Decision Usefulness of the Equity Method of
Accounting
by
Amanda L. Gonzales
Business Administration
Duke University
Date:
Approved:
Katherine Schipper, Supervisor
Jennifer Francis
John Graham
William Mayew
Per Olsson
Dissertation submitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Business Administration
in the Graduate School of Duke University
2013
Abstract
Decision Usefulness of the Equity Method of Accounting
by
Amanda L. Gonzales
Business Administration
Duke University
Date:
Approved:
Katherine Schipper, Supervisor
Jennifer Francis
John Graham
William Mayew
Per Olsson
An abstract of a dissertation submitted in partial fulfillment of the requirements for
the degree of Doctor of Philosophy in Business Administration
in the Graduate School of Duke University
2013
c 2013 by Amanda L. Gonzales
Copyright All rights reserved except the rights granted by the
Creative Commons Attribution-Noncommercial Licence
Abstract
I examine the decision usefulness of the equity method of accounting from two perspectives. First, I examine the value relevance of information provided under the
equity method relative to the value relevance of information resulting from measuring investments in affiliates at fair value. For a sample of 221 U.S. investors with
publicly-traded affiliates during 1993-2011, I find that balance sheet measures of investments in publicly-traded affiliates provided under the equity method are associated with investors’ stock prices, but income from these affiliates recognized under
the equity method is not associated with investors’ stock prices. In addition, fair
value balance sheet and income measures of investments in publicly-traded affiliates
are incrementally associated with investors’ stock prices after controlling for information provided under the equity method. The incremental value relevance of fair
value measures for investments in publicly-traded affiliates exists for both investments
identified as held for sale and those identified as strategic, with no evidence that the
incremental value relevance is higher (lower) for investments identified as held for
sale (strategic). This result suggests that the 2010 and 2013 proposals by the U.S.
Financial Accounting Standards Board to distinguish between investments in affiliates based on management’s intended method of value realization are not supported
by differences in the relative value relevance of fair value measures for these types of
investments. Second, I evaluate whether equity method investors use their significant
influence to manage income reported under the equity method. For a sample of 202
iv
U.S. firms from 1993-2011, I find that signed discretionary accruals of affiliates are
higher when income from affiliates allows investors to meet earnings targets. This
result is consistent with equity method investors influencing the financial reporting
of affiliates to achieve earnings targets.
v
To my husband, Kevin, whose unending support, encouragement, and sacrificial
love made it possible for me to complete this dissertation.
vi
Contents
Abstract
iv
List of Tables
ix
Acknowledgements
xi
1 Introduction
1
2 The Equity Method of Accounting
8
3 Sample and Data
10
4 Descriptive Statistics
14
4.1
Variables Specific to the Equity Method . . . . . . . . . . . . . . . .
14
4.2
Comparison of Equity Method Investors and Affiliates with the Full
Compustat Population . . . . . . . . . . . . . . . . . . . . . . . . . .
20
Analysis of Changes in Firms When Held As Affiliates . . . . . . . .
21
4.3
5 Value Relevance of Information Provided by the Equity Method
23
5.1
Related Literature and Hypotheses Development . . . . . . . . . . . .
23
5.2
Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
5.3
Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
35
5.4
Supplementary Analyses . . . . . . . . . . . . . . . . . . . . . . . . .
42
5.4.1
Correlation between Investor and Affiliate Returns
. . . . . .
42
5.4.2
Theoretical Values in a Residual Income Model . . . . . . . .
43
vii
6 Earnings Management by Investors with Significant Influence
46
6.1
Related Literature and Hypotheses Development . . . . . . . . . . . .
46
6.2
Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
50
6.3
Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
6.4
Supplementary Analyses . . . . . . . . . . . . . . . . . . . . . . . . .
60
6.4.1
Effect of “True” Significant Influence . . . . . . . . . . . . . .
60
6.4.2
Differences in Earnings Management of Public Affiliates and
Private Affiliates . . . . . . . . . . . . . . . . . . . . . . . . .
63
Potential for Mechanical Relation Between Investor Incentive
and Outcome Variables . . . . . . . . . . . . . . . . . . . . . .
66
6.4.4
Additional Investigation into Effects of Accruals Reversal . . .
67
6.4.5
Other Potential Methods of Managing Earnings from Investments in Affiliates . . . . . . . . . . . . . . . . . . . . . . . . .
68
6.4.6
Reverse Causality/Endogeneity . . . . . . . . . . . . . . . . .
72
6.4.7
Effect of Equity Method Ownership on Accounting Conservatism of Affiliates . . . . . . . . . . . . . . . . . . . . . . . .
72
6.4.3
7 Link between Value Relevance and Earnings Management
78
8 Conclusion
82
A Description of the Cost Method
84
Bibliography
87
Biography
94
viii
List of Tables
3.1
Sample Identification . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
4.1
Descriptive Statistics for Full Sample of Equity Method Investors . .
16
4.2
Descriptive Statistics for Full Sample of Affiliates . . . . . . . . . . .
17
4.3
Investors’ Prior and Subsequent Ownership of Affiliates . . . . . . . .
19
4.4
Industry Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
5.1
Descriptive Statistics for the Analysis of the Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of Investments in Affiliates . . . . . . . . . . . . . . . . . . . . . . . . . .
36
Correlation Coefficients for the Analysis of the Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of Investments in Affiliates . . . . . . . . . . . . . . . . . . . . . . . . . .
37
Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of Investments in Affiliates . . . . . . . . . . . . . . .
40
Effect of the Correlation of Investor and Affiliate Stock Returns on
the Relation between Investors’ Stock Prices and Balance Sheet and
Income Measures of Investments in Affiliates . . . . . . . . . . . . . .
43
Evaluation of Theoretical Values of Coefficients Relating Investors’
Stock Prices to Balance Sheet and Income Measures of Investments
in Affiliates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
6.1
Descriptive Statistics for the Discretionary Accruals Analysis . . . . .
56
6.2
Correlation Coefficients for the Discretionary Accruals Analysis . . .
57
6.3
Relation between Equity Method Ownership and Discretionary Accruals of Affiliates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
5.2
5.3
5.4
5.5
ix
6.4
6.5
6.6
6.7
6.8
7.1
7.2
Effect of “True” Significant Influence on the Relation between Equity
Method Ownership and Discretionary Accruals of Affiliates . . . . . .
64
Effect of Existence of Private Affiliates on the Relation between Equity
Method Ownership and Discretionary Accruals of Affiliates . . . . . .
66
Effect of Defining Investor Incentive Variable as Just Meeting or Beating an Earnings Target on the Relation between Equity Method Ownership and Discretionary Accruals of Affiliates . . . . . . . . . . . . .
67
Relation between Equity Method Ownership and Discretionary Accruals of Affiliates for Investors with INCENTIVEIN V EST ORk,t = 0 in All
Periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
Relation between Equity Method Ownership and Accounting Conservatism of Affiliates . . . . . . . . . . . . . . . . . . . . . . . . . . . .
77
Effect of Earnings Management by Investors on the Relation between
Investors’ Stock Prices and Balance Sheet and Income Measures of
Investments in Affiliates . . . . . . . . . . . . . . . . . . . . . . . . .
80
Effect of Separating Publicly-Traded and Private Affiliates on the Relation between Investors’ Stock Prices and Balance Sheet and Income
Measures of Investments in Affiliates . . . . . . . . . . . . . . . . . .
81
x
Acknowledgements
I thank my dissertation committee members Jennifer Francis, John Graham, Bill
Mayew, Per Olsson, and Katherine Schipper (chair) for valuable comments and direction. I also thank Dirk Black, Thomas Steffen, and workshop participants at Duke
University, Georgetown University, Indiana University, Pennsylvania State University, Texas A&M University, the University of Nebraska—Lincoln, and the University
of Washington for helpful suggestions.
xi
1
Introduction
This paper examines two aspects of the decision usefulness of information provided
by applying the equity method of accounting. First, I examine the value relevance of
balance sheet and income statement information provided under the equity method
relative to the value relevance of balance sheet and income statement information
resulting from measuring investments in affiliates at fair value.1 Second, I evaluate
whether equity method investors use their significant influence to manage income
reported under the equity method.
The equity method of accounting was promulgated in its current form over forty
years ago by APB Opinion No. 18 “The Equity Method of Accounting for Investments in Common Stock” (APB 18). APB 18 preceded the Conceptual Framework
of the U.S. Financial Accounting Standards Board (FASB) and does not specify a
measurement attribute for investments in affiliates.2 Instead, it describes the me1
Throughout this paper, I use the term “affiliate” to refer to a firm that has a corporate investor
applying the equity method of accounting to its investment in that firm.
2
The FASB (ASC Glossary) defines a measurement attribute as the “quantifiable characteristic of
an item that is measured for accounting purposes.” For example, fair value is the “amount at which
an asset (or liability) could be bought (or incurred) or sold (or settled) in a current transaction
between willing parties, that is, other than in a forced or liquidation sale.” In contrast, the carrying
1
chanics of calculating the carrying value for such investments (a calculated value, not
a measurement). Since APB 18 was issued, accounting standard setters have required
fair value measurement for certain financial assets (e.g., SFAS No. 12, Accounting
for Certain Marketable Securities (1975); SFAS No. 115, Accounting for Certain
Investments in Debt and Equity Securities (1993); SFAS No. 133, Accounting for
Derivative Instruments and Hedging Activities (1998)). However, the FASB has not
changed the accounting requirements for investments in affiliates. The durability of
the equity method raises questions about whether some distinctive feature of investments in which investors have the ability to exercise “significant influence” over the
operating and financial policies of investees makes the equity method a decision useful
accounting treatment for these types of investments.
My research aims to shed light on the decision usefulness of the information provided by applying the equity method from two perspectives. First, I examine the
value relevance of balance sheet and income statement information provided under
the equity method relative to the value relevance of information resulting from measuring investments in affiliates at fair value; that is, accounting for them as if they
were trading securities. This analysis can be thought of as a joint test of relevance
and faithful representation, the two fundamental qualitative characteristics of useful
financial information identified in Chapter 3, Qualitative Characteristics of Useful
Financial Information, of the FASB’s Conceptual Framework. For a sample of 221
U.S. investors with publicly-traded affiliates during 1993-2011, I find that balance
sheet measures of investments in publicly-traded affiliates provided under the equity
method are associated with investors’ stock prices, but income from these affiliates
recognized under the equity method is not associated with investors’ stock prices. In
addition, fair value balance sheet and income measures of investments in publiclyvalue of an equity method investment can only be described by explaining how it is calculated; it is
not a characteristic of the underlying equity securities held.
2
traded affiliates are incrementally associated with investors’ stock prices after controlling for information provided under the equity method. This result suggests that
fair value measurement provides incremental relevant information, compared to the
equity method, for investments in affiliates.3
I refine this analysis to incorporate the FASB’s most recent deliberations on the
accounting for investments in affiliates, which reflect the idea that an entity’s plan for
realizing value from its investment in an affiliate (i.e., its business model) affects the
relevance of various measurement attributes. In particular, the FASB’s 2010 Exposure
Draft, Accounting for Financial Instruments and Revisions to the Accounting for
Derivative Instruments and Hedging Activities, proposed restricting the application
of the equity method to cases where the investor has significant influence over the
investee and the operations of the investee are considered related to the investor’s
consolidated operations (“strategic” investments in affiliates). If these two criteria
are not satisfied, the investment would be measured at fair value with changes in
fair value included in net income. The FASB’s 2013 Exposure Draft, Recognition and
Measurement of Financial Assets and Financial Liabilities, revised the proposals to
require investors with the ability to exercise significant influence over an investee to
use the equity method unless the plan is to realize value by selling the investment. In
that case, the investment should be measured at fair value with changes in fair value
included in net income. I separately analyze the incremental value relevance of fair
value measures for investments in affiliates identified as held for sale and strategic. I
find that fair value measurement provides incremental relevant information, compared
to the equity method, for both categories of investments in affiliates. In addition, I
find no evidence that the incremental value relevance of fair value measures is higher
3
I acknowledge that fair value measurements might be less reliable in other samples, including
investments in non-publicly traded affiliates (e.g., investments in some joint ventures). However,
that is a measurement (reliability) issue, separate from the relevance of the information provided
by measuring investments in affiliates at fair value. The issue of reliability is discussed further in
Chapter 5.1.
3
(lower) for investments in affiliates identified as held for sale (strategic). This result
suggests that the FASB’s 2010 and 2013 proposals to distinguish between investments
in affiliates based on management’s intended method of value realization are not
supported by differences in the relative value relevance of fair value measures for
these types of investments.
My second research question views the equity method from an income statement
perspective and examines the neutrality of income reported under the equity method.
Neutrality is identified as a component of faithful representation in the FASB’s Conceptual Framework. Specifically, I investigate whether the equity method achieves
its objective of reducing earnings management. In developing APB 18, a reason
noted for requiring the equity method when investors can exercise significant influence is that the cost method might invite earnings management through the timing of
dividend payments. However, the equity method does not eliminate the potential opportunity for earnings management; it shifts the opportunity to a different channel.
Equity method investors might have the ability to use their “significant influence”
to manage the earnings of affiliates and thereby affect their own income, because a
proportionate share of the affiliates’ income flows into investors’ net income.
I investigate this issue by examining signed discretionary accruals of affiliates,
using a sample of 202 publicly-traded U.S. firms that were affiliates for a portion
of the period 1993-2011. I find that signed discretionary accruals of publicly-traded
affiliates are higher when income from the affiliates allows investors to meet earnings
targets. This result indicates that the equity method does not eliminate concerns
about earnings management (such as those expressed in relation to the cost method);
instead, it changes the mechanism through which earnings are managed. In supplementary analyses, I find evidence that the earnings management result is stronger (a)
for equity method investors that are more likely to exercise significant influence (e.g.,
those that held a subsidiary-level interest in the affiliate prior to application of the
4
equity method, those with ownership of less than 20%, and those with longer holding
periods) and (b) when equity method investors do not have an option to manage
income from private affiliates and therefore might be more likely to manage income
from publicly-traded affiliates even though it might be more difficult or costly.
Two previous studies compare the decision usefulness of the equity method of accounting with that of proportional consolidation for Canadian firms during 1995-2001.
Graham et al. (2003a) and Kothavala (2003) find that proportionate consolidation
better predicts future return on equity and price volatility, respectively. Kothavala
also finds that the equity method is more risk relevant than proportionate consolidation for explaining bond ratings. In addition, Lee et al. (2013) find that analysts’
annual earnings forecasts are less accurate and more dispersed for firms with affiliates
relative to firms without affiliates for a sample of 21,336 U.S firm-year observations
from 1985-2010. The authors suggest that this is due to the lack of disclosure of components of affiliate earnings under the equity method. Related to my study, Graham
et al. (2003b) examine a sample of 55 U.S. investors from 1993-1997 and find that
both the carrying values reported under the equity method and the fair values of investments in publicly-traded affiliates have a positive association with stock prices of
investor firms. The authors also find no association between income from investments
in affiliates reported under the equity method and investors’ stock prices. I confirm
this result for a larger sample and longer time period and also provide evidence of
the incremental value relevance of income from investments in affiliates measured
using fair value. I also contribute to the equity method literature by providing evidence that equity method investors influence the financial reporting of affiliates to
achieve their own earnings objectives; that is, use of the equity method does not eliminate concerns about earnings management. This identification of ex post earnings
management extends previous literature documenting ex ante actions taken by firms
to manage earnings in relation to equity method investments (e.g., Comiskey and
5
Mulford (1986, 1988); Morris and Gordon (2006)).
I also extend the literature investigating the relation between large shareholder
ownership and financial reporting characteristics of investees. The literature finds
divergent effects of large shareholders on financial reporting quality and suggests the
importance of considering the heterogeneity of large shareholders (e.g., Cronqvist and
Fahlenbrach (2009) and Hope (2013)). I contribute to this literature by identifying
the financial reporting requirements of the large shareholder as a potential factor that
can influence the large shareholder’s effect on the financial reporting of its investee.
In particular, I find that the mechanics of the equity method (the flow-through of
affiliate income to investor income) create both an incentive and an opportunity for
earnings management.
Finally, I contribute to the ongoing debate at the FASB and International Accounting Standards Board (IASB) about the appropriateness of business-model-based
accounting for financial instruments. The IASB has promulgated IFRS 9 Financial
Instruments which includes a business-model-based provision for the measurement of
financial assets; the FASB is proposing a similar provision for investments in debt
securities in its project on financial instruments.4 However, the FASB previously
stated an objective to improve and simplify the accounting for financial instruments
by requiring “all financial instruments to be measured at fair value with realized and
unrealized gains and losses recognized in the period in which they occur.”5 These
differences in the standard setters’ decisions illustrate a fundamental difference of
views about whether measurement attributes should be based on management’s intended method of value realization for the items in question (the IFRS 9 position) or
4
IFRS 9 requires entities to measure financial assets at amortized cost subject to impairment if the
assets meet specified criteria, including that they are “held within a business model whose objective
is to hold assets in order to collect contractual cash flows” (para 4.2(a)).
5
FASB website, Accounting for Financial Instruments—Joint Project of the FASB and IASB.
Available at http://www.fasb.org/cs/ContentServer?c=FASBContent C&pagename=FASB%2F
FASBContent C%2F ProjectUpdatePage&cid=1175801889654.
6
whether the same measurement attribute should be applied to similar items regardless of management’s intent. The issue is also part of the FASB’s and IASB’s joint
work to develop conceptual guidance for selecting measurement attributes that satisfy the objectives and qualitative characteristics of financial reporting and contribute
to the reporting of decision useful information. My results suggest that significant
influence might not be a foundation for requiring or permitting the use of a different
measurement method for investments in affiliates than other investments in marketable equity securities (which are required to be measured at fair value under U.S.
GAAP). The results might generalize to other investments in securities accounted for
under the business-model-based guidance in SFAS No. 115. For example, the results
suggest that fair value might be an appropriate measurement attribute for debt securities classified as held to maturity. Even though a firm might intend to realize the
value of a debt security through the collection of cash flows, fair value measurement
might provide a similar level of information about that investment as is provided for
investments that the firm intends to sell.
This paper proceeds as follows. Chapter 2 provides a description of the equity
method of accounting. Chapter 3 summarizes my sample and data. Chapter 4
presents descriptive statistics. Chapters 5 and 6 summarize related literature, formulate hypotheses, outline my research design, discuss the results of my empirical
tests, and present supplementary analyses for the value relevance and earnings management tests, respectively. Chapter 7 describes the link between the value relevance
and earnings management analyses. Chapter 8 summarizes and concludes.
7
2
The Equity Method of Accounting
In March 1971, the Accounting Principles Board issued APB 18, requiring the equity method of accounting to be used by an investor whose investment in voting stock
gives it the ability to exercise significant influence over the operating and financial
policies of an investee. APB 18 describes significant influence and specifies that an
investment of 20% or more indicates significant influence in the absence of evidence
to the contrary and an investment of less than 20% indicates that an investor does
not have significant influence unless it can be demonstrated to exist.
Under the equity method of accounting, the investor initially recognizes its investment at acquisition cost as a single line item in its balance sheet. Subsequent
measurement is based on a calculated value, rather than on a measurement attribute.
Each period, the investor recognizes its proportionate share of the affiliate’s net income (loss) as income (expense) with the offsetting amount increasing (decreasing)
the investment asset balance. Dividends received from the affiliate are recognized as
a reduction in the carrying value of the investment.
In calculating the amount of income from affiliates, the investor eliminates profits
and losses arising from transactions between the investor and affiliate and accounts
8
for any difference between the acquisition cost of the investment and the investor’s
share of the underlying equity in the net assets of the affiliate as if the affiliate were a
consolidated subsidiary. For example, if the fair value of the affiliate’s property, plant
and equipment (PPE) at the time of the investor’s purchase exceeds its carrying
amount, the investor adjusts the affiliate’s income to reflect the additional depreciation expense associated with the higher value of PPE.
If an investor’s share of affiliate losses exceeds the carrying value of its investment,
the investor suspends application of the equity method and does not recognize additional losses (unless it has otherwise committed to provide financial support to the
affiliate). The investor resumes applying the equity method only after its proportionate share of income exceeds any losses that were not recognized during the time the
equity method was suspended.
