets of consolidated statements. Thus, it is difficult for capital
providers to understand the entire resources of such family-owned groups with any of the
available consolidated statements. Our results show that the improvements brought by the IAS
27 control-based approach over the ownership-based approach are more pronounced for firms
with simple ownership structures than for firms with complex structures. The implication is that
the scope of consolidation required under the controlling entity model is not wide enough to
handle complicated ownership structures without further guidance on the lines of the common
control model. To better serve existing and potential investors in “sibling entities” or familyowned groups, consolidation at the common control level should be considered.
While this study presents empirical results relating to the DP’s inquires directly, there are
limitations to this study. First, the current US GAAP consolidation rules do not encompass a
single consolidation model. Rather, an enterprise is required to first assess which of two
31
consolidation models is applicable, a model based solely on ownership for voting entities, or a
model for special purpose entities based solely on risk and rewards (FIN 46R). It is therefore
acknowledged that this study only focuses on voting entities without covering special purpose
entities. Secondly, for making inferences on the common control model proposed by the DP, a
better proxy can be developed to index complicated common control in family owned or
affiliated sibling entities. This study uses pyramid structures and cross-holding as an indicator of
common control, following previous studies (Claessens et al. 2000, Fang & Wong 2002);
however, until “common control” is better defined by either the IFRS or the US GAAP, there is
no telling whether these complex structures embody the kind of control meant by the standards.
Further studies are needed on how the common control model might best be used to define the
scope of an accounting reporting entity.
32
References
Abad, C., Laffarga, J., Garcia-Borbolla, A., Larran, M., Pinero, J.M., & Garrod, N. (2000). An
evaluation of the value relevance of consolidated versus unconsolidated accounting information:
Evidence from quoted Spanish firms. Journal of International Financial Management and
Accounting, 11, 156–177.
Aboody, D., & Lev, B. (1998). The value relevance of intangibles: The case of software
capitalization. Journal of Accounting Research, 36 (supplement), 161–199.
American Institute of Certified Public Accountants (AICPA). (1959). Consolidated Financial
Statements. Accounting Research Bulletin 51. New York, NY: AICPA.
Barth, M., Landsman, W., & Lang, M. (2008). International accounting standards and accounting
quality, Journal of Accounting Research, 46, 467–98.
Beams, F. A., Brozovsky, J.A. & Shoulders, C. D. (2009). Advanced Accounting. Pearson
Prentice Hall, Inc, 10th Edition.
Benis, M. (1979). The Non-Consolidated Finance Company Subsidiary. The Accounting Review,
54, 808-814.
Claessens, S., Djankov, S., & Lang, H.P. (2000). The separation of ownership and control in east
asian corporations. Journal of Financial Economics, 58, 81-112.
Collins, D., Maydew, E., & Weiss, I. (1997). Changes in the value-relevance of earnings and
book values over the past forty years. Journal of Accounting and Economics, 24, 39-67.
Collins, D. W., Pincus, M., & Xie, H. (1999). Equity valuation and negative earnings: the role of
book value of equity. The Accounting Review, 74: 29-61.
Conklin, D., & Lecraw, D. (1997). Restrictions on foreign ownership during 1984-1994:
Developments and alternative policies. Transnational Corporations, 6 (April), 1-30.
33
Dechow, P.M., & Dichev, I. (2002). The Quality of Accruals and Earnings: The Role of
Accrual Estimation Errors. The Accounting Review, 77, (Supplement), 35-59.
DeFond, M.L., & Jiambalvo, J. (1994). Debt Covenant Violation and Manipulation of Accruals.
Journal of Accounting and Economics, 17, (January),145-176.
Duchac, J. (2004). The dilemma of bright line accounting rules and professional judgment:
Insights from special purpose entity consolidation rules. International Journal of Disclosure and
Governance, 1(4), 324-338.
Duh, R.R., Lee, W., & Lin, C.C. (2009). Reversing an impairment loss and earnings
management: The role of corporate governance. International Journal of Accounting, 44, 113137.