The investor evaluates its equity method investment (as a whole) to determine
whether an other-than-temporary impairment has occurred. The equity method investor does not separately test the affiliate’s underlying assets for impairment. If
the affiliate recognizes an impairment charge, the investor considers the effect of that
impairment on the investor’s basis difference in the assets giving rise to the affiliate’s
impairment charge.1
1
ASC 323-10-35 (based on APB 18 and EITF Issue No. 08-6, “Equity Method Investment Accounting Considerations) clarifies, “A series of operating losses of an investee or other factors may
indicate that a decrease in value of the investment has occurred that is other than temporary and
that shall be recognized even though the decrease in value is in excess of what would otherwise be
recognized by application of the equity method. A loss in value of an investment that is other than
a temporary decline shall be recognized. Evidence of a loss in value might include, but would not
necessarily be limited to, absence of an ability to recover the carrying amount of the investment
or inability of the investee to sustain an earnings capacity that would justify the carrying amount
of the investment. A current fair value of an investment that is less than its carrying amount may
indicate a loss in value of the investment. However, a decline in the quoted market price below the
carrying amount or the existence of operating losses is not necessarily indicative of a loss in value
that is other than temporary. All are factors that shall be evaluated.”
9
3
Sample and Data
I identify 283 U.S. firms (“equity method investors”) with investments in equity
securities of 414 publicly-traded firms (“affiliates”) accounted for under the equity
method at some point during the period 1993-2011.1 Although the equity method
applies to investments in both publicly-traded and private affiliates, I focus on firms
with publicly-traded affiliates because of the data required for my analyses.2 Table
3.1 summarizes the sample identification process.
I begin by identifying all U.S. firms in the Compustat annual dataset with at least
1
My study is based on U.S. firms. Although similar equity method accounting requirements exist
for international firms, I focus on U.S. firms because of the availability of Schedule 13D data to
assist me in identifying publicly-traded investees. I am exploring the use of the Amadeus database
to expand my sample to European firms.
2
Equity method investors are required (by APB 18) to disclose the fair values of investments in
affiliates only when a quoted market price is available. Investments in affiliates are excluded from
the scope of SFAS No. 107, Disclosures about Fair Value of Financial Instruments, and therefore
no fair value information is required to be disclosed for private affiliates. It is not practicable to
construct an “as if” fair value measure of private affiliates in my sample. One possibility would be
to use a price-earnings ratio. Equity method investors are required to disclose the name of each
affiliate and percentage of ownership of common stock (ASC 323-10-50-3a). However, investors are
not required to provide summarized financial information (including net income) about affiliates
unless the investments are “material” in relation to the financial position or results of operations of
the investor (ASC 323-10-50-3c) and often the summarized financial information presented combines
several affiliates.
10
Table 3.1: Sample Identification
Investors
U.S. firms with at least one equity method investment in Compustat (non-zero
IVAEQ or ESUB) from 1993-2011
Firms with an available 10-K on the SEC website
Firms with U.S. publicly-traded investees identified through Schedule 13D filings
(includes investments in trading and AFS securities, equity method investees,
and subsidiaries)
Final sample of U.S. firms with publicly-traded affiliates (i.e., investees
accounted for using the equity method)
Investees
5,946
4,658
1,291
2,610
283
414
one equity method investment (non-zero, non-missing IVAEQ or ESUB) during fiscal
years 1993-2011. I restrict my sample to begin in 1993 because I review investors
10-K reports from the SEC’s EDGAR system (“SEC website”) to determine which
investments are accounted for under the equity method. Of the 5,946 firms in Compustat with an equity method investment, 4,658 also have at least one 10-K available
on the SEC website.
To identify investors with publicly-traded investees, I write a Perl program that
identifies all Schedule 13D filings (original and amendments) on the SEC website made
by the 4,658 equity method investors during my sample period.3 The Perl program
creates a table with the investor name, investor CIK number, investee name, investee
CIK number, and percentage of the equity class owned for each Schedule 13D filing.
After excluding investees with no 10-K reports on the SEC website and no data
available in Compustat, the sample includes 1,291 investors and 2,610 investees with
3,003 unique investor-investee pairs.
Finally, I examine the investors’ 10-K reports to identify publicly-traded investments accounted for under the equity method. For each of the 3,003 unique investorinvestee pairs identified above, I identify the filing date of the related Schedule 13D
with the highest percentage of ownership of the investee. I then review the investor’s
3
When an investor acquires beneficial ownership of more than five percent of the voting class of a
company’s equity securities that are registered with the SEC, the investor must file a Schedule 13D
with the SEC. Schedule 13D filings include all publicly-traded investments, including those accounted
for as trading securities, available-for-sale securities, equity method securities, and subsidiaries.
11
10-K report for the period including that Schedule 13D filing date. I search for “equity method”, “equity basis”, and “equity in” to identify investments accounted for
by the equity method. I determine which of those investments are publicly traded by
searching for the investees’ names using the “Company Search” feature on the SEC
website.
When I identify an investor with a publicly-traded affiliate, I examine all of that
investor’s 10-K reports available on the SEC website during my sample period to
identify any additional publicly-traded affiliates. For each publicly-traded affiliate
identified, I record the dates that the equity method began and ceased to be applied.4 After excluding affiliates with no data available in Compustat, my final sample
includes data for 283 investors and 414 affiliates during the period 1993-2011.
For each investor 10-K in which a publicly-traded affiliate is identified, I handcollect the following data (where disclosed) for each publicly-traded affiliate: number
of shares held, percentage of ownership, carrying value of investment in affiliate, and
income from investment in affiliate.5 APB 18 (ASC 323-10-50-3b) requires disclosure
of the aggregate value (i.e., fair value) of each identified investment in an affiliate when
a quoted market price is available. I record this value as my measure of the fair value
of an investment in a publicly-traded affiliate when it is available. However, I observe
noncompliance with this disclosure requirement in 54.43% (768 out of 1,411 firmyears) of my sample. In those cases, I calculate the fair value of the investment in a
4
Some firms do not explicitly disclose the start and end dates of the application of the equity
method to a particular investment. In those cases, I assume that the start date is the fiscal year-end
date of the first fiscal year in which the investment appears as an equity method investment. I
assume that the end date is the fiscal year-end date of the last fiscal year in which the investment
appears as an equity method investment. If an investment is accounted for under the equity method
before the start of the 1993 fiscal year, I assume that the start date is the first day of the 1993 fiscal
year. If an investment is accounted for under the equity method after the end of the 2011 fiscal year,
I assume that the end date is the last day of the 2011 fiscal year.
5
If the number of shares held or percentage of ownership is not disclosed, I calculate the amount
based on disclosed information and publicly-available data. For example, if the percentage of ownership is disclosed but the number of shares held is not, I calculate the number of shares held as the
percentage of ownership multiplied by the number of shares outstanding recorded in CRSP.
12
publicly-traded affiliate using information disclosed by the investor (either the number
of shares held in the affiliate or the percentage of shares held in the affiliate) and
publicly-available data from CRSP (the share price of the affiliate and, if necessary,
the total number of outstanding shares of the affiliate). If I cannot calculate the
fair value of an equity method investment using publicly-available data (329/1,411
firm-years), I do not estimate or impute it. I also construct a measure of income from
affiliates as if the investment in the affiliate were remeasured to fair value at each
reporting date (“fair value income”), by multiplying the number of shares held at a
reporting date by the change in the affiliate’s share price during the period.6
The additional data required for my analyses are from Compustat (accounting
data), CRSP (stock price and returns data) and the Thomson Reuters Institutional
Holdings Database (institutional ownership data). I use the institutional investor
classification data available on Brian Bushee’s website to classify institutions as transient; quasi-indexing; monitoring; and dedicated, non-independent.
6
If the equity method investment was acquired during the reporting period, I calculate fair value
income as the number of shares held in the affiliate at the reporting date multiplied by the change
in the price per share of the affiliate from the date that the investment was made to the reporting
date.
13
4
Descriptive Statistics
4.1 Variables Specific to the Equity Method
Tables 4.1 and 4.2 provide descriptive statistics for the 283 equity method investors
and 414 publicly-traded affiliates in my sample. As shown in Table 4.1, equity method
investors invest in an average (median) of 3.90 (2.00) affiliates, including an average
(median) of 1.51 (1.00) publicly-traded affiliates and 2.48 (1.00) private affiliates.1
Equity method investments comprise an average (median) of 11% (5%) of total assets,
including average (median) carrying values of investments in publicly-traded affiliates
of $541.40 million ($58.94 million) and in private affiliates of $445.34 million ($3.70
million). The absolute value of income from affiliates comprises an average (median) of
5% (1%) of the absolute value of net income, including average (median) income from
public affiliates of $1.83 million ($0.05 million) and private affiliates of $4.98 million
($0.00 million). The average (median) difference between fair value and carrying
value of publicly-traded affiliates is $517.76 million ($32.15 million). If income were
1
In an untabulated result, I find that 47.70% (135 out of 283) of the equity method investors in
my sample have both publicly-traded and private affiliates.
14
calculated by remeasuring investments in publicly-traded affiliates to fair value each
reporting period, some firms would report much lower income (25th percentile reports
decrease in income of $22.81 million) and some firms would report much higher income
(75th percentile reports increase in income of $18.23 million).
I find that 22.61% (64 out of 283) of investors and 6.45% (91 out of 1,411) of
investor firm-years in my sample report an impairment of a publicly-traded equity
method investment (untabulated). As shown in Table 4.1, for those firm-years, the
average (median) impairment totals $98.48 million ($23.70 million), which is an average (median) of 44% (37%) of the lagged carrying value of the impaired investment.2
In an untabulated investigation, I find that equity method investors do not recognize
impairments of their investments in 275 cases in which the fair value of the investment
is below its carrying value at the reporting date. The investor often states that it
believes that the decline in fair value is temporary in these cases. In 13.10% of these
cases (36/275), an impairment of the investment is recognized in a future period.
In untabulated results, I also find that 9.54% (27/283) of firms in my sample cease
applying the equity method for at least one investment because of the recognition
of losses in excess of the carrying value of the investment. This represents 4.96%
(70/1,411) of available firm-year observations.
Table 4.2 documents that for my sample of 414 publicly-traded affiliates, the
average (median) equity method holding period during 1993-2011 is 4.17 years (3.25
years) and the average (median) percentage of equity method ownership is 29% (29%).
In untabulated results, I find that 83 of the 392 affiliates with data available have a
percentage of equity method ownership less than 20% and 7 of the 392 affiliates have
a percentage of equity method ownership more than 50%. This suggests that, in some
circumstances, firms “rebut” the presumption in APB 18 that significant influence
2
The number of observations for the Impairments of Public Affiliates/Lagged Book Value variable
is less than 91 because the lagged book value of the affiliate that was impaired is not available in 41
cases.
15
16
|| |
|| |
|| |
|| |
26240.64
555.62
-0.02
17313.58
3.53
0.33
0.51
0.57
0.11
11.73
24.44
0.11
0.30
0.01
0.07
0.50
0.03
3.90
1.51
2.48
1145.77
541.40
445.34
5.83
1.83
4.98
0.11
0.09
0.03
0.05
0.05
0.01
517.76
0.11
-43.41
3.81
98.48
0.44
0.49
Mean
103220.07
3957.32
0.44
40307.02
19.90
0.41
0.42
0.99
0.37
12.39
17.01
0.09
0.17
0.02
0.07
0.27
0.07
5.89
1.42
5.60
3421.79
1738.93
1659.46
54.26
15.90
40.68
0.16
0.16
0.07
0.29
0.36
0.06
1654.60
0.49
616.48
36.44
178.12
0.37
1.07
Std Dev
727.11
-3.87
-0.01
436.88
1.15
0.14
0.14
0.00
0.02
0.00
10.00
0.04
0.16
0.00
0.02
0.28
0.00
1.00
1.00
0.00
24.91
12.70
0.00
-0.24
-0.13
0.00
0.02
0.01
0.00
0.00
0.00
0.00
0.94
0.00
-22.81
0.07
4.25
0.19
0.05
p25
3697.00
70.20
0.03
2223.79
1.88
0.29
0.43
0.16
0.05
8.00
20.00
0.09
0.34
0.00
0.06
0.56
0.00
2.00
1.00
1.00
128.76
58.94
3.70
0.25
0.05
0.00
0.05
0.03
0.00
0.01
0.01
0.00
32.15
0.01
-0.14
0.33
23.70
0.37
0.10
Median
p75
16941.95
452.78
0.06
11318.16
3.10
0.44
0.83
0.88
0.10
20.00
39.00
0.15
0.42
0.00
0.11
0.70
0.01
4.00
2.00
3.00
793.95
314.10
160.59
3.74
1.27
0.11
0.12
0.08
0.03
0.03
0.02
0.01
202.65
0.06
18.23
1.09
107.88
0.64
0.51
Equity Method Investors with Publicly-Traded Affiliates
166,670
19,196
166,670
166,670
166,056
144,355
144,061
165,363
144,771
156,114
155,472
166,670
166,670
92,506
92,918
84,429
90,028
97,973
22,806
N
6439.22
107.14
-2.40
2565.39
3.49
1.13
0.57
0.26
2.55
3.98
13.31
0.10
0.23
0.01
0.05
0.39
0.02
64229.98
1087.54
337.19
13976.06
340.27
69.85
3.85
1.60
342.25
6.92
12.87
0.19
1.14
0.03
0.08
1.27
0.06
Std Dev
27.03
-2.98
-0.09
24.57
0.92
0.03
0.15
0.00
0.03
0.00
4.00
0.01
0.05
0.00
0.00
0.10
0.00
p25
189.27
2.10
0.01
127.53
1.69
0.19
0.38
0.00
0.07
0.00
9.00
0.06
0.18
0.00
0.02
0.33
0.00
Median
Full Compustat Population
Mean
1077.69
27.69
0.06
744.76
3.12
0.39
0.77
0.20
0.14
5.00
18.00
0.15
0.36
0.00
0.09
0.64
0.01
p75
This table reports pooled annual descriptive statistics for 283 U.S. firms with publicly-traded affiliates and for the full Compustat population (all observations with
total assets and net income data available) from 1993-2011. Bold-faced font for the mean (median) values represents differences between the mean (median) value of
that variable for my sample vs. the full Compustat population that are significant at a two-tailed p-value of 0.05 or better. The variables are defined following Table
4.2.
|| |
1,411
283
|
|
|
Firm-years
Firms
|
|
|
|
1411
1411
1411
1357
1354
1408
1304
1389
1401
1411
1411
965
965
916
950
988
468
1052
1397
1046
1272
1007
996
1339
859
857
1272
1007
996
1339
859
857
1082
1082
659
659
91
50
91
Total Assets (in millions)
Net Income (in millions)
Return on Assets
Market Value of Equity (in millions)
Market-to-Book Ratio
Debt
PPE/Total Assets
Dividends per share
Total Accruals /Total Assets
Analyst Coverage
Firm Age
TRA Ownership
QIX Ownership
DNI Ownership
MON Ownership
Total Institutional Ownership
Management Ownership
Total Number of Affiliates Owned
Number of Public Affiliates
Number of Private Affiliates
Book Value of Affiliates (in millions)
Book Value of Public Affiliates (in millions)
Book Value of Private Affiliates (in millions)
Income from Affiliates (in millions)
Income from Public Affiliates (in millions)
Income from Private Affiliates (in millions)
Book Value of Affiliates/Total Assets
Book Value of Public Affiliates/Total Assets
Book Value of Private Affiliates/Total Assets
Income from Affiliates / NI
Income from Public Affiliates / NI
Income from Private Affiliates / NI
FV less BV of Public Affiliates (in millions)
FV less BV of Public Affiliates/Total Assets
FV less Reported Public Affiliate Income (in millions)
FV less Reported Public Affiliate Income / NI
Impairments of Public Affiliates (in millions)
Impairments of Public Affiliates/Lagged Book Value
Impairments of Public Affiliates / NI
N
Table 4.1: Descriptive Statistics for Full Sample of Equity Method Investors
17
0.35
0.26
0.28
0.30
0.29
84
75
112
35
86
414
Firms
2248.27
44.81
1.10
1795.19
24.44
0.33
0.47
0.22
2.69
4.13
9.19
0.08
0.16
0.01
0.05
0.32
0.02
4.17
0.33
0.26
0.29
0.28
767.57
17.48
0.04
790.92
5.16
0.45
0.75
0.00
0.19
2.40
6.50
0.10
0.21
0.00
0.07
0.45
0.02
0.13
0.10
0.12
0.12
0.15
0.11
0.11
0.14
0.12
2.98
2.89
2.90
5.67
2.39
0.13
2.51
3.58
3.09
3.80
6588.43
371.97
39.34
5321.37
458.52
0.46
0.42
0.55
43.45
5.13
9.93
0.08
0.14
0.06
0.09
0.25
0.04
3.27
Std Dev
0.26
0.20
0.19
0.21
0.18
0.27
0.21
0.20
0.20
1.72
1.75
1.84
3.16
1.79
0.21
1.66
1.33
1.84
2.66
80.13
-20.67
-0.15
72.06
0.95
0.08
0.14
0.00
0.05
0.00
3.50
0.02
0.05
0.00
0.00
0.10
0.00
1.84
p25
0.36
0.26
0.28
0.30
0.28
0.34
0.24
0.28
0.28
3.01
2.89
3.73
6.00
3.00
0.29
2.75
2.99
2.97
4.42
359.22
0.54
0.00
325.96
1.76
0.26
0.36
0.00
0.09
2.00
6.00
0.06
0.13
0.00
0.02
0.29
0.00
3.25
Median
p75
0.43
0.34
0.38
0.36
0.38
0.40
0.34
0.39
0.38
5.00
4.59
5.38
11.51
5.35
0.38
4.71
5.19
5.00
6.71
1521.35
32.52
0.04
1088.69
3.52
0.47
0.73
0.12
0.16
6.25
10.50
0.13
0.25
0.00
0.08
0.50
0.01
5.38
Period as an Affiliate (“Event” Period)
Mean
87
28
176
101
125.92
-0.22
-0.00
201.43
2.33
0.23
0.40
0.00
0.10
0.00
2.50
0.04
0.10
0.00
0.02
0.20
0.00
p75
3.86
3.74
4.03
7.29
3.82
0.29
22.89
-6.32
-0.21
44.36
1.34
0.06
0.17
0.00
0.06
0.00
1.00
0.01
0.02
0.00
0.00
0.05
0.00
Median
90
78
117
35
94
392
3231.77
160.09
1.64
3374.23
12.49
0.45
0.47
0.40
1.53
3.97
9.52
0.08
0.17
0.05
0.06
0.27
0.05
p25
3.47
3.91
3.90
5.28
|
1073.55
7.44
-0.34
1184.64
3.27
0.34
0.51
0.15
0.28
1.97
6.15
0.07
0.14
0.01
0.05
0.27
0.02
Std Dev
92
29
184
109
|
414
414
414
384
384
413
395
404
412
414
414
309
309
302
306
312
65
414
Mean
Period Before Becoming an Affiliate (“Pre” Period)
Total Assets (in millions)
Net Income (in millions)
Return on Assets
Market Value of Equity (in millions)
Market-to-Book Ratio
Debt
PPE/Total Assets
Dividends per share
Total Accruals /Total Assets
Analyst Coverage
Firm Age
TRA Ownership
QIX Ownership
DNI Ownership
MON Ownership
Total Institutional Ownership
Management Ownership
Holding Period
Prior Ownership:
Subsidiary
Less Than Significant Influence
None
Other
Subsequent Ownership:
Subsidiary
Less Than Significant Influence
None
End of Sample Period
Other
Equity Method Percentage
Prior Ownership:
Subsidiary
Less Than Significant Influence
None
Other
Subsequent Ownership:
Subsidiary
Less Than Significant Influence
None
End of Sample Period
Other
N
Table 4.2: Descriptive Statistics for Full Sample of Affiliates
1417.20
43.43
2.25
841.30
46.62
-0.03
-0.04
0.09
3.92
2.44
3.96
0.02
0.04
0.01
0.01
0.08
-0.01
3.81
1.68
0.73
2.83
0.94
-1.02
-2.50
3.03
1.14
8.54
34.97
2.44
2.83
1.71
1.78
3.94
-2.09
0.094
0.467
0.005
0.348
0.307
0.013
0.003
0.256
.001
.001
0.016
0.006
0.091
0.078
.001
0.049
.001
Paired t-tests for
“Event” vs. “Pre” Periods
Difference
t-stat
p-value
18
|
3740.15
10.45
-0.42
1869.60
4.65
0.94
0.55
0.33
0.50
5.27
14.37
0.12
0.25
0.00
0.06
0.45
0.02
14318.74
867.61
1.93
5423.78
31.75
6.75
0.66
0.96
2.31
6.63
10.76
0.10
0.20
0.02
0.08
0.33
0.05
Std Dev
84.48
-16.88
-0.16
52.76
0.78
0.11
0.15
0.00
0.06
0.00
8.00
0.02
0.06
0.00
0.00
0.11
0.00
p25
585.82
0.32
-0.00
268.79
1.63
0.28
0.42
0.00
0.10
2.56
11.00
0.11
0.21
0.00
0.04
0.40
0.00
Median
1934.52
51.25
0.04
1280.19
2.84
0.47
0.81
0.12
0.17
8.00
16.00
0.20
0.42
0.00
0.09
0.76
0.01
p75
1590.30
-25.95
-0.15
240.50
1.94
0.61
0.09
0.10
0.26
1.17
5.35
0.05
0.10
-0.01
0.02
0.15
-0.00
3.04
-0.42
-1.31
0.97
0.89
1.39
2.21
1.90
1.79
3.59
34.33
5.49
7.78
-2.09
3.77
8.06
-0.76
0.003
0.673
0.190
0.334
0.373
0.167
0.028
0.058
0.074
.001
.001
.001
.001
0.038
.001
.001
0.455
Paired t-tests for
“Post” vs. “Event” Periods
Difference
t-stat
p-value
3536.70
35.07
-0.27
1685.50
1.49
0.92
0.11
0.18
0.11
3.38
8.81
0.09
0.20
-0.01
0.02
0.29
-0.02
2.48
0.68
-1.31
2.71
0.63
1.21
1.70
2.46
1.37
6.33
37.52
5.77
6.92
-0.88
1.89
5.48
-1.28
0.014
0.500
0.193
0.008
0.529
0.228
0.091
0.016
0.173
.001
.001
.001
.001
0.384
0.065
.001
0.243
Paired t-tests for
“Post” vs. “Pre” Periods
Difference t-stat
p-value
|
|| |
|
|
|
|| |
|
|| |
|
|
The variables used in Tables 4.1 and 4.2 are defined as follows. Total Assets: total assets (AT) of firm j in year t. Net Income: net income (NI) of firm j in year t. Return on Assets: return on
assets (NI/AT) of firm j in year t. Market Value of Equity: market value of equity (CSHO*PRCC F) of firm j in year t. Market-to-Book Ratio: market-to-book ratio (CSHO*PRCC F/CEQ)
of firm j in year t. Debt: debt-to-assets ratio (DLTT+DLC/AT) of firm j in year t. PPE/Total Assets: ratio of PPE to total assets (PPEGT/AT) of firm j in year t. Dividends per share:
dividends per share (DVPSX F) of firm j in year t. Total Accruals /Total Assets: ratio of the absolute value of total accruals to total assets ( NI-OANCF /AT) of firm j in year t. Analyst
Coverage: number of analysts in I/B/E/S issuring forecasts for firm j in year t. Firm Age: firm age, measured as the number of years that firm j has been listed on Compustat in year t.