Easton, P., & Harris, T. (1991). Earnings as an explanatory variable for returns. Journal of
Accounting Research, 29 (Spring), 19-36.
Easton, P. (1999). Security returns and value relevance of accounting data. Accounting Horizons,
3 (December), 399-412.
Edwards, J., & Bell, P. (1961). The theory and measurement of business income, University of
California Press.
Faccio, M., & Lang, L. H. (2002). The ultimate ownership of western European corporations.
Journal of Financial Economics, 65, 365-395.
Fan, J., & Wong, T. (2002). Corporate ownership structure and the informativeness of
accounting earnings in East Asia. Journal of Accounting and Economics, 33, 401-425.
Feltham, G., & Ohlson, J. (1995). Valuation and clean surplus accounting for operating and
financial activities. Contemporary Accounting Research, 11, 689–731.
34
Financial Accounting Standards Board (FASB). (1987). Consolidation of all majority-owned
subsidiaries. Statement of financial accounting standards 94. Stamford, CT: FASB.
Financial accounting standards board (FASB). (2003). Consolidation of variable interest
entities—an interpretation of ARB No. 51. FASB Interpretation 46(R). Stamford, CT: FASB.
Financial accounting standards board (FASB). (2001). Accounting for the impairment or
disposal of long-lived assets. Statement of financial accounting standards 144. Stamford, CT:
FASB.
Ghosh, A., & Moon, D. (2005). Auditor tenure and perceptions of audit quality. The
Accounting Review, 80, 2, 585–612.
Goncharov, I., Werner, J.R., & Zimmermann, J. (2009). Legislative demands and economic
realities: company and group accounts compared. International Journal of Accounting, 44, 334362.
Harris, T. S., Lang, M.H., & Möller, H. P. (1994). The value relevance of German accounting
measures: an empirical analysis. Journal of Accounting Research, 32(2), 187-209.
Haw, I., Hu, L., Hwang, S., & Wu, W. (2004). Ultimate ownership, income management, and
legal and extra-legal institutions. Journal of Accounting Research, 42(2), 423-462.
Heian, J. B. & Thies, J. B. (1989). Consolidation of finance subsidiaries: $230 billion in offbalance-sheet financing comes home to roost. Accounting Horizons, 3(1): 1-9.
Hoyle, J., Schaefer, T., & Doupnik, T. (2008). Advanced Accounting. McGraw-Hill/Irwin, 9th
edition.
Hsu, C., Lin, W., Lee, W., & Cheng, W. (2004). Complex pyramid structure for conglomerate
Fareast. Money Watching & Credit Rating, 65, 42-83.
International Accounting Standards Board (IASB). (2003). Consolidated and separate financial
statements. IAS 27. London, UK: IASB.
35
International Accounting Standards Board (IASB). (2008). Consolidated and Separate Financial
Statements. IAS 27. London, UK: IASB.
Jennings, R. (1990). A note on interpreting incremental information content. The Accounting
Review ,65 (4), 925–932.
Jian, M., & Wong, T. J. (2009). Propping through related party transactions. Review of
Accounting Studies, forthcoming.
Kothari, S. P., & Zimmerman, J. (1995). Price and return models. Journal of Accounting and
Economics, 20 (September), 155–193.
La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (1999). Corporate ownership around the world.
Journal of Finance, 54, 471-518.
Leuz, C., & Verrecchia, R. (2000). The economic consequences of increased disclosure. Journal
of Accounting Research, 38, 91–124.
Leuz, C. (2003). IFRS versus US GAAP: Information asymmetry-based evidence from
Germany’s new market. Journal of Accounting Research, 41, 445–72.
Mohr, R. M. (1988). Unconsolidated finance subsidiaries: Characteristics and debt/equity effects.
Accounting Horizons, 2(1), 27-34.
Niskanen, J., Kinnunen, E., & Kasanen, J. (1998). A note on the information content of parent
company versus consolidated earnings in Finland. The European Accounting Review, 7(1),
31−40.
Ohlson, J. (1995). Earnings, book values and dividends in equity valuation. Contemporary
Accounting Research, 11, 661-687.
36
Ohlson, J., & Shroff, P. (1992). Changes versus levels in earnings as explanatory variables for
returns: Some theoretical considerations. Journal of Accounting and Research, 30 (Autumn),
210-226.
Peasnell, K. (1982). Some formal connections between economic values and yields and
accounting numbers. Journal of Business Finance and Accounting, 9, 361–381.
Scofield, B.W. (1996). Alcoa: Toward understanding minority interest. Journal of Accounting
Education, 14, 579–606.
Sheleifer, A. & Vishny, R. (1997). A survey of corporate governance. Journal of Finance, 52,
737-783.
Sherman, H. D., & Young, S.D. (2001). Tread lightly through these accounting minefields.
Harvard Business Review, 79(7):129-35.
Shin, H. & Park, Y. S. (1999). Financing constraints and internal capital markets: Evidence from
Korea “cheobols”, Journal of Corporate Finance, 5, 169-191.
So, S., & Smith, M. (2009). Value relevance of IAS 27 (2003) revision on presentation of noncontrolling interest: Evidence from Hong Kong. Journal of International Financial Management
and Accounting, 20(2), 166-198.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test
for heteroskedasticity. Econometrica, 48(4), 817-838.
37
Figure 1 Ownership Structure of Hsu Family-owned Groups as of Dec. 2008
Sources: updated from Hsu, Lin, Lee and Cheng (2004).
38
Table 1 Descriptive statistics
Panel A: Summary Statistics
N
Mean
STD
Q1
Median
Q3
MVit
3,935
20.081
17.808
9.200
14.650
24.720
RETit
3,935
0.083
0.527
-0.285
-0.013
0.307
EQUITYit
3,935
16.973
15.355
12.115
15.355
20.228
C_ASSETit
3,935
33.504
16.366
22.050
29.240
41.131
C_LIABILITYit
3,935
15.698
11.219
7.698
13.017
20.437
NCIit
3,935
0.833
1.921
0.000
0.098
0.758
C_NIit
3,935
1.597
2.454
0.167
1.110
2.685
NCIEit
3,935
0.039
0.279
-0.005
0.000
0.023
Panel B:Summary Statistics across two sub-periods
MVit
RETit
EQUITYit
C_ASSETit
C_LIABILITYit
NCIit
C_NIit
NCIEit
N
2,213
2,213
2,213
2,213
2,213
2,213
2,213
2,213
2000-2004
Mean
Median
20.061
14.205
0.093
0.001
16.402
15.021
32.433
28.476
15.357
13.051
0.674
0.062
1.458
1.019
0.029
0.000
Std
14.857
0.543
7.121
14.923
10.305
1.598
2.375
0.253
N
1,722
1,722
1,722
1,722
1,722
1,722
1,722
1,722
2005-2008
Mean
Median
20.106
15.31
0.071
-0.030
17.709
16.027
34.881
30.330
16.136
12.963
1.036
0.150
1.5776
1.243
0.051
0.000
STD
20.129
0.505
8.526
17.964
12.284
2.255
2.542
0.309
Panel C: Correlation Table
Variables
MVit
MVit
1
.
RETit
RETit
C_ASSETit
0.2453
(0.000)
1
0.4433
(0.000)
0.0271
(0.089)
1
C_LIABILITYit
0.1719
(0.000)
0.018
(0.258)
0.8523
(0.000)
1.000
C_NIit
0.6506
(0.000)
0.1853
(0.000)
0.5198
(0.000)
0.205
(0.000)
1.000
0.1825
(0.000) .
C_ASSETit
0.3839
0.0111
(0.000)
(0.485) .
C_LIABILITYit
0.1557
0.0242
0.8926
(0.000)
(0.130)
(0.000) .