TRA Ownership: percentage ownership of firm j at the end of year t by institutional investors described as transient by Bushee (2001). QIX Ownership: percentage ownership of firm j at the
end of year t by institutional investors described as quasi-indexing by Bushee (2001). DNI Ownership: percentage ownership of firm j at the end of year t by institutional investors described
as dedicated by Bushee (2001) and non-independent by Brickley et al. (1988). MON Ownership: percentage ownership of firm j at the end of year t by institutional investors described as
monitoring institutions in Ramalingegowda and Yu (2012) based on classification as dedicated by Bushee (2001) and independent by Brickley et al. (1988). Total Institutional Ownership:
percentage ownership of firm j at the end of year t by all institutional investors. Management Ownership: percentage ownership (excluding options) of firm j at the end of year t by the CEO
of firm j. Total Number of Affiliates Owned: total number of affiliates owned by firm j in year t. Book Value of Affiliates: book value (IVAEQ) of firm j’s investments in affiliates in year
t. Income from Affiliates: income (ESUB) from firm j’s affiliates in year t. Book Value of Affiliates/Total Assets: book value (IVAEQ) of firm j’s investments in affiliates in year t divided
by total assets (AT) of firm j in year t. Income from Affiliates / NI : absolute value of income (ESUB) from firm j’s affiliates in year t divided by the absolute value of net income (NI) of
firm j in year t. FV less BV of Public Affiliates: fair value of firm j’s investments in publicly-traded affiliates less the carrying value of those affiliates in year t. FV less BV of Public
Affiliates/Total Assets: FV less BV of Affiliate divided by total assets (AT) of firm j in year t. FV less Reported Public Affiliate Income: fair value income from firm j’s publicly-traded
affiliates less reported income for those affiliates in year t, where fair value income is calculated as the number of shares held in an affiliate at the end of year t multiplied by the change in the
fair value of the affiliate’s share price during year t. FV less Reported Public Affiliate Income / NI : absolute value of FV less Reported Affiliate Income divided by the absolute value of
net income (NI) of firm j in year t. Impairments of Public Affiliates: impairments of publicly-traded affiliates reported by firm j in year t. Statistics include only those firms reporting an
impairment. Impairments of Public Affiliates/Lagged Book Value: Impairments of Public Affiliates of firm j in year t divided by the book value of the impaired affiliates in year t - 1.
Impairments of Public Affiliates / NI : absolute value of Impairments of Public Affiliates divided by the absolute value of net income (NI) of firm j in year t. Holding Period: number of
years in which firm j is held as an affiliate in the period from 1993-2011. Equity Method Percentage: percentage ownership of firm j in year t by an equity method owner.
This table reports annual descriptive statistics for 414 U.S. publicly-traded firms from 1993-2011. The table reports separate measures for the period prior to the firms becoming affiliates (the
“pre” period), the period in which the firms are affiliates (the “event” period), and the period after the firms are affiliates (the “post” period). The table presents the average of the average values
for each firm. That is, the average value for a variable is calculated separately for a firm over each period (“pre”, “event”, and “post”). Then the table presents the averages of these values over
the 414 firms in the sample. The table also presents the results of paired t-tests comparing the “event” period to the “pre” period, the “post” period to the “event” period, and the “post” period
to the “pre” period. Bold-faced font for the mean (median) values during the “event” period represents differences between the mean (median) value of that variable for my sample vs. the full
Compustat population that are significant at a two-tailed p-value of 0.05 or better. Descriptive statistics for the full Compustat population are presented in Table 4.1.
|
Total Assets (in millions)
Net Income (in millions)
Return on Assets
Market Value of Equity (in millions)
Market-to-Book Ratio
Debt
PPE/Total Assets
Dividends per share
Total Accruals /Total Assets
Analyst Coverage
Firm Age
TRA Ownership
QIX Ownership
DNI Ownership
MON Ownership
Total Institutional Ownership
Management Ownership
Mean
Period After Becoming an Affiliate (“Post” Period)
exists when ownership is between 20-50%.
I also analyze the relationships of the investors and affiliates prior to and after
application of the equity method. The results are summarized in Table 4.3. I find that
prior to application of the equity method, the investor has no holding in the affiliate
in 44.44% of cases, a holding of less than significant influence in 7.00% of cases, and
a subsidiary-level holding in 22.22% of cases. After application of the equity method,
the investor retains no interest in the affiliate in 28.26% of cases, a holding of less
than significant influence in 18.84% of cases, and a subsidiary-level holding in 21.74%
of cases.
Table 4.3: Investors’ Prior and Subsequent Ownership of Affiliates
Investors’ Prior Ownership of Affiliates
# of Affiliates
% of Affiliates
Subsidiary
Less than significant influence
None
Debtholder
Held shares in another entity that was reorganized to form affiliate
Unknown: no previous investor 10Ks available
Unknown: could not determine from available investor 10Ks
92
29
184
4
13
10
82
22.22
7.00
44.44
0.97
3.14
2.42
19.81
Total
414
Investors’ Subsequent Ownership of Affiliates
# of Affiliates
% of Affiliates
Subsidiary
Less than significant influence
None
Distributed to shareholders as dividend
Bankruptcy/dissolution of affiliate
Reorganization/transfer/exchange of shares of affiliate
Unknown: no further investor 10Ks available (prior to end of sample period)
Unknown: end of sample period
Unknown: could not determine from available investor 10Ks
90
78
117
5
10
8
38
35
33
21.74
18.84
28.26
1.21
2.42
1.93
9.18
8.45
7.97
Total
414
Table 4.4 shows the distribution of equity method investors and affiliates across
industries. In untabulated results, I find that 50% (39%) of publicly-traded affiliates
are in the same 1-digit (2-digit) SIC code as their investors.
19
Table 4.4: Industry Analysis
Industry
Agriculture, Forestry, and Fishing
Mining and Construction
Manufacturing
Manufacturing
Transportation and Public Utilities
Trade, Wholesale, and Retail
Finance, Insurance, and Real Estate
Services
Services
Public Administration
Missing
Total
1-digit
SIC
# of
Investors
% of
Investors
# of
Affiliates
% of
Affiliates
# of Full
Compustat
Sample
% of Full
Compustat
Sample
0
1
2
3
4
5
6
7
8
9
-
0
18
47
44
56
19
54
27
11
3
4
0.00
6.36
16.61
15.55
19.79
6.71
19.08
9.54
3.89
1.06
1.41
3
23
38
62
73
28
64
79
10
3
31
0.72
5.56
9.18
14.98
17.63
6.76
15.46
19.08
2.42
0.72
7.49
51
919
2,136
3,724
1,706
1,644
2,894
3,024
929
655
1,514
0.26
4.79
11.13
19.40
8.89
8.56
15.08
15.75
4.84
3.41
7.89
283
414
19,196
The full Compustat sample represents all Compustat firms from 1993-2011 with nonmissing total assets and net
income.
4.2 Comparison of Equity Method Investors and Affiliates with the
Full Compustat Population
Tables 4.1 and 4.2 also compare accounting, market, and ownership measures
of equity method investors and affiliates with the full Compustat population (those
observations with minimum data requirements of total assets and net income).3 Bold
font in the mean and median columns of Table 4.1 and the “event” period of Table
4.2 represents the existence of a difference between investors (Table 4.1) or affiliates
(Table 4.2) and the Compustat population that is significant at a two-tailed p-value
of 0.05 or better. As shown in Table 4.1, equity method investors are larger than
other firms in Compustat both in terms of total assets (average of $26,240.64 million
compared to $6,439.22 million) and in terms of market value of equity (average of
$17,313.58 million compared to $2,565.39 million). They are also more profitable in
terms of net income (average of $555.62 million compared to $107.14 million) and
return on assets (average of -0.02% compared to -2.40%). In addition, equity method
3
In untabulated results, I find that 31% of firms in Compustat (5,946/19,196) during the period
1993-2011 have an equity method investment. I compare my sample of equity method investors with
publicly-traded affiliates to this sample of all equity method investors in Compustat. I find that my
sample firms are an average of 8.3 years older and have an average of 4.2 more analysts following
them relative to the full population of equity method investors. I also find that the ratio of the
book value of affiliates to total assets for my sample firms is 5.5% higher on average than that for
all equity method investors. I do not find differences in other measures, including total assets, net
income, dividends per share, and institutional ownership at a two-tailed p-value of 0.10 or better.
20
investors have a higher institutional ownership percentage (average of 50% compared
to 39%) and more analyst following (average of 11.73 analysts compared to 3.98
analysts).
As shown in Table 4.2, although the average affiliate is smaller than the average firm in the Compustat population (total assets of $2,248.27 million compared to
$6,439.22 million), the median affiliate is larger than the median firm in the Compustat population (total assets of $359.22 million compared to $189.27 million). In
addition, I find that the average affiliate is younger (firm age of 9.19 years compared
to 13.31 years), has more analyst coverage (4.13 analysts compared to 3.98 analysts),
and has less institutional ownership (32% compared to 39%) relative to the average
firm in the Compustat population.
4.3 Analysis of Changes in Firms When Held As Affiliates
Finally, Table 4.2 documents accounting, market, and ownership measures for the
period before a firm became an affiliate (the “pre” period), the period in which the
firm is an affiliate (the “event” period), and the period after the firm is an affiliate
(the “post” period). The results of paired t-tests indicate that firms become larger
in the “event” period relative to the “pre” period both in terms of total assets (an
average increase of $1,417.20 million, p-value
0.001) and market value of equity
(an average increase of $841.30 million, p-value = 0.005). This trend continues in
the “post” period, with an average additional growth in total assets of $1,590.30
million (p-value = 0.003). Firms also appear to attract more institutional ownership
and more analyst attention when they become affiliates, with institutional ownership
0.001) and analyst coverage increasing an
average of 2.44 analysts per firm (p-value 0.001). This trend also continues during
increasing an average of 8% (p-value
the “post” period, with institutional ownership increasing an average of an additional
15% (p-value
0.001) and analyst coverage increasing an average of an additional
21
1.17 analysts per firm (p-value
0.001). In contrast, firms do not exhibit changes in
net income or return on assets from the “pre” period to the “post” period (two-tailed
p-values = 0.500 and 0.193, respectively).
I examine several potential manifestations of “significant influence” by equity
method investors. First, I calculate a measure of leverage (debt-to-assets) to evaluate
whether equity method owners are prompting changes in affiliates’ financing strategies. I do not find evidence supporting this type of influence (p-value = 0.307 for the
paired t-test from the “pre” period to the “event” period). I find an average decrease
of 4% in the ratio of property, plant, and equipment to total assets from the “pre”
period to the “event” period (p-value = 0.013). This might reflect changes in the
operating structure of the affiliate. I also find that dividends increase an average of
$0.09 per share (p-value = 0.003), which represents a more than 50% increase from
the average of $0.15 dividends per share in the “pre” period.4 This is consistent with
concerns expressed during the development of APB 18 that firms with “significant
influence” could affect the amounts and timing of dividend payments. However, the
upward trend in dividends per share continues during the “post” period, with an
average additional increase in dividends per share of $0.10 (p-value = 0.058), which
might indicate that the increase in dividends is due to factors other than equity
method ownership.
4
In my sample, 161 firms do not pay dividends prior to becoming affiliates. Of those 161 firms, 18
begin paying dividends when they become affiliates.
22
5
Value Relevance of Information Provided by the
Equity Method
5.1 Related Literature and Hypotheses Development
Debate about the appropriate accounting for investments in affiliates predates the
issuance of APB 18 over 40 years ago.1 APB 18 (para. 7) describes the concerns about
the cost method as follows (see Appendix A for a description of the cost method):
“Financial statements of an investor prepared under the cost method may not reflect
substantial changes in the affairs of an investee. Dividends included in income of
an investor for a period may be unrelated to the earnings (or losses) of an investee
for that period. For example, an investee may pay no dividends for several periods
1
Nobes (2002) describes the international development of the equity method. He states that
the earliest uses of the equity method were in the early 1900s as a way to include subsidiaries
in parents’ financial statements before the practice of consolidation was fully established. After
consolidated financial statements became common, some jurisdictions still required/permitted the
use of the equity method in parent company financial statements. The equity method is described as
a “useful arithmetic device for preparing consolidated balance sheets when there are several layers of
subsidiaries” (Nobes, page 19). The application of the equity method to joint ventures and affiliates
arose during the 1960s as this type of ownership structure became more common. Nobes (page 40)
states that the “uses [of the equity method for inclusion of joint ventures and associates in investor
statements or consolidated statements] seem to have arisen with little theoretical justification and
no prior research into their usefulness.”
23
and then pay dividends substantially in excess of the earnings of a period. Losses of
an investee of one period may be offset against earnings of another period because
the investor reports neither in results of operations at the time they are reported by
the investee. Some dividends received from an investee do not cover the carrying
costs of an investment whereas the investor’s share of the investee’s earnings more
than covers those costs. Those characteristics of the cost method may prevent an
investor from reflecting adequately the earnings related to an investment in common
stock—either cumulatively or in the appropriate periods.” However, APB 18 (para.
13) also indicates that some supported the cost method because “the investor is not
entitled to recognize earnings on its investment until a right to claim the earnings
arises, and that claim arises only to the extent dividends are declared. The investor
is considered to have no earnings on its investment unless it is in a position to control
the distribution of earnings. Likewise, an investment or an investor’s operations are
not affected by losses of an investee unless those losses indicate a loss in value of the
investment that should be recognized.”
APB 18 (paras. 10 and 12) indicates that the Accounting Principles Board believed that the equity method provides useful information to investors, stating “the
equity method is an appropriate means of recognizing increases or decreases measured
by generally accepted accounting principles in the economic resources underlying the
investments. Furthermore, the equity method of accounting more closely meets the
objectives of accrual accounting than does the cost method since the investor recognizes its share of the earnings and losses of the investee in the periods in which
they are reflected in the accounts of the investee. ... The equity method tends to be
most appropriate if an investment enables the investor to influence the operating or
financial decisions of the investee. The investor then has a degree of responsibility
for the return on its investment, and it is appropriate to include in the results of
operations of the investor its share of the earnings or losses of the investee.”
24
However, APB 18 also describes a market value method analogous to fair value
measurement and states that the market value method “is considered to meet most
closely the objective of reporting the economic consequences of holding the investment” (para. 9). At the time that APB 18 was issued, the market value method
was used only in special circumstances. APB 18 (para. 9) states, “While the Board
believes the market value method provides the best presentation of investments in
some situations, it concludes that further study is necessary before the market value
method is extended beyond current practice.” The APB did not describe the nature
of that further study.
Since the issuance of APB 18 in 1971, accounting standard setters have required
fair value measurement for certain financial assets (e.g., SFAS No. 12, Accounting
for Certain Marketable Securities (1975); SFAS No. 115, Accounting for Certain
Investments in Debt and Equity Securities (1993); SFAS No. 133, Accounting for
Derivative Instruments and Hedging Activities (1998)). It is also the stated objective
of the FASB to improve and simplify the accounting for financial instruments by
requiring “all financial instruments to be measured at fair value with realized and
unrealized gains and losses recognized in the period in which they occur.”2
However, the FASB has not required fair value measurement for investments in
affiliates.3 One possibility is that the FASB is concerned about the reliability or
difficulty of fair value measurements for investments in affiliates that are not publiclytraded (as discussed in Chapter 4.1, many firms in my sample have private affiliates).
2
FASB website, Accounting for Financial Instruments —Joint Project of the FASB and IASB.
Available at http://www.fasb.org/cs/ContentServer?c=FASBContent C&pagename=FASB%2F
FASBContent C%2FProjectUpdatePage&cid=1175801889654.
3
SFAS No. 159, The Fair Value Option for Financial Assets and Financial Liabilities (2007),
permits entities to measure investments otherwise accounted for under the equity method at fair
value with realized and unrealized gains and losses recognized in earnings at each reporting date. I
have identified one firm (out of 283) applying the fair value option to an investment that otherwise
would be accounted for under the equity method. The FASB’s 2013 Exposure Draft would remove
the option to measure investments in affiliates at fair value and instead require investments in
affiliates classified as held for sale to be measured at fair value.