C_NIit
0.6292
0.1305
0.5217
0.251
(0.000)
(0.000)
(0.000)
(0.000) .
All firms were listed on the Taiwan Stock Exchange from 2000 to 2008, and all the data were collected from the
Taiwan Economic Journal (TEJ) database. All variables except for RETit are in New Taiwan Dollars. MVit denotes
the market value of firm I’s shareholders’ equity four months after the end of fiscal year t; RETit denotes the firm’s
annual stock return, cumulated from eight months before the end of fiscal year t through four months after the end of
fiscal year t; EQUITYit denotes the book value of firm i’s shareholder equity at the end of fiscal year t; C_ASSETit
denotes consolidated assets for firm i at the end of year t; C_LIABILITYit denotes consolidated liability for firm i at
the end of year t; NCIit denotes non-controlling interest for firm i at the end of year t; C_NIit denotes consolidated
net income for firm i in year t; NCIEit denotes non-controlling shareholders’ interest in subsidiaries’ income in year
t. All regression variables are deflated by the number of shares outstanding at year-end.
39
Table 2 Value relevance of consolidated assets, liabilities and earnings
Whole sample
Regression (2)
Regression (3)
MV
MV
-31.780
-27.053
(-11.29)***
(-10.54)***
0.822
0.524
(7.19)***
(4.88)***
-0.988
-0.571
(-5.80)***
(-3.57)***
-0.766
-0.743
(-8.13)***
(-6.65)***
1.055
-0.345
(2.14)*
(-0.69)
-3.357
(-2.56)*
0.382
(3.81)***
-0.518
(-4.62)***
-0.110
(-0.67)
2.724
(9.70)***
Intercept
C_ASSETit
C_LIABILITYit
NCIit
C_NIit
POSTt
POSTt C_ASSETit (+)
POSTt C_LIABILITYit (-)
POSTt NCIit
POSTt C_NIit
Control Variables
SIZE
1.494
1.431
(4.43)***
(4.25)***
LOSS
5.924
5.037
(7.78)***
(7.30)***
LEV
5.164
1.812
(1.15)
(0.43)
GROWTH
8.107
8.621
(5.71)***
(5.95)***
N
3935
3935
adj. R2
0.644
0.727
Regression models (2) and (3) are estimated using pooled cross-section and time-series data for
non-financial firms from 2000 to 2008.
4
2008
(2)
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCI it
4
C _ NI it
5,H
CONTROLH
H 1
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCI it
4
C _ NIit
5
POSTt
4
7
POSTt C _ LIABILITYit
8
POSTt NCIit
9
POSTt C _ NIit
6 ,Y
YEARY
it
Y 2000
6
POSTt C _ ASSETit
(3)
2008
10,H
CONTROLH
H 1
11,Y
YEARY
it
Y 2000
MVit denotes the market value of firm i four months after the end of fiscal year t; C_ASSETit denotes
consolidated assets for firm i at the end of year t; C_LIABILITYit denotes consolidated liabilities of firm i
at the end of year t; NCIit denotes non-controlling interest for firm i at the end of year t; C_NIit denotes
consolidated net income for firm i in year t; POSTt equals to 1 when observations are under the controlbased approach (after 2005) and zero when observations are under the ownership-based approach (before
2005); YEARt is a time indicator that equals one if an observation is from fiscal year t and zero otherwise.
All regression variables are deflated by the number of shares outstanding at year-end. t-statistics are given
in parentheses and are adjusted for heteroscedasticity. ***, ** and * denote significance at the 1%, 5% and
10% levels, respectively, in a two-tailed test.