25
However, APB 18 and the most recent FASB discussions on the equity method did
not raise the reliability of fair value measurements as a basis for retaining the equity
method. In fact, the 2013 Exposure Draft would require fair value measurement
for investments in affiliates that an entity intends to sell regardless of whether the
affiliate is public or private. Other standards on the accounting for financial assets
issued since APB 18 tend to exclude investments in affiliates from their scopes. For
example, SFAS 115 (para. 44) states, “The Board decided to limit the scope of the
project because of its desire to expedite resolution of the problems with the current
accounting and reporting practices for investment securities. Accordingly, the Board
decided to address the accounting for only certain financial assets and not to change
the accounting for financial liabilities nor include other assets.”4 In summary, if the
continued existence of the equity method is based on the reluctance of the FASB to
require fair value measurement for financial assets without active markets, the FASB
has not articulated that as a basis.
This raises questions about whether some distinctive feature of investments in
which “significant influence” can be exercised makes fair value a less relevant measurement attribute than for other investments in equity securities. The FASB’s most
recent discussions on the accounting for investments in affiliates reflect the possibility
that management’s plan for realizing value from the investment (i.e., the business
model) affects the relevance of various measurement attributes. For example, investments in affiliates might be undertaken for long-term operating or strategic purposes,
and thus, the method of value realization might differ from management’s method
for realizing value from other investments in equity securities.
4
Similarly, IFRS 9 (para. BC2.1) states, “The scope of IAS 39 has not been raised as a matter
of concern during the financial crisis and, hence, the Board believes that the scope of IFRS 9
should be based on that of IAS 39 until it considers the scope more generally in a later phase
of the project to replace IAS 39.” The IASB is in the preliminary stages of a research project
to fundamentally assess the usefulness of the equity method to investors and the difficulties that
preparers face in implementing the equity method (http://www.ifrs.org/Current-Projects/IASBProjects/equity-method-accounting/Pages/equity-method-accounting.aspx).
26
In its May 2010 Exposure Draft, Accounting for Financial Instruments and Revisions to the Accounting for Derivative Instruments and Hedging Activities, the FASB
proposed restricting the application of the equity method to cases where the investor
has significant influence over the investee and the operations of the investee are
considered related to the investor’s consolidated operations (the “related operations
criterion”). If these two criteria are not satisfied, the investment would be accounted
for at fair value with all changes in fair value recognized in net income. The Basis
for Conclusions (para. BC24) explains that the FASB “believes that an entity generally should measure investments in equity securities at fair value with all changes
recognized in net income because the only way to realize gains or losses from equity
securities is to sell the equity securities. ... However, the Board decided that for those
equity investments in which the entity has significant influence over the investee and
the investee’s operations are related to those of the entity’s consolidated operations ...
the equity method of accounting would provide the most appropriate representation
of the underlying economic activity in the entity’s financial statements.”
In its redeliberations, the FASB tentatively decided not to retain the “related
operations criterion” because of concerns about interpretive and operational issues.
Instead, the FASB’s 2013 Exposure Draft would require investments in affiliates to
be accounted for under the equity method unless the investment is held for sale (i.e.,
the plan is to realize value by selling the investment). In that case, the investment
should be measured at fair value with changes in fair value included in net income.
The classification as a held-for-sale investment would be made upon the investment’s
initial qualification for the equity method and could not be changed. This proposal
reflects the notion that management’s intended method of value realization from the
investment (i.e., sale or strategic operation) should determine whether the item is
measured at fair value or using some other approach.
I investigate this issue in the context of equity method investments by examining
27
the value relevance of information provided under the equity method relative to the
value relevance of information resulting from measuring investments in affiliates at
fair value.5 To do this, I treat investments in affiliates as if they were trading securities
and measure them at fair value in each reporting period and calculate income based
on changes in the fair values of the investments.
Previous research generally supports the value relevance of balance sheet measures
of fair values for investment securities, often in banks and often not in a context where
the method of value realization is considered. For example, Barth (1994) finds that
disclosed fair values of investment securities are incrementally associated with bank
share prices for a sample of U.S. banks from 1971-1990. Three studies (Barth et
al. (1996), Eccher et al. (1996), and Nelson (1996)) examine the value relevance of
fair value disclosures made under SFAS No. 107, Disclosures about Fair Values of
Financial Instruments, by a sample of U.S. banks from 1992-1993. All three studies
find that the fair values of investment securities are incrementally informative relative
to book values in explaining bank share prices in at least some specifications. A recent
study by Song et al. (2010) confirms the value relevance of fair value measurements
for investment securities disclosed under SFAS No. 157, Fair Value Measurements,
using quarterly reports of U.S. banking firms in 2008.
Evidence on the value relevance of unrealized gains and losses on investment securities is less definitive (see Rees and Shane (2012) for a summary). For example,
Dhaliwal et al. (1999) find evidence of the value relevance of unrealized gains and
losses on marketable securities only for the financial services industry. Chambers et
al. (2007) find that unrealized gains and losses on available-for-sale (AFS) securities
5
I use an interpretation of value relevance from Francis and Schipper (1999, pages 326-327): “value
relevance is measured by the ability of financial statement information to capture or summarize
information, regardless of source, that affects share values.” My research design does not let me
determine whether investors use the reported information provided under the equity method or
disclosed about the fair value of affiliates. I cannot rule out that the accounting information I
examine is merely correlated with information used by investors.
28
are positively associated with investor returns for a sample of S&P 500 firms from
1998-2003. However, the magnitude of the valuation coefficient on unrealized gains
and losses on AFS securities in their study is significantly larger than that for other
components of other comprehensive income and net income and the authors suggest
that further study is needed to explain the result.
It is not certain that the value relevance results for AFS securities will apply to
investments in affiliates because of differences in the prominence of the presentation
of information about the fair values of the investments. Hirst and Hopkins (1998)
and Maines and McDaniel (2000) find evidence that the format presentation (e.g.,
performance statement, statement of shareholders’ equity, note disclosure) of other
comprehensive income can affect how investors use that information. In the case of
investments in publicly-traded affiliates, the fair value of the investment is disclosed
in the notes (vs. presented on the face of the balance sheet for AFS securities) and
the change in the fair value is not disclosed (vs. recognized in other comprehensive
income for AFS securities). Therefore, it is possible that the value relevance results
for AFS securities would not hold for investments in publicly-traded affiliates.
Related to my study, Graham et al. (2003b) examine a sample of 55 U.S. investors
with affiliates from 1993-1997 and find that both the carrying values reported under
the equity method and the fair values of investments in publicly-traded affiliates have
a positive association with stock prices and stock returns of investor firms. The
authors also find (a) no association between income from investments in affiliates
reported under the equity method and investors’ stock prices and returns and (b) a
positive association between investors’ stock returns and the annual change in fair
value of an equity method investment less the annual change in the income from the
equity investment. Given the FASB’s recent discussions about expanding the use of
fair value measurement for investments in affiliates, I revisit these findings for a larger
sample and a longer and more recent time period. I also examine the value relevance
29
of income from investments in affiliates using fair value measurement by constructing
a “levels” measure of fair value income from affiliates by multiplying the number of
shares held at a reporting date by the change in the affiliate’s share price during the
period. Based on the balance sheet results in Graham et al. (2003b), I expect fair
value income from investments in affiliates to be incrementally value relevant relative
to information provided under the equity method.
The main focus of my value relevance analysis is on evaluating whether management’s plan for realizing value from an investment in an affiliate affects the relevance
of fair value measurement. In particular, I separately analyze the incremental value
relevance of fair value measures for investments in publicly-traded affiliates based on
the distinguishing criteria proposed by the FASB in 2010 and 2013. The FASB’s
2010 Exposure Draft (para. BC24) suggests that the FASB “believes that an entity generally should measure investments in equity securities at fair value with all
changes recognized in net income because the only way to realize gains or losses from
equity securities is to sell the equity securities ... However, the Board decided that
for those equity investments in which the entity has significant influence over the
investee and the investee’s operations are related to those of the entity’s consolidated
operations...the equity method of accounting would provide the most appropriate representation of the underlying economic activity in the entity’s financial statements.”
This suggests that the FASB believes that there is something distinct about investments in affiliates related to the investor’s operations (“strategic” investments in
affiliates) that makes fair value a less relevant measurement attribute than for other
investments in equity securities. I operationalize this proposal by identifying investments in publicly-traded affiliates as “strategic” when the investor and affiliate have
the same 2-digit SIC code and propose the following hypothesis (in alternative form):
H1A: Fair value measures of investments in publicly-traded affiliates identified
30
as “strategic” have less incremental value relevance relative to information
provided under the equity method of accounting than fair value measures of
other investments in publicly-traded affiliates.
The FASB’s 2013 Exposure Draft revised the proposals for investments in affiliates
and would require the equity method unless the investment is classified as “held
for sale” upon the investment’s initial qualification for the equity method. In that
case, the investment would be measured at fair value with changes in fair value
included in net income. This suggests that the FASB believes that fair value is a
more relevant measurement attribute for investments in affiliates classified as held
for sale than for other investments in affiliates. I operationalize this proposal by
identifying investments in publicly-traded affiliates as “held for sale” when they are
sold either in part or in entirety after application of the equity method and propose
the following hypothesis (in alternative form):
H1B: Fair value measures of investments in publicly-traded affiliates identified as “held for sale” have more incremental value relevance relative to
information provided under the equity method of accounting than fair value
measures of other investments in publicly-traded affiliates.
In contrast, fair value might be the appropriate measurement attribute for all investments in affiliates regardless of how management intends to realize value from the
investment. For example, Leisenring et al. (2012) suggest that basing measurement
on management intent impairs comparability by permitting alternative accounting
models for the same economic phenomenon (unless management’s intended method
of value realization is considered to alter the economic characteristics of the underlying equity securities held). The authors suggest (page 339) that “both relevance
and comparability are achievable by basing recognition on the rights and obligations
in an item or arrangement and not on management’s plans for those rights and obli31
gations, and by applying the same measurement attribute to rights and obligations
that are similar, regardless of management’s intentions. Differences in presentation,
not differences in measurement, can be used to convey management’s intent for the
use, disposition or settlement of items.”
5.2 Research Design
The design of my value relevance tests is based on the residual income valuation
model (Edwards and Bell (1961); Peasnell (1982)), as further developed by Ohlson
(1995) and used in the value relevance literature (e.g., Song et al. (2010), Graham
et al. (2003b)).6 I examine the value relevance of information provided under the
equity method relative to the value relevance of information resulting from measuring
investments in affiliates at fair value as follows:
M V Ej ,t
β0
β1 BVotherj ,t
β4 IN Cotherj ,t
β2 BVpublicaffiliatej ,t
BVpublicaffiliatej ,t
β6 F V IN C IN Cpublicaffiliatej ,t
β3 F V
β5 IN Cpublicaffiliatej ,t
εj ,t
(5.1)
where (all variables are on a per-share basis):
MVEj ,t : market value of equity (CSHO * PRCC F) of firm j three months
after the end of fiscal year t;
BVotherj ,t : book value (CEQ) of firm j less the book value of its investments
in publicly-traded affiliates at the end of year t;
BVpublicaffiliatej ,t : book value of firm j’s investments in publicly-traded affiliates
at the end of year t;
6
See Chapter 5.4.2 for additional details on the theoretical values predicted under such a model.
32
FV-BVpublicaffiliatej ,t : fair value less book value of investments in publiclytraded affiliates of firm j at the end of year t;
INCotherj ,t : income before extraordinary items (IB) less reported income from
publicly-traded affiliates of firm j in year t;
INCpublicaffiliatej ,t : reported income from publicly-traded affiliates of firm j in
year t; and
FVINC-INCpublicaffiliatej ,t : fair value income less reported income of publiclytraded affiliates of firm j in year t, where the fair value income of an affiliate
is calculated as the number of shares held in the affiliate at the end of year t
multiplied by the change in the price per share of the affiliate from the end
of year t - 1 to the end of year t.
I scale all variables by number of shares outstanding as Barth and Clinch (2009) show
that this deflator generally performs better than other deflators in mitigating scale
effects in this type of regression.
I am interested in the value relevance of information provided by the equity
method, captured by β 2 and β 5 . A positive β 2 (β 5 ) indicates that the balance sheet
(income statement) information about investments in affiliates is associated with investors’ share prices. I am also interested in whether fair value measurement of investments in affiliates is incrementally value relevant relative to the information provided
under the equity method, captured by β 3 and β 6 . A positive β 3 (β 6 ) indicates that the
balance sheet (income statement) measurement of affiliates at fair value has an incremental association with investors’ share prices relative to the carrying value (income)
derived under the equity method.
In addition, I examine whether the incremental value relevance of fair value measures of investments in affiliates depends on the investor’s intended method of realizing
the value from an investment. In particular, I examine the incremental value rele33
vance of fair value measures separately for strategic and held-for-sale investments in
affiliates as follows:
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β3 F V
BVstrategicpublicaffiliatej ,t
BVnonstrategicpublicaffiliatej ,t β5 IN Cotherj ,t
β6 IN Cpublicaffiliatej ,t β7 F V IN C IN Cstrategicpublicaffiliatej ,t
β8 F V IN C IN Cnonstrategicpublicaffiliatej ,t εj ,t
β4 F V
(5.2)
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β3 F V
BVheldforsalepublicaffiliatej ,t
BVnonheldforsalepublicaffiliatej ,t β5 IN Cotherj ,t
β6 IN Cpublicaffiliatej ,t β7 F V IN C IN Cheldforsalepublicaffiliatej ,t
β8 F V IN C IN Cnonheldforsalepublicaffiliatej ,t εj ,t
β4 F V
(5.3)
where (other variables defined above):
FV-BVstrategicpublicaffiliatej ,t : fair value less book value of investments in publiclytraded affiliates of firm j at the end of year t classified as strategic;
FV-BVnonstrategicpublicaffiliatej ,t : fair value less book value of investments in
publicly-traded affiliates of firm j at the end of year t not classified as strategic;
FV-BVheldforsalepublicaffiliatej ,t : fair value less book value of investments in
publicly-traded affiliates of firm j at the end of year t classified as held-forsale;
FV-BVnonheldforsalepublicaffiliatej ,t : fair value less book value of investments
in publicly-traded affiliates of firm j at the end of year t not classified as
held-for-sale;
34
FVINC-INCstrategicpublicaffiliatej ,t : fair value income less reported income of
publicly-traded affiliates of firm j in year t classified as strategic;
FVINC-INCnonstrategicpublicaffiliatej ,t : fair value income less reported income of
publicly-traded affiliates of firm j in year t not classified as strategic;
FVINC-INCheldforsalepublicaffiliatej ,t : fair value income less reported income of
publicly-traded affiliates of firm j in year t classified as held-for-sale;
FVINC-INCnonheldforsalepublicaffiliatej ,t : fair value income less reported income of publicly-traded affiliates of firm j in year t not classified as held-forsale;
As noted in Chapter 5.1, I identify an affiliate as “strategic” when the investor
and affiliate have the same 2-digit SIC code. I identify an affiliate as “held for sale”
when it is sold either in part or in entirety after application of the equity method.
In this analysis, I am interested in the incremental value relevance of fair value
measures of investments in affiliates, both on the balance sheet (β 3 -β 4 ) and on the
income statement (β 7 -β 8 ). A positive coefficient indicates that the balance sheet (income statement) measurement of affiliates at fair value has an incremental association
with investors’ share prices relative to the carrying value (income) derived under the
equity method. I am also interested in comparing the relative magnitudes of β 3 -β 4
and β 7 -β 8 , respectively.
5.3 Empirical Results
I test the value relevance of information provided under the equity method and the
incremental value relevance of fair value measures of investments in publicly-traded
affiliates by estimating Equation 5.1 using pooled OLS regression for all firm-years
from 1993-2011 for the 221 equity method investors in my sample with the required
data available. I winsorize all variables at the 1st and 99th percentiles and include
35
firm and year fixed effects in the regression. The related descriptive statistics and
correlation coefficients are presented in Tables 5.1 and 5.2. The estimation results
are reported in Table 5.3.
Table 5.1: Descriptive Statistics for the Analysis of the Relation between Investors’
Stock Prices and Balance Sheet and Income Measures of Investments in Affiliates
MVEj ,t
BVotherj ,t
BVpublicaffiliatej ,t
FV-BVpublicaffiliatej ,t
FV-BVstrategicpublicaffiliatej ,t
FV-BVnonstrategicpublicaffiliatej ,t
FV-BVheldforsalepublicaffiliatej ,t
FV-BVnonheldforsalepublicaffiliatej ,t
INCotherj ,t
INCpublicaffiliatej ,t
FVINC-INCpublicaffiliatej ,t
FVINC-INCstrategicpublicaffiliatej ,t
FVINC-INCnonstrategicpublicaffiliatej ,t
FVINC-INCheldforsalepublicaffiliatej ,t
FVINC-INCnonheldforsalepublicaffiliatej ,t
Mean
Std Dev
p25
Median
p75
30.038
12.353
1.659
1.574
0.309
1.227
0.610
0.968
1.018
0.088
-0.159
-0.023
-0.063
-0.004
-0.088
26.683
14.498
3.289
3.864
1.497
3.321
2.116
3.041
3.059
0.463
2.601
0.875
1.779
1.213
1.550
9.610
3.748
0.033
0.015
0.000
0.000
0.000
0.000
-0.093
-0.018
-0.378
0.000
-0.107
0.000
-0.095
23.875
9.190
0.597
0.295
0.000
0.032
0.000
0.000
0.955
0.017
-0.002
0.000
0.000
0.000
0.000
42.750
17.362
1.754
1.515
0.119
0.907
0.200
0.651
2.149
0.140
0.300
0.000
0.074
0.000
0.026
Firm-years
857
The table reports descriptive statistics for the variables used in the estimation of Equations 5.1, 5.2, and 5.3 for
the 221 equity method investors in my sample with the required data available from 1993-2011. The variables are
defined as follows. All variables are on a per-share basis and winsorized at the 1st and 99th percentiles. MVEj ,t :
market value of equity (CSHO * PRCC F) of firm j three months after the end of fiscal year t. BVotherj ,t :
book value (CEQ) of firm j less the book value of its investments in publicly-traded affiliates at the end of year
t. BVpublicaffiliatej ,t : book value of firm j’s investments in publicly-traded affiliates at the end of year t. FVBVpublicaffiliatej ,t : fair value less book value of investments in publicly-traded affiliates of firm j at the end of year
t. FV-BVstrategicpublicaffiliatej ,t : fair value less book value of investments in publicly-traded affiliates of firm j at the
end of year t classified as strategic. FV-BVnonstrategicpublicaffiliatej ,t : fair value less book value of investments in
publicly-traded affiliates of firm j at the end of year t not classified as strategic. FV-BVheldforsalepublicaffiliatej ,t :
fair value less book value of investments in publicly-traded affiliates of firm j at the end of year t classified as held for
sale. FV-BVnonheldforsalepublicaffiliatej ,t : fair value less book value of investments in publicly-traded affiliates
of firm j at the end of year t not classified as held for sale. INCotherj ,t : income before extraordinary items (IB)
less reported income from publicly-traded affiliates of firm j in year t. INCpublicaffiliatej ,t : reported income from
publicly-traded affiliates of firm j in year t. FVINC-INCpublicaffiliatej ,t : fair value income less reported income of
publicly-traded affiliates of firm j in year t, where the fair value income of an affiliate is calculated as the number of
shares held in the affiliate at the end of year t multiplied by the change in the price per share of the affiliate from the
end of year t - 1 to the end of year t. FVINC-INCstrategicpublicaffiliatej ,t : fair value income less reported income
of publicly-traded affiliates of firm j in year t classified as strategic. FVINC-INCnonstrategicpublicaffiliatej ,t : fair
value income less reported income of publicly-traded affiliates of firm j in year t not classified as strategic. FVINCINCheldforsalepublicaffiliatej ,t : fair value income less reported income of publicly-traded affiliates of firm j in year
t classified as held for sale. FVINC-INCnonheldforsalepublicaffiliatej ,t : fair value income less reported income of
publicly-traded affiliates of firm j in year t not classified as held for sale.
36
37
0.616
0.061
0.180
0.189
0.165
0.090
0.218
0.665
0.242
0.176
0.141
0.116
0.085
0.137
(1)
(2)
0.027
0.094
0.098
0.059
0.129
0.042
0.538
0.173
0.061
0.051
0.024
0.033
0.044
0.638
(3)
-0.193
0.031
-0.139
-0.035
0.113
0.465
0.045
0.019
0.025
0.012
0.021
0.214
0.275
(4)
0.273
0.211
0.311
0.463
0.429
0.732
0.461
0.662
0.279
0.297
0.228
0.163
0.157
0.161
0.129
(5)
-0.120
0.296
0.380
0.134
0.051
0.147
0.212
0.051
0.089
0.119
0.336
0.277
0.204
0.860
(6)
0.373
0.542
0.140
0.188
0.149
0.037
0.191
0.120
0.086
0.129
0.085
0.244
0.611
-0.025
(7)
-0.179
0.069
0.056
0.168
0.097
0.142
0.251
0.038
0.208
0.257
0.305
0.769
0.262
0.564
(8)
0.179
0.172
0.115
0.103
0.082
-0.015
0.146
0.209
0.108
0.177
0.276
0.356
0.643
0.008
(9)
0.212
0.129
0.063
0.092
0.033
0.103
0.560
0.505
0.211
0.291
0.243
0.170
0.242
0.180
(10)
0.006
0.052
-0.024
-0.002
-0.011
0.276
0.276
0.589
0.212
0.256
0.207
0.281
0.152
0.179
(11)
0.547
0.801
0.605
0.742
0.046
-0.055
0.013
0.141
0.146
0.164
0.166
0.086
0.030
-0.012
(12)
0.078
0.416
0.479
0.071
0.005
-0.017
0.265
0.254
0.027
0.049
0.138
-0.001
0.010
0.486
(13)
0.519
0.633
0.077
-0.010
0.086
0.210
0.023
0.295
0.245
0.098
0.102
0.032
0.818
0.024
(14)
0.060
0.042
0.007
0.085
0.183
0.088
0.216
0.338
-0.038
0.079
0.101
0.608
0.302
0.591
(15)
0.152
0.117
0.015
0.183
0.019
-0.047
0.769
0.440
0.652
0.069
0.071
-0.035
0.014
The table reports correlation coefficients for the variables used in the estimation of Equations 5.1, 5.2, and 5.3 for the 221 equity method investors in my sample with the required data
available from 1993-2011. Pearson (Spearman) correlation coefficients are presented in the upper-right (lower-left) corner. Bold text indicates two-tailed p-values at the 0.10 level or better.
The variables are defined in Table 5.1. All variables are on a per-share basis and winsorized at the 1st and 99th percentiles.
MVEj ,t (1)
BVotherj ,t (2)
BVpublicaffiliatej ,t (3)
FV-BVpublicaffiliatej ,t (4)
FV-BVstrategicpublicaffiliatej ,t (5)
FV-BVnonstrategicpublicaffiliatej ,t (6)
FV-BVheldforsalepublicaffiliatej ,t (7)
FV-BVnonheldforsalepublicaffiliatej ,t (8)
INCotherj ,t (9)
INCpublicaffiliatej ,t (10)
FVINC-INCpublicaffiliatej ,t (11)
FVINC-INCstrategicpublicaffiliatej ,t (12)
FVINC-INCnonstrategicpublicaffiliatej ,t (13)
FVINC-INCheldforsalepublicaffiliatej ,t (14)
FVINC-INCnonheldforsalepublicaffiliatej ,t (15)
Table 5.2: Correlation Coefficients for the Analysis of the Relation between Investors’ Stock Prices and Balance Sheet and
Income Measures of Investments in Affiliates
The correlation coefficients presented in Table 5.2 indicate a positive univariate
correlation between investors’ market value of equity and both the carrying value of
investments in publicly-traded affiliates reported under the equity method [Spearman
(Pearson) coefficient = 0.061 (0.214)] and the difference between fair value and book
value of investments in publicly-traded affiliates [Spearman (Pearson) coefficient =
0.180 (0.273)]. This provides preliminary confirmation of the results from Graham et
al. (2003b). There is also a positive univariate correlation between investors’ market
value of equity and both income from publicly-traded affiliates reported under the
equity method [Spearman (Pearson) coefficient = 0.242 (0.276)] and the difference
between fair value and reported income from publicly-traded affiliates [Spearman
(Pearson) coefficient = 0.176 (0.046)]. This provides preliminary evidence supporting
the incremental value relevance of fair value measures of income from affiliates relative
to the equity method. There is also a positive correlation between reported income
from affiliates and investors’ other income [Spearman (Pearson) coefficient = 0.212
(0.179)], which might be due to similarities in the industries of some investors and
affiliates.
Regression (1) in Table 5.3 presents results of the multivariate regression analysis. Turning first to the balance sheet measures for investments in affiliates, the
coefficient on the carrying value of investments in publicly-traded affiliates under the
equity method, BVpublicaffiliatej ,t , is positive (β 2 = 1.113, t-statistic = 3.673).7 The
coefficient on FV-BVpublicaffiliatej ,t is also positive (β 3 = 0.712, t-statistic = 3.726).
The coefficient on income from publicly-traded affiliates reported under the equity
method, INCpublicaffiliatej ,t , is not distinguishable from zero (β 5 = 2.132, t-statistic =
1.591). These results are consistent with the findings in Graham et al. (2003b), for a
7
Unless otherwise indicated, the empirical results discussed in Chapters 5.3 and 6.3 have two-tailed
p-values of 0.10 or better.
38
smaller sample (55 firms) and shorter time period (1993-1997).8 I also find evidence
supporting the incremental value relevance of fair value measures of income from
publicly-traded affiliates. In particular, the coefficient on FVINC-INCpublicaffiliatej ,t is
positive (β 6 = 0.396, t-statistic = 2.186).
I then extend this analysis to examine whether the incremental value relevance
of fair value measures of investments in affiliates depends on the investor’s intended
method of realizing the value from an investment. I estimate Equation 5.2 identifying
investments in affiliates as either strategic or non-strategic. Regression (2) in Table
5.3 presents the results. The coefficients on FV-BVstrategicpublicaffiliatej ,t (β 3 = 1.293, tstatistic = 2.530) and FVINC-INCstrategicpublicaffiliatej ,t (β 7 = 1.420, t-statistic = 2.482)
are positive. In addition, tests of the equality of these coefficients with the related
coefficients for non-strategic affiliates reveal that they are not statistically different
from one another (p-value of 0.1626 for test of β 3 = β 4 and p-value of 0.1021 for
test of β 7 = β 8 ). I then estimate Equation 5.3 identifying investments in affiliates
as either held-for-sale or non-held-for-sale. Regression (3) in Table 5.3 presents the
results. The coefficient on FV-BVheldforsalepublicaffiliatej ,t (β 3 = 0.794, t-statistic =
2.443) is positive, but the coefficient on FVINC-INCheldforsalepublicaffiliatej ,t (β 7 =
0.472, t-statistic = 1.178) is indistinguishable from zero. Tests of the equality of
these coefficients with the related coefficients for non-held-for-sale affiliates reveal
that they are not statistically different from one another (p-value of 0.9986 for test
of β 3 = β 4 and p-value of 0.7761 for test of β 7 = β 8 ).
8
The finding of an insignificant coefficient on income reported under the equity method might
be due to a lack of power. In untabulated results, I estimate Equation 5.1 for the full Compustat
population of equity method investors (non-zero IVAEQ or ESUB) from 1993-2011, excluding the
FV-BVpublicaffiliatej ,t and FVINC-INCpublicaffiliatej ,t terms. I find that the coefficient on INCaffiliatej ,t
(β 5 = 0.648, t-statistic = 8.500) is positive.
39
40
5.560
14.408
Included
Included
857
0.828
4.285
1.591
2.186
10.092
3.673
3.726
0.811
2.132
0.396
0.774
1.113
0.712
t-value
.1626
.9986
.1021
.7761
p-value
857
0.829
0.472
0.615
14.600
0.794
0.793
0.794
2.102
0.760
1.088
Coefficient
Included
Included
1.178
2.060
5.641
2.443
3.205
4.190
1.561
9.836
3.615
t-value
(3)
Dependent variable = MVEj ,t
The table reports the results of estimation of Equations 5.1, 5.2, and 5.3 using pooled OLS regression with annual data from 1993-2011 for the 221 equity method investors
in my sample with the required data available. Two-tailed p-values are reported as follows: p
0.10, p
0.05, p
0.01. The variables are defined in Table 5.1.
All variables are on a per-share basis and winsorized at the 1st and 99th percentiles.
Included
Included
857
0.830
5.708
2.482
1.577
1.420
0.413
14.749
4.121
1.432
2.530
2.636
1.293
0.561
0.789
1.922
10.007
3.921
t-value
0.766
1.190
Coefficient
(2)
Dependent variable = MVEj ,t
FV-BVstrategicpublicaffiliatej ,t = FV-BVnonstrategicpublicaffiliatej ,t
FV-BVheldforsalepublicaffiliatej ,t = FV-BVnonheldforsalepublicaffiliatej ,t
FVINC-INCstrategicpublicaffiliatej ,t = FVINC-INCnonstrategicpublicaffiliatej ,t
FVINC-INCheldforsalepublicaffiliatej ,t = FVINC-INCnonheldforsalepublicaffiliatej ,t
Tests of Equality of Coefficients:
Year fixed effects
Firm fixed effects
Firm-years
Adjusted R2
BVotherj ,t
BVpublicaffiliatej ,t
FV-BVpublicaffiliatej ,t
FV-BVstrategicpublicaffiliatej ,t
FV-BVnonstrategicpublicaffiliatej ,t
FV-BVheldforsalepublicaffiliatej ,t
FV-BVnonheldforsalepublicaffiliatej ,t
INCotherj ,t
INCpublicaffiliatej ,t
FVINC-INCpublicaffiliatej ,t
FVINC-INCstrategicpublicaffiliatej ,t
FVINC-INCnonstrategicpublicaffiliatej ,t
FVINC-INCheldforsalepublicaffiliatej ,t
FVINC-INCnonheldforsalepublicaffiliatej ,t
Intercept
Coefficient
(1)
Dependent variable = MVEj ,t
β6 F V IN C
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β3 F V BVpublicaffiliatej ,t
β4 IN Cotherj ,t
β5 IN Cpublicaffiliatej ,t
IN Cpublicaffiliatej ,t εj ,t
M V Ej ,t β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β3 F V BVstrategicpublicaffiliatej ,t
β4 F V BVnonstrategicpublicaffiliatej ,t
β5 IN Cotherj ,t
β6 IN Cpublicaffiliatej ,t
β7 F V IN C IN Cstrategicpublicaffiliatej ,t
β8 F V IN C IN Cnonstrategicpublicaffiliatej ,t
εj ,t
M V Ej ,t β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β3 F V BVheldforsalepublicaffiliatej ,t
β4 F V BVnonheldforsalepublicaffiliatej ,t
β5 IN Cotherj ,t
β6 IN Cpublicaffiliatej ,t
β7 F V IN C IN Cheldforsalepublicaffiliatej ,t
β8 F V IN C IN Cnonheldforsalepublicaffiliatej ,t
εj ,t
M V Ej ,t
β0
(5.3)
(5.2)
(5.1)
Table 5.3: Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of Investments in Affiliates
These results suggest that the incremental value relevance of fair value measures
for investments in affiliates exists for both investments identified as held for sale
and those identified as strategic. The results do not provide evidence in support
of H1A (fair value measures of investments in publicly-traded affiliates identified as
“strategic” have less incremental value relevance relative to information provided
under the equity method of accounting than fair value measures of other investments
in publicly-traded affiliates) and H1B (fair value measures of investments in publiclytraded affiliates identified as “held for sale” have more incremental value relevance
relative to information provided under the equity method of accounting than fair
value measures of other investments in publicly-traded affiliates).
In summary, I find that balance sheet measures of investments in affiliates provided
under the equity method are associated with investors’ stock prices and that fair
value balance sheet and income measures of investments in affiliates are incrementally
associated with investors’ stock prices after controlling for information provided under
the equity method. Income from affiliates recognized under the equity method is not
associated with investors’ stock prices. These results support the incremental value
relevance of fair value measurement of investments in affiliates relative to the equity
method. I also find that the incremental value relevance of fair value measures for
investments in affiliates exists for both investments identified as held for sale and those
identified as strategic, with no evidence that the incremental value relevance is higher
(lower) for investments identified as held for sale (strategic). This result suggests that
the FASB’s 2010 and 2013 proposals to distinguish between investments in affiliates
based on management’s intended method of value realization are not supported by
differences in the relative value relevance of fair value measures for these types of
investments.
41
5.4 Supplementary Analyses
5.4.1
Correlation between Investor and Affiliate Returns
I investigate whether the results in Table 5.3 of a positive association between
investors’ stock prices and the fair value measures (both balance sheet and income
statement) of investments in affiliates simply capture the correlation between investor
and affiliate stock returns. This concern arises because of the overlap in industry memberships of investors and affiliates noted in Chapter 4.1. To do this, I first calculate
the correlation of investor and affiliate monthly stock returns for each investor-affiliate
pair in my sample during the period in which the equity method is applied. The average (median) number of monthly observations used to calculate this correlation is
43.59 (32.00):
Variable
Number of monthly observations used to calculate correlations
N
Mean
Std Dev
P25
Median
P75
253
43.59
39.99
17.00
32.00
56.00
The average Pearson (Spearman) correlation of monthly stock returns of investors
and affiliates across firms in my sample is 0.322 (0.329):
Variable
Pearson correlation between monthly stock returns of investors
and publicly-traded affiliates during the period
in which the equity method is applied
Spearman correlation between monthly stock returns of investors
and publicly-traded affiliates during the period
in which the equity method is applied
N
Mean
Std Dev
P25
Median
P75
253
0.322
.0272
0.138
0.327
0.511
253
0.329
0.260
0.154
0.335
0.500
I then include in my value relevance analysis an indicator variable (and related
interaction terms) equal to one when the investor-affiliate correlation is above the
sample median. As shown in Table 5.4, I find that the incremental value relevance
of fair value balance sheet measures exists in the low correlation group (coefficient
on FV-BVpublicaffiliatej ,t = 0.965, t-statistic = 2.931) and that the incremental value
relevance of fair value income statement measures is not higher for the high correlation
42
group than the low correlation group (untabulated test of equality of coefficients on
FVINC-INCpublicaffiliatej ,t and FVINC-INCpublicaffiliatej ,t + FVINC-INCpublicaffiliatej ,t *
HIGH CORRELATIONj ,t has p-value of 0.8910). Together, this suggests that my
results do not seem to be driven by the correlation in investor returns and affiliate
returns.
Table 5.4: Effect of the Correlation of Investor and Affiliate Stock Returns on the
Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of
Investments in Affiliates
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β5 IN Cpublicaffiliatej ,t
β6 F V IN C
β3 F V BVpublicaffiliatej ,t
IN Cpublicaffiliatej ,t εj ,t
Dependent variable = MVEj ,t
Coefficient
BVotherj ,t
BVotherj ,t * HIGH CORRELATIONj ,t
BVpublicaffiliatej ,t
BVpublicaffiliatej ,t * HIGH CORRELATIONj ,t
FV-BVpublicaffiliatej ,t
FV-BVpublicaffiliatej ,t * HIGH CORRELATIONj ,t
INCotherj ,t
INCotherj ,t * HIGH CORRELATIONj ,t
INCpublicaffiliatej ,t
INCpublicaffiliatej ,t * HIGH CORRELATIONj ,t
FVINC-INCpublicaffiliatej ,t
FVINC-INCpublicaffiliatej ,t * HIGH CORRELATIONj ,t
HIGH CORRELATIONj ,t
Intercept
Year fixed effects
Firm fixed effects
Firm-years
Adjusted R2
t-value
0.774
10.092
1.113
3.673
0.712
3.726
0.811
β4 IN Cotherj ,t
(5.1)
Dependent variable = MVEj ,t
Coefficient
0.763
t-value
8.377
4.285
0.004
1.159
0.079
0.965
0.304
0.035
2.444
0.134
2.931
0.793
2.132
1.591
0.396
2.186
14.408
5.560
Included
Included
784
0.833
0.870
5.815
1.061
5.815
0.353
0.051
1.792
15.405
3.605
2.135
0.542
2.135
1.163
0.137
0.670
5.327
Included
Included
784
0.839
The table incorporates into Equation 5.1 the correlation of montly stock returns of equity method investors and affiliates.
HIGH CORRELATIONj ,t : indicator variable equal to 1 if firm j has a correlation of its monthly stock returns with those
of its publicly-traded affiliates that is above the sample median. Additional variables are defined in Table 5.1. Two-tailed
p-values are reported as follows: p
0.10, p
0.05, p
0.01.
5.4.2
Theoretical Values in a Residual Income Model
The design of the value relevance tests is based on the residual income valuation
model (Edwards and Bell (1961); Peasnell (1982)). In particular, I use a modified
version of the valuation framework derived in Equation (7) of Ohlson (1995):
Pt
k pφxtq p1 kqyt
43
(5.4)
where:
Pt : the market value, or price, of the firm’s equity at date t;
xt : earnings for the period t - 1 to t;
yt : (net) book value at date t;
Rf : the risk-free rate plus one;
ω: persistence of abnormal earnings, i.e., xat
1
ωxat
ε1t 1 ;
xat : xt - (Rf - 1) yt1 ;
k:
pRf 1qω ; and
Rf ω
φ:
Rf
.
Rf 1
For my sample period, the average monthly risk-free rate of return (one-month
treasury bill rate) is .00259474, which implies an average annual risk-free rate of
return of approximately 3.11%. I also calculate an average persistence of abnormal
earnings for my sample firms. In particular, I estimate xat
1
ωxat
vt
ε1t
1
for
each sample firm and calculate the average ω across firms. I do this in two ways: (a)
using all fiscal years from 1993-2011 for my sample firms and (b) using only those
fiscal years which appear in my value relevance regression. I calculate an average ω
of 0.517 and 0.806, respectively.9
Therefore, the theoretical value for the balance sheet variables in my value relevance regression should be either 0.969 or 0.889 based on the average persistence
measures calculated and the following formula:
1k
1 pRRf 1ωqω
(5.5)
f
9
In a t-test, I find no statistical difference between these average persistence values (p-value of
0.286).
44
The theoretical value for the income statement variables in my value relevance
regression should be either 1.037 or 3.692 based on the average persistence measures
calculated and the following formula:
kφ pRf 1qω Rf
R f ω Rf 1
(5.6)
As shown in Table 5.5, in general, the valuation coefficients are indistinguishable
from their theoretical values in at least one of the specifications.
Table 5.5: Evaluation of Theoretical Values of Coefficients Relating Investors’ Stock
Prices to Balance Sheet and Income Measures of Investments in Affiliates
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β5 IN Cpublicaffiliatej ,t
β6 F V IN C
β3 F V BVpublicaffiliatej ,t
IN Cpublicaffiliatej ,t εj ,t
Dependent variable
MVEj ,t
Coefficient
BVotherj ,t
BVpublicaffiliatej ,t
BVpublicaffiliatej ,t + FV-BVpublicaffiliatej ,t
INCotherj ,t
INCpublicaffiliatej ,t
INCpublicaffiliatej ,t + FVINC-INCpublicaffiliatej ,t
Year fixed effects
Firm fixed effects
Firm-years
Adjusted R2
0.774
1.113
1.825
0.811
2.132
2.528
β4 IN Cotherj ,t
(5.1)
p-values for tests
of equality with theoretical values
Rf = 1.0311
ω = 0.517
BV = 0.969
INC = 1.037
.0112
.6355
.0043
.2335
.4142
.2749
Rf = 1.0311
ω = 0.806
BV = 0.889
INC = 3.692
.1337
.4605
.0018
.0000
.2445
.3938
Included
Included
857
0.828
The table evaluates the coefficients derived from estimating Equation 5.1 with their theoretical
values as described in Section 5.4.2. Variables
are defined in Section 5.4.2 and Table 5.1. Twotailed p-values are reported as follows: p 0.10, p 0.05, p 0.01.