40
Table 3 Value relevance of consolidated assets, liabilities and earnings with respect to
different complex structures
Intercept
C_ASSETit
C_LIABILITYit
NCIit
C_NIit
POSTt
POSTt C_ASSETit
POSTt C_LIABILITYit
POSTt NCIit
POSTt C_NIit
(1)
No pyramid/No
crossholdings
MV
-37.046
(-6.50)***
0.431
(2.83)**
-0.468
(-2.24)*
-0.652
(-4.04)***
-0.073
(-0.10)
-4.029
(-1.85)
0.421
(2.59)**
-0.597
(-3.34)***
-0.200
(-0.88)
2.852
(6.73)***
Control Variables
SIZE
(2)
Only Pyramid
2.383
(3.58)***
6.034
(5.68)***
-0.448
(-0.09)
7.208
(3.50)***
2081
0.706
LOSS
LEV
GROWTH
N
adj. R2
MV
-24.557
(-6.56)***
0.740
(5.82)***
-0.934
(-4.90)***
-0.746
(-3.39)***
-0.604
(-1.55)
-1.814
(-0.92)
0.198
(1.40)
-0.210
(-1.26)
-0.062
(-0.23)
2.689
(6.71)***
(3)
Only Crossholdings
MV
-32.492
(-4.05)***
0.538
(1.87)
-0.787
(-1.78)
-0.480
(-0.49)
-0.651
(-0.95)
-5.749
(-1.42)
0.432
(1.47)
-0.571
(-1.51)
-0.129
(-0.13)
3.183
(5.20)***
0.773
(3.19)**
5.319
(5.42)***
9.956
(2.03)*
10.749
(18.46)***
755
0.616
1.242
(2.85)**
8.203
(4.13)***
20.243
(2.21)*
11.433
(16.16)***
425
0.836
(4)
Both structures
MV
-5.024
(-0.86)
0.654
(5.53)***
-0.871
(-4.34)***
-0.814
(-3.19)**
0.939
(1.71)
1.396
(0.70)
-0.078
(-0.56)
0.190
(1.15)
0.080
(0.24)
2.537
(4.08)***
1.108
(3.20)**
8.086
(7.00)***
7.488
(1.14)
10.025
(7.42)***
674
0.639
Regression model (3) is estimated using pooled cross-section and time-series data for non-financial
firms from 2000 to 2008.
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCI it
4
C _ NIit
5
POSTt
4
7
POSTt C _ LIABILITYit
8
POSTt NCIit
9
POSTt C _ NIit
6
POSTt C _ ASSETit
(3)
2008
10,H
CONTROLH
H 1
11,Y
YEARY
it
Y 2000
MVit denotes the market value of firm i four months after the end of fiscal year t; C_ASSETit denotes
consolidated assets for firm i at the end of year t; C_LIABILITYit denotes consolidated liability for firm i
at the end of year t; NCIit denotes non-controlling interest for firm i at the end of year t; C_NIit denotes
consolidated net income for firm i in year t; POSTt equals to 1 when observations are under the controlbased approach (after 2005) and zero when observations are under the ownership-based approach (before
2005); YEARt is a time indicator that equals one if an observation is from fiscal year t and zero otherwise.
All regression variables are deflated by the number of shares outstanding at year-end. t-statistics are given
in parentheses and are adjusted for heteroscedasticity. ***, ** and * denote significance at the 1%, 5% and
10% levels, respectively, in a two-tailed test.
41
Table 4 Value relevance of consolidated assets, liabilities and earnings for firms that are
affected by IAS 27 relative to firms that are not.
Group A
Regression (3)
MV
Group B
Regression (3)
MV
Intercept
-23.350
-32.740
(-9.68)***
(-4.62)***
C_ASSETit
0.599
0.402
(8.05)***
(1.43)
C_LIABILITYit
-0.690
-0.311
(-6.41)***
(-0.78)
NCIit
-0.626
-1.099
(-5.16)***
(-3.03)**
C_NIit
-0.819
0.490
(-3.50)***
(0.50)
POSTt
-5.041
-0.932
(-4.21)***
(-0.42)
0.520
0.253
POSTt C_ASSETit (+)
(5.40)***
(1.39)
-0.626
-0.431
POSTt C_LIABILITYit (-)
(-5.54)***
(-1.07)
-0.484
0.492
POSTt NCIit
**
(-3.07)
(1.29)
2.232
2.592
POSTt C_NIit
(7.74)***
(4.18)***
Control Variables
0.969
2.297
SIZE
(6.47)***
(2.73)**
3.932
6.448
LOSS
(7.85)***
(4.32)***
4.207
-7.051
GROWTH
(1.60)
(-0.73)
10.382
5.637
N
(25.34)***
(2.31)*
adj. R2
3076
859
0.751
0.707
Regression model (3) is estimated using pooled cross-section and time-series data for non-financial
firms from 2000 to 2008.
MVit
0
1
C _ ASSETit
2
C _ LIABILITYit
3
NCI it
4
C _ NIit
5
POSTt
4
7
POSTt C _ LIABILITYit
8
POSTt NCIit
9
POSTt C _ NIit
6
POSTt C _ ASSETit
(3)
2008
10,H
CONTROLH
H 1
11,Y
YEARY
it
Y 2000
MVit denotes the market value of firm i four months after the end of fiscal year t; C_ASSETit denotes
consolidated assets for firm i at the end of year t; C_LIABILITYit denotes consolidated liabilities of firm i
at the end of year t; NCIit denotes non-controlling interest for firm i at the end of year t; C_NIit denotes
consolidated net income for firm i in year t; POSTt equals to 1 when observations are under the controlbased approach (after 2005) and zero when observations are under the ownership-based approach (before
2005); YEARt is a time indicator that equals one if an observation is from fiscal year t and zero otherwise.
All regression variables are deflated by the number of shares outstanding at year-end. t-statistics are given
in parentheses and are adjusted for heteroscedasticity. ***, ** and * denote significance at the 1%, 5% and
10% levels, respectively, in a two-tailed test.
42
Table 5 Value relevance of consolidated assets, liabilities and earnings: returns model
(1)
Whole Sample
Intercept
C_NIit
△C_NIit
POSTt
POSTt C_NIit
POSTt △C_NIit
-0.779
(-9.61)***
-0.032
(-6.57)***
0.057
(9.85)***
0.755
(24.54)***
0.012
(2.29)*
0.021
(2.76)**
(2)
No pyramid/No
crossholdings
-0.781
(-5.32)***
-0.045
(-6.05)***
0.065
(6.57)***
0.013
(0.49)
-0.007
(-0.86)
0.045
(3.61)***
(3)
Only Pyramid
-0.794
(-3.87)***
-0.042
(-3.06)**
0.062
(5.14)***
0.573
(11.62)***
-0.003
(-0.24)
0.034
(1.74)
(4)
Only Cross-holdings
-0.550
(-2.46)*
-0.031
(-1.47)
0.063
(2.52)*
0.600
(6.37)***
-0.021
(-1.03)
-0.001
(-0.05)
(5)
Both structures
-0.189
(-1.03)
-0.047
(-3.36)***
0.041
(2.66)**
-0.375
(-6.28)***
-0.001
(-0.09)
0.036
(1.86)
Control Variables
SIZE
0.017
0.044
0.016
-0.005
0.008
(3.26)**
(4.44)***
(1.09)
(-0.33)
(0.77)
LOSS
-0.103
-0.157
-0.061
-0.205
-0.076
(-5.31)***
(-4.71)***
(-1.57)
(-2.78)**
(-1.56)
LEV
0.095
0.120
0.057
0.189
0.077
(2.32)*
(1.83)
(0.56)
(1.28)
(0.81)
GROWTH
0.140
0.164
0.166
0.182
0.180
(19.34)***
(14.83)***
(5.70)***
(6.60)***
(6.89)***
N
3935
2081
755
425
674
adj. R2
0.473
0.229
0.505
0.459
0.481
Regression model (4) is estimated using pooled cross-section and time-series data for non-financial firms from 2000 to 2008. The models are as follows:
4
2007
(4)
RETit 0 1 C _ NI it 2 C _ NI it 3 POSTt 4 POSTt C _ NI it 5 POSTt C _ NIit
6 ,H CONTROLH
7 ,Y YEARY
it
H 1
Y 2000
where: RETit denotes the firm’s annual stock returns, cumulated from eight months before the end of fiscal year t through four months after the end of fiscal year
t; C_NIit denotes consolidated net income for firm i in year t; C_NI it is the change in consolidated net income per share during each accounting period t; POSTt
equals to 1 when observations are under the control-based approach (after 2005) and zero when observations are under the ownership-based approach (before
2005); YEARt is a time indicator that equals one if an observation is from fiscal year t and zero otherwise. All regression variables are deflated by the number of
shares outstanding at year end. t-statistics are given in parentheses and are adjusted for heteroscedasticity. ***, ** and * denote significance at the 1%, 5% and
10% levels, respectively, in a two-tailed test.