45
6
Earnings Management by Investors with Significant
Influence
6.1 Related Literature and Hypotheses Development
I also examine the neutrality of income reported under the equity method. Specifically, I investigate whether the equity method achieves its objective of reducing earnings management. During the development of APB 18, concern was expressed that
when investors can exercise significant influence, the cost method might invite earnings management through the timing of dividend payments (e.g., Miller and Bahnson
(2007)).1 Earnings management is inconsistent with neutral financial reporting. The
FASB’s Conceptual Framework (SFAC No. 8, para. QC12) confirms the importance
of neutrality by including it as a component of faithful representation, one of the
fundamental qualitative characteristics of decision useful financial information.2
However, the equity method does not eliminate all opportunities for earnings
1
Under the cost method, income recognition only occurs through dividend payments (or impairments of investments, when required).
2
SFAC No. 8 (para. QC14) states, “A neutral depiction is without bias in the selection or
presentation of financial information. A neutral depiction is not slanted, weighted, emphasized,
deemphasized, or otherwise manipulated to increase the probability that financial information will
be received favorably or unfavorably by users.”
46
management. Because a proportionate share of the affiliates’ income flows into the
investors’ net income, investors might use their “significant influence” to manage the
earnings of affiliates for the investors’ benefit. I explore whether the equity method
achieves its objective of reducing earnings management by examining the effect of
equity method ownership on the signed discretionary accruals of affiliates.3,4
There is an extensive literature examining earnings management (see, for example,
Dechow et al. (2010) for a review). The literature (e.g., Libby and Seybert (2009))
suggests a variety of factors that might drive earnings management, including capital
market pressures, individual reputation concerns, tax savings, and bonus compensation. For example, Graham et al. (2005) survey more than 400 chief financial officers
and find more than 80% of participants believe that meeting earnings benchmarks
builds credibility with the capital market and helps maintain or increase the firm’s
stock price (Table 4). The authors (page 28) also find that more than 75% of participants “agree or strongly agree that a manager’s concern about her external reputation
helps explain the desire to hit the earnings benchmark” with interviews confirming
that “the desire to hit the earnings target appears to be driven less by short-run
compensation motivations than by career concerns.” This paper does not attempt to
isolate the specific motivation(s) driving management of income from affiliates.
The one-line balance sheet and income statement presentation of information
about investments in affiliates might make managing income from this source less
transparent than managing income from other sources. For example, Lee et al. (2013)
find that analysts’ annual earnings forecasts are less accurate and more dispersed for
firms with affiliates relative to firms without affiliates for a sample of 21,336 U.S firm3
In future work, I plan to examine whether equity method investors influence operating decisions
of affiliates (“real actions”) for the investors’ benefit (e.g., to achieve earnings targets).
4
Accounting for investments in affiliates similar to trading securities or available-for-sale securities
would result in recognizing dividends received as income. Therefore, those methods would not eliminate concerns about earnings management, but the effects would appear in the timing of dividends,
rather than in discretionary accruals.
47
year observations from 1985-2010. The authors suggest that this is due to the lack of
disclosure of components of affiliate earnings under the equity method.
Previous literature documents two ex ante actions taken by firms to manage earnings in relation to equity method investments. First, Comiskey and Mulford (1986,
1988) find evidence that the requirements of the equity method affect the level of
ownership taken by investor firms. In particular, they find that for a sample of 1,255
investment positions taken in U.S firms in 1982, a statistically significant concentration of positions were taken just under and just over 20 percent, and investees that
were owned in the 19-19.99 percent range (applying the cost method) reported net
losses significantly more often than investees that were owned in the 20-20.99 percent
range (applying the equity method). Second, Morris and Gordon (2006) examine the
reporting choices of Australian firms holding investments in affiliates before and after
the first accounting standard on the subject was implemented in 1984. They find that
firms choose to show the effects of the equity method in a third column on the face of
the financial statements if doing so increases reported earnings, but in a footnote if
the equity method decreases reported earnings. I extend this literature by examining
whether investors manage earnings from affiliates ex post. That is, once firms make
investments with “significant influence” and apply the equity method of accounting,
do those investors manage reported earnings from affiliates?
I propose the following hypothesis (in alternative form):
H2A: Signed discretionary accruals of affiliates are higher when equity method
investors have an incentive to manage income from affiliates.
Equity method investors might also affect the discretionary accruals of affiliates
for reasons apart from a desire to achieve their own earnings targets. For example, the
literature examining the effects of large shareholders (mainly institutional investors)
on the financial reporting of investees suggests monitoring or the consumption of
48
private benefits by large shareholders as two possible examples of factors affecting
financial reporting. The monitoring effect exists when large shareholders can mitigate
the effects of self-interested actions taken by an affiliate’s management (e.g., reporting
higher discretionary accruals during their tenure and deferring the recognition of
losses until after their tenure) by monitoring the actions of affiliate’s management.
Previous research (e.g., Chung et al. (2002), Yeo et al. (2002), Velury and Jenkins
(2006), Koh (2007), Hadani et al. (2011)) documents a negative relation between
earnings management and the presence of large shareholders. The private benefits
effect exists when insiders use their positions of influence to benefit themselves at
the expense of other shareholders (e.g., Shleifer and Vishny (1997) and La Porta et
al. (1999)) and then manage reported amounts so as to deter or impede scrutiny.
This includes engaging in self-serving transactions, transferring assets to the insider
or another firm owned by the insider, or consuming perquisites. Previous research
documents the effects of concealing private benefits consumption through a positive
relation between earnings management and the presence of “insiders” (e.g., Leuz et
al. (2003), Haw et al. (2004), Gopalan and Jayaraman (2012)).
The list of factors above is not exhaustive. In addition, it is possible that there will
be no association between an equity method investor and the financial reporting of
an affiliate. Equity method investors might not use or wish to influence the external
financial reporting of affiliates. By definition, equity method investors have “significant influence” over the operating and financial policies of affiliates. This implies
that equity method investors have access to information necessary to affect operating
and financial decisions of affiliates and might not need to rely on external financial
reporting to monitor affiliates’ management. In addition, as discussed in Chapter 5.1,
equity method investments are undertaken for a variety of reasons (including strategic
operational benefits) unrelated to influencing the financial reporting of affiliates.
The focus of my analysis of the relation between equity method ownership and
49
signed discretionary accruals of affiliates is to identify the existence of earnings management (if any) arising under the equity method of accounting. I do not attempt to
isolate the factors affecting the relation between equity method ownership and signed
discretionary accruals of affiliates in periods when investors have weaker incentives to
manage income from affiliates and therefore make no prediction about which factors
will dominate:
H2B: There is no association between signed discretionary accruals of affiliates and equity method ownership when equity method investors have
no/weaker incentives to manage income from affiliates.
6.2 Research Design
I follow prior research examining the relation between earnings management and
ownership characteristics (e.g., Gopalan and Jayaraman (2012)) and analyze the relation between equity method ownership and signed discretionary accruals of affiliates
using the following model:
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
β2 AF F ILIAT Ek,t IN CEN T IV EIN V EST ORk,t
β3 IN CEN T IV EAF F ILIAT Ek,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
β7 CAP IT ALk ,t
β8 DEBTk ,t
β10 SALESGROW T Hk ,t
β13 SALESV OLk ,t
β16 DN IOW Nk ,t
β6 OP CY CLEk ,t
β9 M T Bk ,t
β11 LOSSk ,t
β14 T RAOW Nk ,t
β17 M ON OW Nk ,t
β12 ASSET Sk ,t
β15 QIXOW Nk ,t
εk ,t
(6.1)
where:
DISC ACCk ,t : discretionary accruals of firm k in quarter t estimated using
the Jones (1991) model (estimation described in detail below);
50
AFFILIATEk,t : indicator variable equal to 1 if firm k is an affiliate (i.e., has
an equity method investor) during quarter t, and 0 otherwise;
INCENTIVEIN V EST ORk,t : indicator variable equal to 1 if income from affiliates allows the equity method investor of firm k to meet one of three earnings
targets (non-negative income, non-negative change in income, analysts’ forecasted earnings) in year t, and 0 otherwise;
CONTROLS:
INCENTIVEAF F ILIAT Ek,t : indicator variable equal to 1 if firm k just meets
one of three earnings targets (non-negative income, non-negative change in
income, analysts’ forecasted earnings) in quarter t, and 0 otherwise;5
INCENTIVE REVERSALIN V EST ORk,t : indicator variable equal to 1 if quarter t is identified as the reversal period for accruals from an INCENTIVEIN V EST OR
period using the Baber et al. (2011) methodology;
INCENTIVE REVERSALAF F ILIAT Ek,t : indicator variable equal to 1 if quarter t is identified as the reversal period for accruals from an INCENTIVEAF F ILIAT E
period using the Baber et al. (2011) methodology;
OPCYCLEk ,t : length of the operating cycle of firm k in quarter t, measured
as days receivable plus days inventory less days payable at the beginning of
the quarter, as defined in Zang (2012);
CAPITALk ,t : capital intensity of firm k in quarter t, measured as the ratio
of noncurrent assets (ATQ - ACTQ) to lagged total assets;
5
In particular, INCENTIVEAF F ILIAT Ek,t = 1 when an affiliate has small positive earnings, a
small positive change in earnings, or just meets analyst expectations. To define “small”, I calculate
the ratio of the absolute value of net income to total assets for my sample. Firms are then defined as
having “small” positive earnings (change in earnings) if their earnings (change in earnings) divided
by total assets in period t are less than the 10th percentile of this ratio for my overall sample of
firms. I define “just meeting” analyst expectations as situations in which actual I/B/E/S earnings
per share less the median analyst forecast (using the most recent forecast for each analyst within 90
days of the earnings announcement) is greater than zero and less than or equal to 0.01.
51
LEVk ,t : leverage ((DLTTQ + DLCQ)/(CSHOQ * PRCCQ) of firm k at the
end of quarter t;
MTBk ,t : market-to-book ratio (CSHOQ*PRCCQ/CEQQ) of firm k in quarter t;
SALESGROWTHk ,t : quarterly sales (SALEQ) growth rate of firm k from
quarter t - 1 to quarter t;
LOSSk ,t : an indicator variable equal to 1 if firm k has a loss in quarter t;
ASSETSk ,t : logarithm of total assets (ATQ) of firm k in quarter t;
SALESVOLk ,t : sales volatility, measured as the standard deviation of quarterly sales (SALEQ) of firm k in quarter t based on the most recent five
quarterly observations;
TRAOWNk ,t : percentage ownership of firm k at the end of quarter t by
institutional investors described as transient by Bushee (2001);
QIXOWNk ,t : percentage ownership of firm k at the end of quarter t by institutional investors described as quasi-indexing by Bushee (2001);
DNIOWNk ,t : percentage ownership of firm k at the end of quarter t by
institutional investors described as dedicated by Bushee (2001) and nonindependent by Brickley et al. (1988); and
MONOWNk ,t : percentage ownership of firm k at the end of quarter t by institutional investors described as monitoring institutions in Ramalingegowda
and Yu (2012) based on classification as dedicated by Bushee (2001) and
independent by Brickley et al. (1988).
The main coefficient of interest is β 2 , which investigates whether earnings management (investors using the flow-through of affiliate income to investor income to
achieve their own earnings targets) exists in my sample. I include an interaction term
52
representing situations when it is likely that equity method investors would have an
incentive to manage the earnings of affiliates. I capture this construct by identifying
situations in which income from an affiliate allows an equity method investor to meet
an earnings target (non-negative income, non-negative change in income, or analysts’
forecasted earnings). That is, I identify situations in which the earnings of an equity method investor without incorporating income from an affiliate are below the
earnings target, but the inclusion of income from the affiliate in the investor’s earnings results in the investor meeting or exceeding its earnings target. A positive β 2
indicates that affiliates have higher signed discretionary accruals when income from
an affiliate allows their equity method investors to meet an earnings target and thus
provides evidence of earnings management.
I also investigate the relation between equity method ownership and discretionary
accruals of affiliates in periods when investors have weaker incentives to manage income from affiliates. A negative (positive) β 1 indicates that firms have lower (higher)
signed discretionary accruals when they are affiliates. As discussed in Chapter 6.1,
this relation might arise due to the effects of monitoring, concealment of private
benefits, or other factors related to equity method ownership.
I measure discretionary accruals using the Jones (1991) model. I estimate Equation 6.2 for each firm in my sample, using all quarterly observations for that firm
from 1993-2011 in periods in which the firm is not an affiliate (with a minimum of 10
observations required).6,7 This approach allows me to control for firm-specific factors
affecting accruals. I control for time-specific factors affecting accruals by including
year fixed effects in my regression.
6
The use of a “non-event” period to develop firm-specific estimates of discretionary accruals is
discussed, for example, by Dechow et al. (1995).
7
In untabulated tests, I find that my main results are robust to excluding the intercept in Equation
6.2, estimating discretionary accruals using the modified Jones model (Dechow et al. (1995)), and
using annual data rather than quarterly data.
53
T ACCk ,t
α0
α1
1
ASSETSk ,t 1
α2 4SALESk ,t
α3 P P Ek ,t
µk ,t
(6.2)
where:
TACCk ,t : total accruals of firm k in quarter t, calculated as net income (NIQ)
less cash flows from operations (OANCF), scaled by lagged total assets;
ASSETSk ,t 1 : total assets (ATQ) of firm k in quarter t - 1;
4SALESk ,t : change in firm k’s sales (SALEQ) from quarter t - 1 to quarter
t, scaled by lagged total assets; and
PPEk ,t : firm k’s gross property, plant and equipment (PPEGTQ) in quarter
t, scaled by lagged total assets.
The coefficient estimates from Equation 6.2 are used to estimate firm-specific
discretionary accruals (DISC ACCk ,t ) for each firm-quarter as follows:
DISC ACCk ,t
T ACCk ,t αx0
1
x ASSETS
α1
k ,t 1
x
α2 4SALESk ,t
x
α3 P P Ek ,t
(6.3)
I include the following firm-specific control variables in Equation 6.1. First, I
include a proxy for the affiliate’s own incentives to manage earnings to meet or
beat earnings targets (INCENTIVEAF F ILIAT Ek,t ). I also include indicator variables
(INCENTIVE REVERSAL IN V EST ORk,t and INCENTIVE REVERSALAF F ILIAT Ek,t )
identifying “accruals reversal periods” associated with any build-up of discretionary
accruals created in periods in which the investor or affiliate had an incentive to manage earnings to meet an earnings target. In particular, I follow the methodology in
Baber et al. (2011) and for each firm, I estimate:
54
DISC ACCk ,t
αk ,t
ρm DISC ACCk,t
m
εk,t
(6.4)
for m = 1 to 16 quarters. Following Proposition 1 in Baber et al. (2011), the minimum
value of ρm is achieved in the period in which period t discretionary accruals reverse.
Prior studies (e.g., Dechow (1994), Dechow and Dichev (2002), Hribar and Nichols
(2007)) find that accruals can be affected by differences in the following factors:
operating cycle (OPCYCLEk ,t ), capital intensity (CAPITALk ,t ), leverage (LEVk ,t ),
size (ASSETSk ,t ), and sales volatility (SALESVOLk ,t ). I also control for differences
in growth opportunities (MTBk ,t and SALESGROWTHk ,t ). Finally, I control for
the effects of various types of institutional ownership (TRAOWNk ,t , QIXOWNk ,t ,
DNIOWNk ,t , MONOWNk ,t ) as monitoring or other actions taken by these investors
might also affect discretionary accruals (e.g., Chung et al. (2002), Yeo et al. (2002)).
6.3 Empirical Results
I investigate the relation between equity method ownership and signed discretionary accruals of affiliates (H2) by estimating Equation 6.1 using a pooled OLS
regression for all firm-quarters from 1993-2011 for the publicly-traded affiliates in my
sample with the required data available. I include year and firm fixed effects in the
regression and winsorize all variables at the 1st and 99th percentiles. The related
descriptive statistics and correlation coefficients are presented in Tables 6.1 and 6.2.
The estimation results are reported in Table 6.3.
As shown in Table 6.1, the mean of INCENTIVEIN V EST ORk,t indicates that in
3.5% of my firm-quarter observations (175/4,937), the firm has an equity method
investor and income from the firm calculated under the equity method allows the equity method investor to meet an earnings target (non-negative income, non-negative
change in income, or analysts’ forecasted earnings).
55
This translates into equity
Table 6.1: Descriptive Statistics for the Discretionary Accruals Analysis
DISC ACCk ,t
AFFILIATEk ,t
INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
OPCYCLEk ,t
CAPITALk ,t
LEVk ,t
MTBk ,t
SALESGROWTHk ,t
LOSSk ,t
ASSETSk ,t
SALESVOLk ,t
TRAOWNk ,t
QIXOWNk ,t
DNIOWNk ,t
MONOWNk ,t
Firm-quarters
Firms
Mean
Std Dev
p25
Median
p75
-0.009
0.260
0.035
0.262
0.031
0.222
57.063
0.587
0.880
2.942
0.080
0.384
6.052
64.898
0.110
0.231
0.012
0.063
0.121
0.438
0.185
0.440
0.174
0.415
168.541
0.272
2.244
6.907
0.436
0.486
2.018
173.401
0.110
0.189
0.053
0.074
-0.019
0.000
0.000
0.000
0.000
0.000
-5.130
0.386
0.052
1.128
-0.060
0.000
4.484
2.561
0.016
0.073
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
39.702
0.594
0.274
1.962
0.025
0.000
6.131
9.282
0.075
0.183
0.000
0.038
0.021
1.000
0.000
1.000
0.000
0.000
95.480
0.797
0.741
3.381
0.124
1.000
7.424
43.067
0.175
0.375
0.001
0.100
4,937
202
The table reports descriptive statistics for the variables used in the analysis of the level of discretionary accruals
of affiliates (Equation 6.1) for the 202 U.S. publicly-traded affiliates in my sample with the required quarterly data
available from 1993-2011. The variables are defined as follows. All variables (except indicator variables) are winsorized at the 1st and 99th percentiles. DISC ACCk ,t : discretionary accruals of firm k in quarter t estimated using
the Jones (1991) model (estimation described in detail below). AFFILIATEk,t : indicator variable equal to 1 if firm
k is an affiliate (i.e., has an equity method investor) during quarter t, and 0 otherwise. INCENTIVEIN V EST ORk,t :
indicator variable equal to 1 if income from affiliates allows the equity method investor of firm k to meet one of
three earnings targets (non-negative income, non-negative change in income, analysts’ forecasted earnings) in year
t, and 0 otherwise. INCENTIVEAF F ILIAT Ek,t : indicator variable equal to 1 if firm k just meets one of three
earnings targets (non-negative income, non-negative change in income, analysts’ forecasted earnings) in quarter t,
and 0 otherwise. INCENTIVE REVERSALIN V EST ORk,t : indicator variable equal to 1 if quarter t is identified
as the reversal period for accruals from an INCENTIVEIN V EST OR period using the Baber et al. (2011) methodology. INCENTIVE REVERSALAF F ILIAT Ek,t : indicator variable equal to 1 if quarter t is identified as the
reversal period for accruals from an INCENTIVEAF F ILIAT E period using the Baber et al. (2011) methodology.
OPCYCLEk ,t : length of the operating cycle of firm k in quarter t, measured as days receivable plus days inventory
less days payable at the beginning of the quarter, as defined in Zang (2012). CAPITALk ,t : capital intensity of
firm k in quarter t, measured as the ratio of noncurrent assets (ATQ - ACTQ) to lagged total assets. LEVk ,t :
leverage ((DLTTQ + DLCQ)/(CSHOQ * PRCCQ) of firm k at the end of quarter t. MTBk ,t : market-to-book
ratio (CSHOQ*PRCCQ/CEQQ) of firm k in quarter t. SALESGROWTHk ,t : quarterly sales (SALEQ) growth
rate of firm k from quarter t - 1 to quarter t. LOSSk ,t : an indicator variable equal to 1 if firm k has a loss in quarter
t. ASSETSk ,t : logarithm of total assets (ATQ) of firm k in quarter t. SALESVOLk ,t : sales volatility, measured
as the standard deviation of quarterly sales (SALEQ) of firm k in quarter t based on the most recent five quarterly
observations. TRAOWNk ,t : percentage ownership of firm k at the end of quarter t by institutional investors
described as transient by Bushee (2001). QIXOWNk ,t : percentage ownership of firm k at the end of quarter t by
institutional investors described as quasi-indexing by Bushee (2001). DNIOWNk ,t : percentage ownership of firm
k at the end of quarter t by institutional investors described as dedicated by Bushee (2001) and non-independent
by Brickley et al. (1988). MONOWNk ,t : percentage ownership of firm k at the end of quarter t by institutional
investors described as monitoring institutions in Ramalingegowda and Yu (2012) based on classification as dedicated
by Bushee (2001) and independent by Brickley et al. (1988).
ESTIMATION OF DISCRETIONARY ACCRUALS: Discretionary accruals are measured as the residuals
from the Jones (1991) model, with parameters estimated at the firm level using all quarters from 1993-2011 in which
the firm is not an affiliate with a minimum of 10 observations required.
TACCk ,t = α0 + α1
1
ASSETSk ,t 1
+ α2 4SALESk ,t + α3 PPEk ,t + µk,t (6.2)
TACCk ,t : total accruals of firm k in quarter t, calculated as net income (NIQ) less cash flows from operations
(OANCF), scaled by lagged total assets. ASSETSk ,t 1 : total assets (ATQ) of firm k in quarter t - 1. 4SALESk ,t :
change in firm k’s sales (SALEQ) from quarter t - 1 to quarter t, scaled by lagged total assets. PPEk ,t : firm k’s
gross property, plant and equipment (PPEGTQ) in quarter t, scaled by lagged total assets.
56
57
-0.029
0.024
0.031
0.001
0.017
-0.015
0.003
-0.050
0.030
0.015
-0.177
0.005
0.007
0.022
0.006
0.009
0.034
(1)
(2)
0.324
0.016
0.166
-0.023
0.068
-0.049
0.051
-0.049
0.014
0.027
-0.027
0.009
-0.139
-0.105
0.005
0.012
-0.083
(3)
0.005
0.148
-0.005
-0.042
0.111
0.044
-0.004
-0.004
-0.093
0.094
0.118
-0.033
-0.024
0.042
0.040
0.027
0.324
(4)
-0.017
0.100
-0.033
0.094
0.027
0.021
0.019
-0.247
0.163
0.103
0.091
0.145
0.102
0.110
0.068
0.016
0.005
(5)
0.019
-0.028
0.080
0.087
-0.009
-0.023
-0.044
0.090
0.117
0.001
0.040
0.062
0.104
0.010
0.166
0.148
-0.017
(6)
-0.033
0.056
0.039
-0.007
-0.019
-0.128
0.173
0.135
0.104
0.169
0.095
0.106
0.043
-0.023
-0.005
0.100
0.019
(7)
-0.397
0.028
-0.077
-0.112
0.042
-0.186
-0.113
-0.106
-0.033
-0.090
-0.072
-0.051
0.021
-0.033
-0.029
-0.025
-0.015
(8)
0.400
-0.038
0.065
-0.174
0.582
0.404
0.212
0.185
0.197
0.206
0.100
-0.032
0.102
0.090
0.073
0.056
-0.262
(9)
-0.394
-0.036
-0.030
0.391
0.378
-0.011
0.086
0.002
0.077
-0.066
-0.015
-0.012
-0.039
-0.002
0.011
-0.005
0.106
(10)
0.089
-0.139
0.008
0.021
0.195
0.095
0.106
0.079
0.032
-0.037
0.020
-0.019
-0.017
-0.010
-0.016
-0.057
-0.087
(11)
-0.111
0.019
0.014
0.043
-0.008
0.016
0.010
0.020
0.014
0.000
-0.036
-0.001
-0.032
-0.052
0.089
-0.009
0.029
(12)
-0.291
-0.297
-0.171
-0.208
-0.175
-0.202
-0.185
0.027
-0.093
-0.247
-0.044
-0.128
0.075
-0.167
0.121
-0.009
-0.045
(13)
0.853
0.449
0.506
0.434
0.417
0.096
-0.037
0.083
0.169
0.075
0.175
-0.129
0.547
0.131
-0.029
0.003
-0.301
(14)
0.027
-0.080
0.011
0.059
-0.001
0.098
-0.012
0.165
0.022
-0.021
0.037
-0.126
0.503
1.000
0.406
0.436
0.372
0.368
(15)
0.668
0.351
0.479
0.072
0.174
-0.058
0.050
-0.011
0.084
-0.063
0.169
-0.046
0.053
0.021
-0.142
0.370
0.052
(16)
0.429
0.572
0.060
-0.155
-0.052
0.136
0.021
0.174
-0.012
0.176
-0.032
-0.010
-0.029
-0.206
0.496
0.241
0.559
(17)
0.372
0.019
0.126
0.043
0.002
0.000
0.009
0.002
-0.028
0.052
-0.039
-0.013
-0.050
0.110
0.014
-0.002
0.037
(18)
0.055
0.014
0.054
0.080
0.119
0.094
-0.044
0.170
-0.017
0.004
0.021
-0.150
0.326
0.072
0.318
0.462
0.058
The table reports Pearson (Spearman) correlation coefficients in the upper-right (lower-left) corner for the variables used in the analysis of the level of discretionary accruals of affiliates (Equation
6.1) for the 202 U.S. publicly-traded affiliates in my sample with the required quarterly data available from 1993-2011. Bold text indicates two-tailed p-values at the 0.10 level or better. The
variables are defined in Table 6.1. All variables (except indicator variables) are winsorized at the 1st and 99th percentiles.
DISC ACCk ,t (1)
AFFILIATEk ,t (2)
INCENTIVEIN V EST ORk,t (3)
INCENTIVEAF F ILIAT Ek,t (4)
INCENTIVE REVIN V EST ORk,t (5)
INCENTIVE REVAF F ILIAT Ek,t (6)
OPCYCLEk ,t (7)
CAPITALk ,t (8)
LEVk ,t (9)
MTBk ,t (10)
SALESGROWTHk ,t (11)
LOSSk ,t (12)
ASSETSk ,t (13)
SALESVOLk ,t (14)
TRAOWNk ,t (15)
QIXOWNk ,t (16)
DNIOWNk ,t (17)
MONOWNk ,t (18)
Table 6.2: Correlation Coefficients for the Discretionary Accruals Analysis
method investors having an incentive to manage earnings from affiliates in 13.7%
(175/1,282) of the firm-quarters in which an equity method relationship exists. In untabulated results, I find evidence of meeting the non-negative income target in 12.6%
(22/175) of the cases, the non-negative change in income target in 17.1% (30/175) of
the cases, and analysts’ forecasted earnings in 80.0% (140/175) of the cases.8 Table
6.2 shows that there is a positive correlation between signed discretionary accruals and
INCENTIVEIN V EST ORk,t [Spearman (Pearson) coefficient = 0.024 (0.027)], providing
preliminary evidence that equity method investors manage earnings of affiliates to
achieve their own income targets. There is also a negative correlation between signed
discretionary accruals and equity method ownership [Spearman (Pearson) coefficient
= -0.029 (-0.083)].
The data required to calculate the control variables included in Equation 6.1 reduce the sample from 250 affiliates to 202 affiliates. In the first two columns of Table
6.3, I report the results of an estimation of the coefficients on AFFILIATEk,t and
AFFILIATEk,t * INCENTIVEIN V EST ORk,t including control variables for the affiliate
having
an
incentive
to
manage
its
own
discretionary
accruals
(INCENTIVEAF F ILIAT Ek,t ) and for the reversal of any build-up of discretionary accruals created in periods in which the investor or affiliate had an incentive to manage
earnings to meet an earnings target (INCENTIVE REVERSALIN V EST ORk,t and INCENTIVE
REVERSALAF F ILIAT Ek,t ).
The
coefficient
on
AFFILIATEk,t
* INCENTIVEIN V EST ORk,t is positive (β 2 = 0.030, t-statistic = 2.929). This result
indicates that signed discretionary accruals of affiliates are higher when income from
affiliates allows equity method investors to meet earnings targets. This result provides
evidence that equity method investors take advantage of the mechanics of the equity
method (the flow-through of affiliate income to investor income) to manage earnings
8
The total exceeds 100% because in 17 cases income from the affiliate allows the investor to meet
two earnings targets.
58
of affiliates to achieve their own income targets. When I include all of the control
variables (third and fourth columns of Table 8), the coefficient on AFFILIATEk,t *
INCENTIVEIN V EST ORk,t remains positive (β 2 = 0.015, t-statistic = 1.724).
Table 6.3: Relation between Equity Method Ownership and Discretionary Accruals
of Affiliates
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
IN CEN T IV EIN V EST ORk,t
β2 AF F ILIAT Ek,t
β3 IN CEN T IV EAF F ILIAT Ek,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
β7 CAP IT ALk ,t
β8 DEBTk ,t
β10 SALESGROW T Hk ,t
β13 SALESV OLk ,t
β16 DN IOW Nk ,t
β6 OP CY CLEk ,t
β9 M T Bk ,t
β11 LOSSk ,t
β14 T RAOW Nk ,t
β17 M ON OW Nk ,t
(6.1)
β12 ASSET Sk ,t
β15 QIXOW Nk ,t
εk ,t
Dependent variable
DISC ACCk ,t
Coefficient
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
OPCYCLEk ,t
CAPITALk ,t
LEVk ,t
MTBk ,t
SALESGROWTHk ,t
LOSSk ,t
ASSETSk ,t
SALESVOLk ,t
TRAOWNk ,t
QIXOWNk ,t
DNIOWNk ,t
MONOWNk ,t
Intercept
Coefficient
0.016
3.972
0.014
0.021
2.492
0.030
0.015
0.002
0.005
Year fixed effects
Firm fixed effects
Included
Included
Firm-quarters
Firms
Adjusted R2
t-statistic
Dependent variable
DISC ACCk ,t
11,446
250
0.140
2.929
3.942
0.165
1.229
0.015
0.003
0.003
0.001
0.000
0.009
0.003
0.000
0.003
0.037
0.001
0.000
0.017
0.005
0.036
0.011
0.007
t-statistic
3.579
1.724
1.090
0.360
0.422
2.851
0.950
3.748
0.092
1.065
10.179
0.455
1.004
0.874
0.337
0.897
0.379
0.474
Included
Included
4,937
202
0.474
The table reports the results of estimation of Equation 6.1 using pooled OLS regression with quarterly data from
1993-2011 for the U.S. publicly-traded affiliates in my sample with the required data available. Two-tailed p-values
are reported as follows: p 0.10, p 0.05, p 0.01. The variables are defined in Table 6.1. All variables
(except indicator variables) are winsorized at the 1st and 99th percentiles.
I also find that the coefficient on AFFILIATEk,t is negative (β 1 = -0.016, t-statistic
= 3.972 with limited control variables; β 1 = -0.014, t-statistic = 3.579 with the full set
of control variables), indicating that equity method ownership is negatively associated
59
with signed discretionary accruals of affiliates. One possible explanation is that equity
method investors monitor management actions and mitigate the effects of actions
taken by affiliates’ management to inflate reported earnings. I do not further explore
or isolate the cause of the negative association in this paper.
In summary, I find that signed discretionary accruals of affiliates are higher when
income from affiliates allows equity method investors to meet earnings targets. I also
find a negative relation between equity method ownership and signed discretionary
accruals of affiliates in other periods. The result provides evidence that the equity
method does not eliminate concerns about earnings management, such as those expressed in relation to the cost method; it changes the mechanism through which
earnings are managed. The result also suggests that an incentive to manage the
earnings of affiliates can change the effect of equity method ownership on signed discretionary accruals of affiliates and confirms the importance of considering investors’
characteristics in studies of the effects of large shareholders on investees.
6.4 Supplementary Analyses
6.4.1
Effect of “True” Significant Influence
In this section, I investigate whether there is a difference between “nominal” significant influence and “true” significant influence in my sample and re-run my earnings
management tests to examine whether the results exist only in, or are stronger in,
the portion of the sample with “true” significant influence. There is a presumption
in APB 18 (para. 17, ASC 323-10-15-8) that, absent evidence to the contrary, an
investment of 20% or more indicates the ability to exercise significant influence over
the operating and financial policies of an investee. As a result, it is possible that
the equity method is applied in situations where ownership is between 20-50%, but
significant influence is not exercised (“nominal” significant influence).
APB 18 (para. 17, ASC 323-10-15-6) identifies the following factors that might
60
indicate an ability to exercise significant influence:
Representation on the board of directors
Participation in policy-making processes
Material intra-entity transactions9
Interchange of managerial personnel
Technological dependency
Extent of ownership by an investor in relation to the concentration of other
shareholdings (but substantial or majority ownership of the voting stock of
an investee by another investor does not necessarily preclude the ability to
exercise significant influence by the investor).
I use these factors to investigate whether there is a difference between “nominal”
significant influence and “true” significant influence in my sample. In particular, I
collect data on the following manifestations of these factors in my sample:10
Existence of a subsidiary relationship with the affiliate either prior to or
after the application of the equity method (proxy for participation in policymaking processes). I expect that if an investor plans to acquire an affiliate
in the future or previously had control of an affiliate, that investor is more
likely to exert “true” significant influence.
Material transactions between the equity method investor and affiliate (measured as the magnitude of eliminations from the investor’s proportionate
9
I interpret the phrase “material intra-entity transactions” in APB 18 to mean material transactions between the equity method investor and affiliate.
10
In future work, I plan to extend my analysis to collect data on the investor’s representation on
the board of directors and management of the affiliate. Investors are not required to disclose this
information in their 10-K reports, except if it is the basis for the investor’s application of the equity
method when owning less than 20% of the affiliate. I plan to collect background information on
affiliates’ directors from the affiliates’ annual proxy statements.
61
share of the affiliate’s reported net income, including gains/losses arising
from transactions between the equity method investor and affiliate, used to
calculate income from an affiliate). I expect a positive relation between “true”
significant influence and the level of transactions between the equity method
investor and affiliate.
Level of ownership of affiliate (less than 20% vs. greater than 20%). I expect
that ownership of less than 20% indicates “true” significant influence because
investors at that ownership level are required to demonstrate the existence
of significant influence in order to apply the equity method. I do not have
a prediction for the relation between “true” significant influence and level of
ownership above 20%.
Holding period. I expect a positive relation between “true” significant influence and holding period.
I calculate each of these variables at the affiliate level. That is, I calculate the
average value for a variable separately for each affiliate over the sample period. I then
present descriptive statistics for each of the variables averaged across my sample:
Variable
Affiliate is subsidiary prior to application of equity method
(indicator variable)
Affiliate is subsidiary after application of equity method
(indicator variable)
Equity method investor’s proportionate share of affiliate’s
reported net income - income from affiliate recognized
by investor (in $ millions)
Percentage of ownership of affiliate
Number of years in which firm is held as an affiliate
|
|
N
Mean
Std Dev
P25
Median
P75
414
0.22
0.41
0.00
0.00
0.00
414
0.22
0.42
0.00
0.00
0.00
353
392
414
26.94
0.29
4.17
83.26
0.13
3.27
0.73
0.21
1.84
4.03
0.29
3.25
17.04
0.38
5.38
The variation in these variables indicates that it is possible that there is a difference
between “nominal” significant influence and “true” significant influence in my sample.
I re-run my earnings management tests splitting my sample on these factors and find
some evidence that the results exist only in, or are stronger in, the portion of the
62
sample with “true” significant influence. As shown in Table 6.4, I find that the
earnings management results are stronger for the subsample of firms (a) that were
subsidiaries prior to becoming affiliates (p-value for test of equality of coefficients
of 0.0000), (b) with ownership percentage less than 20% (p-value of 0.0000), and
(c) with longer holding periods (p-value of 0.0001). These results are consistent
with the intuition that earnings management should be stronger for those investors
that exercise “true” significant influence. I do not find the predicted results for the
existence of a subsidiary relationship after the application of the equity method or
material transactions between the equity method investor and affiliate. It is possible
that the measure of material transactions between the equity method investor and
affiliate is too noisy to capture the effect even if one exists. This is because the
measure includes effects (e.g., treating any difference between the acquisition cost of
the investment and the investor’s share of the underlying equity in the net assets of the
affiliate as if the affiliate were a consolidated subsidiary) other than the elimination
of gains/losses arising from transactions between the equity method investor and
affiliate.
6.4.2
Differences in Earnings Management of Public Affiliates and Private Affiliates
I also consider whether there are differences in the incentives of equity method
investors to exercise significant influence over a private affiliate rather than a public
affiliate and the related implications for my earnings management findings.11 In
Chapter 6.1, I suggest that one motivation for equity method investors to choose to
11
In addition, it might be easier to exercise significant influence over the financial reporting of
an affiliate when there is a well-defined channel through which the influence can be exerted. One
example of a well-defined channel is overlap in the management of the equity method investor and
the affiliate (e.g., the affiliate’s CEO is appointed by/associated with the equity method investor).
Unfortunately, I do not have data available to permit me to assess whether managerial overlap between equity method investors and affiliates is more common for private affiliates or public affiliates.
If it is the case that there is more managerial overlap for private affiliates, I would expect that it is
easier for equity method investors to manage earnings from private affiliates.
63
Table 6.4: Effect of “True” Significant Influence on the Relation between Equity
Method Ownership and Discretionary Accruals of Affiliates
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
β2 AF F ILIAT Ek,t
β3 IN CEN T IV EAF F ILIAT Ek,t
IN CEN T IV EIN V EST ORk,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
Dependent variable
DISC ACCk ,t
Coefficient
Full Sample
0.016
0.030
0.015
0.002
0.005
0.021
Included
Included
11,446
0.140
Dependent variable
DISC ACCk ,t
t-stat
Coefficient
Subsidiary Prior to Affiliate Relationship
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
Subsidiary After Affiliate Relationship
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
Material Intra-Entity Transactions
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
Ownership Percentage
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t r
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
Length of Holding Period
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
0.016
3.972
εk ,t
3.972
2.929
3.942
0.165
1.229
2.492
Indicator = 0
0.014
0.024
0.017
0.004
0.006
0.023
Included
Included
9,180
0.140
Dependent variable
DISC ACCk ,t
t-stat
Coefficient
3.034
2.053
3.744
0.305
1.333
2.460
t-stat
3.412
3.167
1.412
0.440
0.121
0.880
0.016
3.972
0.024
5.116
0.016
3.972
Median
0.023
3.736
¡ Median
0.008
1.428
0.016
3.972
20%
0.040
5.465
¡20%
0.009
1.771
Full Sample
0.030
0.015
0.002
0.005
0.021
Included
Included
11,446
0.140
Full Sample
0.030
0.015
0.002
0.005
0.021
Included
Included
11,446
0.140
Full Sample
0.030
0.015
0.002
0.005
0.021
Included
Included
11,446
0.140
Full Sample
0.030
0.015
0.002
0.005
0.021
Included
Included
11,446
0.140
2.929
3.942
0.165
1.229
2.492
2.929
3.942
0.165
1.229
2.492
2.929
3.942
0.165
1.229
2.492
2.929
3.942
0.165
1.229
2.492
Indicator = 0
0.043
0.018
0.002
0.004
0.025
Included
Included
9,048
0.156
0.035
0.027
0.001
0.006
0.031
Included
Included
5,524
0.167
0.059
0.038
0.006
0.001
0.055
Included
Included
2,987
0.243
Median
0.000
0.002
0.017
0.005
0.007
0.021
Included
Included
5,155
0.000
Indicator = 1
0.027
0.061
0.011
0.008
0.001
0.015
Included
Included
2,082
0.170
Test of Equality
of Coefficients
3.387
3.978
0.179
0.924
2.375
Indicator = 1
0.012
0.008
0.004
0.005
0.010
0.013
Included
Included
2,214
0.026
2.337
4.390
0.081
0.909
2.589
2.555
5.570
0.238
0.176
3.674
0.055
0.067
2.681
0.201
1.001
1.690
0.023
0.006
0.005
0.004
0.009
Included
Included
5,922
0.108
0.023
0.006
0.000
0.006
0.008
Included
Included
8,275
0.111
1.729
0.502
0.522
0.317
1.239
1.023
1.676
1.241
0.338
0.714
0.792
1.985
1.353
0.000
1.278
0.805
¡ Median
0.025
5.193
0.039
0.014
0.003
0.004
0.017
Included
Included
6,291
0.243
3.579
2.850
0.300
0.708
1.534
p-value
.0003
.0000
.3733
.5167
.3356
.0001
.0000
.0238
.1214
.3351
.0000
.0004
.0109
.4507
.2237
.0000
.0000
.0000
.0566
.3606
.0391
.0001
.4413
.1247
.0990
The table evaluates the effect of “true” significant influence on the relation between equity method ownership and discretionary accruals of affiliates.