43
Table 6 Value relevance of asset, liability and earnings arising from the parent firm and
its subsidiaries
(1)
Whole Sample
Intercept
P_ASSETit
S_ASSETit
P_LIABILITYit
S_LIABILITYit
P_NIit
S_NIv
NCIit
POSTt
POSTt P_ASSETit
POSTt S_ASSETit
POSTt P_ LIABILITYit
POSTt S_ LIABILITYit
POSTt NCIit
POSTt P_NIit
POSTt S_NIit
-29.272
(-13.40)***
0.579
(10.66)***
0.399
(6.41)***
-0.614
(-7.24)***
-0.649
(-7.72)***
-0.471
(-3.01)**
-0.415
(-2.53)*
-0.180
(-0.77)
-3.639
(-4.19)***
0.372
(5.74)***
0.464
(6.12)***
-0.554
(-6.69)***
-0.406
(-4.56)***
-0.364
(-1.77)
2.936
(14.30)***
1.878
(5.71)***
(2)
No pyramid/No
crossholdings
-42.535
(-11.62)***
0.473
(6.11)***
0.493
(5.48)***
-0.574
(-4.94)***
-0.526
(-4.32)***
-0.771
(-3.50)***
0.096
(0.41)
-0.796
(-2.31)*
0.000
(.)
0.385
(3.94)***
0.337
(2.90)**
-0.597
(-4.95)***
-0.344
(-2.52)*
-0.240
(-0.79)
3.131
(10.41)***
2.368
(4.95)***
(3)
Both structures
-25.309
(-6.80)***
0.892
(8.99)***
0.555
(5.28)***
-0.969
(-5.63)***
-1.079
(-7.23)***
-0.315
(-1.17)
-0.946
(-3.26)**
-0.448
(-1.09)
1.573
(0.80)
0.025
(0.20)
0.220
(1.80)
0.013
(0.08)
0.011
(0.07)
-0.316
(-0.92)
2.471
(6.31)***
1.954
(3.14)**
Control Variables
SIZE, LOSS, LEV,
YES
YES
YES
GROWTH
N
3935
2081
674
adj. R2
0.729
0.716
0.788
where: MVit denotes the market value of firm i four months after the end of fiscal year t; P_ASSETit
denotes parent firm i’s total operating assets and non-controlling financial assets at the end of year t;
S_ASSETit denotes subsidiaries’ assets for firm i at the end of year t; P_LIABILITYit denotes parent firm
i’s liability at the end of year t; S_LIABILITYit denotes subsidiaries’ liabilities for firm i at the end of year t;
P_NIit denotes parent firm i’s total income from its own operations and income from non-controlling
financial assets; S_NIit denotes the proportionate share of subsidiaries’ total income for firm i. POSTt
equals to 1 when observations are under the control-based approach (after 2005) and zero when
observations are under the ownership-based approach (before 2005). All regression variables are deflated
by the number of shares outstanding at year end. t-statistics are given in parentheses and are adjusted for
heteroscedasticity. ***, ** and * denote significance at the 1%, 5% and 10% levels, respectively, in a twotailed test.
44