The table reports the results of estimation using pooled OLS regression with quarterly data from 1993-2011 for the U.S. publicly-traded affiliates
in my sample with the required data available. Two-tailed p-values are reported as follows: p 0.10, p 0.05, p 0.01. The variables
are defined in Table 6.1 and Chapter 6.4.1.
manage income from affiliates as opposed to other avenues of earnings management
is the lack of transparent information available about affiliates. This arises because
of the one-line presentation of the affiliate in the investor’s balance sheet and income
statement and the potential lack of disaggregated information about affiliates in the
notes to the investor’s financial statements. As a result, it might be difficult for users
64
of the investor’s financial statements to identify/detect earnings management in the
affiliate.
In the case of public affiliates, this lack of transparent information provided by the
investor could be mitigated by reviewing the publicly-available financial statements
of the affiliate. Therefore, equity method investors might have a stronger incentive
to manage income from private affiliates because it would be more difficult to detect
the earnings management. In addition, it is possible that some private affiliates are
subject to less scrutiny from auditors. For example, a 50-50 joint venture with no
debt financing might not have a demand for audited financial statements. Therefore,
it might be easier to avoid detection of earnings management when no auditor is
reviewing the related financial statements of the affiliate.
An affiliate might also have its own incentives to manage its earnings. If these
incentives conflict with the investor’s desire to manage earnings, it might be more
difficult for the investor to affect the financial reporting of the affiliate. To the extent
that these incentives arise more often in public affiliates (e.g., the public affiliate
might desire to manage its earnings for an SEO), an equity method investor might be
more likely to try to manage earnings from a private affiliate than a public affiliate.
This discussion suggests that it might be easier/more likely for an equity method
investor to manage the earnings of a private affiliate than a public affiliate. I investigate this issue by evaluating whether evidence of earnings management exists only for
(or is stronger for) the subsample of equity method investors with no private affiliates.
In that case, equity method investors do not have the option of using private affiliates
to manage their earnings and thus might be more likely to manage the earnings of
public affiliates.
I find some preliminary evidence to support this argument as shown in Table 6.5.
In particular, I find that the coefficient on AFFILIATEk,t * INCENTIVEIN V EST ORk,t
is larger when the equity method investor does not have any private affiliates (p-value
65
for test of equality of coefficients of 0.0058). However, the coefficient on AFFILIATEk,t
* INCENTIVEIN V EST ORk,t is still significantly different from zero in the subsample of
equity method investors with both public and private affiliates (β 2 = 0.030, t-statistic
= 2.266).
Table 6.5: Effect of Existence of Private Affiliates on the Relation between Equity
Method Ownership and Discretionary Accruals of Affiliates
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
β2 AF F ILIAT Ek,t
β3 IN CEN T IV EAF F ILIAT Ek,t
IN CEN T IV EIN V EST ORk,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
Dependent variable
DISC ACCk ,t
Coefficient
Investor has only Public Affiliates
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Firm fixed effects
Firm-quarters
Adjusted R2
Full Sample
0.016
0.030
0.015
0.002
0.005
0.021
Included
11,446
0.140
εk ,t
Dependent variable
DISC ACCk ,t
t-stat
Coefficient
Indicator = 0
0.013
0.030
0.015
0.010
0.007
0.026
Included
7,220
0.176
3.972
2.929
3.942
0.165
1.229
2.492
t-stat
2.499
2.266
3.100
0.749
1.368
2.475
Dependent variable
DISC ACCk ,t
Coefficient
Indicator = 1
0.027
0.034
0.015
0.016
0.002
0.013
Included
4,042
0.072
t-stat
Test of Equality
of Coefficients
p-value
3.878
.0006
.0058
.6825
.5308
.2352
2.088
2.314
1.020
0.326
0.919
The table evaluates the effect of the existence of private affiliates on the relation between equity method ownership and discretionary accruals of
affiliates. The table reports the results of estimation using pooled OLS regression with quarterly data from 1993-2011 for the U.S. publicly-traded
affiliates in my sample with the required data available. Two-tailed p-values are reported as follows: p 0.10, p 0.05, p 0.01. The
variables are defined in Table 6.1 and Chapter 6.4.2.
6.4.3
Potential for Mechanical Relation Between Investor Incentive and Outcome
Variables
I consider whether a purely mechanical relation exists between the incentive measures and the outcome measures in my analyses of meeting earnings targets. Income
from affiliates is one component of an equity method investor’s net income. All other
things equal, higher income from affiliates leads to higher investor net income. However, higher income from affiliates in a period does not necessarily equate to being
more likely to meet a specified earnings target if analysts have already factored the
higher income from affiliates into their forecasted earnings. I also investigate this
concern by re-defining my investor incentive variable as an investor just meeting or
beating an earnings target. Higher income from affiliates does not have a mechanical
relation with just meeting or beating an earnings target because higher income from
66
affiliates could also lead to the investor far exceeding an earnings target. As shown
in Table 6.6, I find qualitatively similar earnings management results in that specification (β 2 = 0.026, t-statistic = 3.705), suggesting that a mechanical relation is not
driving my results. I do not choose to use that definition of investor incentive because
it does not capture whether income from affiliates is large enough to be “important”
to investors.
Table 6.6: Effect of Defining Investor Incentive Variable as Just Meeting or Beating
an Earnings Target on the Relation between Equity Method Ownership and Discretionary Accruals of Affiliates
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
β2 AF F ILIAT Ek,t
β3 IN CEN T IV EAF F ILIAT Ek,t
IN CEN T IV EIN V EST ORk,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
εk ,t
Dependent variable
DISC ACCk ,t
Coefficient
AFFILIATEk,t
AFFILIATEk,t *INCENTIVEIN V EST ORk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALIN V EST ORk,t
INCENTIVE REVERSALAF F ILIAT Ek,t
Intercept
Firm fixed effects
Firm-quarters
Adjusted R2
t-statistic
0.019
4.495
0.026
0.015
0.003
0.005
0.020
Included
11,446
0.140
3.705
3.884
0.420
1.087
2.480
The table evaluates the effect of defining the investor incentive variable as just meeting or beating and earnings
target on the relation between equity method ownership and discretionary accruals of affiliates. The table reports the
results of estimation using pooled OLS regression with quarterly data from 1993-2011 for the U.S. publicly-traded
affiliates in my sample with the required data available. Two-tailed p-values are reported as follows: p 0.10, p 0.05, p 0.01. INCENTIVEIN V EST ORk,t : indicator variable equal to 1 if the equity method investor
of firm k just meets one of three earnings targets (non-negative income, non-negative change in income, analysts’
forecasted earnings) in quarter t, and 0 otherwise. Other variables are defined in Table 6.1.
6.4.4
Additional Investigation into Effects of Accruals Reversal
As described in Chapter 6.3, equity method ownership is negatively associated
with signed discretionary accruals of affiliates in periods when investors have weaker
incentives to manage income from affiliates. To address the concern that this relation
is the result of a reversal of an accumulated earnings reserve that was built up using
accruals-based earnings management, I include control variables in my analysis to
67
capture the expected reversal period of any build-up of discretionary accruals. I
also consider the behavior of discretionary accruals for the portion of my sample that
never has an equity method investor with a non-zero incentive variable (189 out of 250
affiliates with required data available). In this subsample, there is still a negative and
statistically significant relation between equity method ownership and discretionary
accruals of affiliates (see Table 6.7; β 1 = -0.018, t-statistic = -3.516). This suggests
that some factor other than accruals reversal is driving the negative relation between
equity method ownership and discretionary accruals of affiliates.
6.4.5
Other Potential Methods of Managing Earnings from Investments in Affiliates
I also examine whether there are opportunities for equity method investors to manage earnings related to investments in affiliates other than influencing discretionary
accruals.12
Small changes in ownership around 20%
As noted above, there is a presumption in APB 18 that, absent evidence to the
contrary, an investment of 20% or more indicates the ability to exercise significant and
an investment of less than 20% indicates a lack of significant influence unless such
influence can be demonstrated. Therefore, if an investor were purely applying the
12
A potential channel for earnings management is the original measurement of individual affiliate assets and liabilities at fair value. These individual measurements are not observable in the
investor’s financial statements. Individual assets and liabilities of the affiliate (including goodwill)
are not tested for impairment and thus reducing potential future impairment losses is not a reason
to manipulate original measurements of assets and liabilities. However, any difference between the
acquisition cost of the investment and the investor’s share of the underlying equity in the net assets
of the affiliate is treated as if the affiliate were a consolidated subsidiary. Prior to the issuance of
SFAS 142, Goodwill and Other Intangible Assets, the residual was treated like purchased goodwill
and amortized. Therefore, there might have been an incentive for investors to reduce the amount of
goodwill recognized as much as possible to mitigate goodwill amortization charges in future periods.
Since the issuance of SFAS 142, investors might try to increase the amount of goodwill recognized
as much as possible and lower the fair values assigned to other depreciable assets (e.g., property,
plant, and equipment that will be depreciated in the investor’s financial statements based on the
assigned fair value at the time of the equity method acquisition). Unfortunately, I cannot observe
these potential effects using publicly-available data.
68
Table 6.7: Relation between Equity Method Ownership and Discretionary Accruals
of Affiliates for Investors with INCENTIVEIN V EST ORk,t = 0 in All Periods
DISC ACCk ,t
β0
β1 AF F ILIAT Ek,t
β2 AF F ILIAT Ek,t
β3 IN CEN T IV EAF F ILIAT Ek,t
IN CEN T IV EIN V EST ORk,t
β4 IN CEN T IV EREV ERSALIN V EST ORk,t
β5 IN CEN T IV EREV ERSALAF F ILIAT Ek,t
β7 CAP IT ALk ,t
β8 DEBTk ,t
β10 SALESGROW T Hk ,t
β13 SALESV OLk ,t
β16 DN IOW Nk ,t
β6 OP CY CLEk ,t
β9 M T Bk ,t
β11 LOSSk ,t
β14 T RAOW Nk ,t
β17 M ON OW Nk ,t
β12 ASSET Sk ,t
β15 QIXOW Nk ,t
εk ,t
Dependent variable
DISC ACCk ,t
Coefficient
AFFILIATEk,t
INCENTIVEAF F ILIAT Ek,t
INCENTIVE REVERSALAF F ILIAT Ek,t
OPCYCLEk ,t
PAYABLEk ,t
CAPITALk ,t
LEVk ,t
MTBk ,t
SALESGROWTHk ,t
LOSSk ,t
ASSETSk ,t
SALESVOLk ,t
TRAOWNk ,t
QIXOWNk ,t
DNIOWNk ,t
MONOWNk ,t
Intercept
Year fixed effects
Firm fixed effects
Firm-quarters
Adjusted R2
t-statistic
0.018
0.005
0.005
0.000
3.516
1.080
1.113
2.055
0.025
0.002
0.004
0.000
0.001
0.043
0.002
0.000
0.017
0.019
0.041
0.011
0.019
2.009
0.219
4.164
0.065
0.273
8.746
0.491
0.220
0.693
0.894
0.679
0.292
0.977
Included
Included
3,564
0.495
The table evaluates the relation between equity method ownership and discretionary accruals of affiliates when
investors do not have an incentive to manage income from affiliates. The table reports the results of estimation
using pooled OLS regression with quarterly data from 1993-2011 for the U.S. publicly-traded affiliates in my sample
with the required data available and with equity method investors with INCENTIVEIN V EST ORk,t = 0 in all
periods. Two-tailed p-values are reported as follows: p
0.10, p
0.05, p
0.01. The variables are
defined in Table 6.1.
quantitative thresholds in APB 18, a small change in ownership around 20% would
result in a change in accounting method (for my sample of public firms, from fair value
measurement to the equity method or vice versa). This presents an opportunity for
the investor to manage its earnings. For example, it is possible for an investee to
have a positive change in its fair value during a reporting period but recognize a loss.
In that situation, the investor might want to change its ownership to 19%, measure
69
its investment at fair value, and recognize positive income from the investment. In
contrast, the investee might recognize positive net income during a reporting period,
but suffer a significant decline in fair value. In that situation, the investor might want
to increase its ownership above 20%, apply the equity method, and recognize positive
income from the investment.
I investigate the occurrence of this type of earnings management in my sample by
examining the periods in which an investor changed from ownership in an investee
below significant influence (e.g., AFS security) to the equity method or vice versa. In
an untabulated result using the Fisher’s exact test, I find no statistically significant
relation between the sign of investee net income and the sign of changes in investee’s
fair value during the reporting period in which a change to the equity method is made
(p-value of 0.245). I find a statistically significant relation between the sign of investee
net income and the sign of changes in investee’s fair value during the reporting period
in which a change from the equity method to fair value is made (p-value of 0.065),
but not in the direction predicted under a hypothesis of earnings management.
Change from ownership in an investee below significant influence (e.g., AFS secu-
Investee’s Net Income during
Reporting Period
rity) to the equity method:
Negative
Change in Investee’s Fair Value
during Reporting Period
Negative
Positive
11
3
Total
14
Positive
8
7
15
Total
19
10
29
Change from the equity method to ownership in an investee below significant
influence (e.g., AFS security):
70
Investee’s Net Income during
Reporting Period
Negative
Change in Investee’s Fair Value
during Reporting Period
Negative
Positive
39
9
Total
48
Positive
18
12
30
Total
57
21
78
Aligning or misaligning the fiscal-year-ends of investors and investees
Paragraph 19(g) of APB 18 states, “If financial statements of an investee are not
sufficiently timely for an investor to apply the equity method currently, the investor
ordinarily should record its share of the earnings or losses of an investee from the
most recent available financial statements. A lag in reporting should be consistent
from period to period.”
An investor makes a decision at the initial application of the equity method about
whether to use current or lagged results of the affiliate. That decision does not change
the overall amount of income from the affiliate recognized in the investor’s net income,
but it can change the timing of the income that is recognized. Therefore, an opportunity for earnings management exists. For example, assume (a) an investor knows
that it is going to just meet its earnings target for the period without incorporating
income from its newly-acquired affiliate and (b) incorporating the results of the affiliate would cause the investor to miss its earnings target. The investor might decide
to use the lagged results and incorporate the loss from its affiliate in the next period.
I identified 29 investor-affiliate pairs where a lagged reporting date for the affiliate
was used. In those cases, net income of the affiliate in the initial ownership period
was negative 65.52% (19/29) of the time. The occurrence of negative affiliate income
in the application period when the investor does not choose to use a lagged reported
71
period was 62.24% (305/490). An untabulated Fisher’s exact test suggests that there
is no association between the incidence of negative net income of the affiliate and
the decision to use a lagged reporting date (p-value = .844). This suggests that the
alignment of fiscal year ends does not seem to be a tool for earnings management.
6.4.6
Reverse Causality/Endogeneity
My results are consistent with equity method investors influencing affiliates to
report lower signed discretionary accruals. It is possible instead that firms with lower
signed discretionary accruals attract equity method investors (a reverse causality
explanation). I investigate the change in signed discretionary accruals in the period
prior to equity method ownership. In an untabulated result, I find that there is
no change in the signed discretionary accruals (p-value = 0.341) of affiliates in the
year prior to equity method ownership. This result mitigates concerns about reverse
causality.
It is also possible that unidentified characteristics of the firms in my sample are
correlated with signed discretionary accruals and also make them attractive equity
method investments. My within-firm research design mitigates this issue by showing
discretionary accruals are lower in periods when a firm is an affiliate relative to periods
when a firm is not an affiliate.13
6.4.7
Effect of Equity Method Ownership on Accounting Conservatism of Affiliates
As noted in Chapter 6.3, I find that equity method ownership is negatively associated with signed discretionary accruals of affiliates when investors have weaker
incentives to manage income from affiliates. I suggest that this result is consistent
with the effects of monitoring by large shareholders, but do not further explore or
13
Another way to mitigate concerns about endogeneity would be use a two-stage test with a
determinant model for equity method investments as the first stage. However, there is no well-defined
determinant model for equity method investments in the literature. I am considering developing such
a model as an extension to my paper.
72
isolate the cause of the negative association. In this section, I examine another
manifestation of monitoring studied in the large shareholder literature: accounting
conservatism. Previous research suggests that “monitoring” shareholders demand accounting conservatism because it provides more timely indications of bad news, which
allows shareholders to influence management’s decisions and/or discipline management earlier (e.g., Watts (2003), Chen et al. (2009), Ahmed and Duellman (2011),
Ramalingegowda and Yu (2012)).
Accounting conservatism and earnings management are related in that both violate the concept of neutrality from Chapter 3, Qualitative Characteristics of Useful
Financial Information, of the FASB’s Conceptual Framework. Paragraph QC14 of
SFAC No. 8 states, “A neutral depiction is without bias in the selection or presentation of financial information. A neutral depiction is not slanted, weighted, emphasized, deemphasized, or otherwise manipulated to increase the probability that
financial information will be received favorably or unfavorably by users.” The Basis
for Conclusions (para. BC3.27) explicitly explains that conservatism is not included
as an aspect of faithful representation because it would be inconsistent with neutrality. My earnings management tests focus on biasing reported results by increasing
reported performance, while accounting conservatism decreases reported performance.
However, an equity method investor might not need to rely on public financial
reports in the same way as other investors (e.g., institutional investors) studied in the
large shareholder literature. By definition, an equity method investor has “significant
influence” over the operating and financial policies of an affiliate. This implies that
equity method investors have access to information necessary to affect operating and
financial decisions of affiliates. Such information likely is not available through public
financial reporting channels. Therefore, I might not expect to find a positive relation
between equity method ownership and accounting conservatism of affiliates.
I follow prior research examining the relation between accounting conservatism
73
and ownership characteristics (e.g., Ramalingegowda and Yu (2012)) by using the
Basu (1997) model, modified as follows:
N Ik ,t
β0
β1 N EGk ,t
β2 RETk ,t
β5 N EGk ,t AF F ILIAT Ek ,t
β3 RETk ,t N EGk ,t
β4 AF F ILIAT Ek ,t
β6 RETk ,t AF F ILIAT Ek ,t
β7 RETk ,t N EGk ,t AF F ILIAT Ek ,t
β18 27 N EGk ,t CON T ROLSk ,t 1
β8 17 CON T ROLSk ,t 1
β28 37 RETk ,t CON T ROLSk ,t 1
β38 47 RETk ,t N EGk ,t CON T ROLSk ,t 1
εk ,t
(6.5)
where:
NIk ,t : income before extraordinary items (IBQ) of firm k in quarter t, scaled
by the market value of equity (CSHOQ*PRCCQ) at the end of quarter t - 1;
NEGk ,t : indicator variable equal to 1 if RETk ,t is negative, and 0 otherwise;
RETk ,t : buy-and-hold stock returns of firm k over quarter t;
AFFILIATEk,t : an indicator variable equal to 1 if firm k is an affiliate (i.e.,
has an owner using the equity method to account for its investment in firm
k) in quarter t, and 0 otherwise;
MONOWNk ,t 1 : percentage ownership of firm k at the end of quarter t - 1
by institutional investors described as monitoring institutions in Ramalingegowda and Yu (2012) based on classification as dedicated by Bushee (2001)
and independent by Brickley et al. (1988);
TRAOWNk ,t 1 : percentage ownership of firm k at the end of quarter t - 1
by institutional investors described as transient by Bushee (2001);
QIXOWNk ,t 1 : percentage ownership of firm k at the end of quarter t - 1 by
institutional investors described as quasi-indexing by Bushee (2001);
74
DNIOWNk ,t 1 : percentage ownership of firm k at the end of quarter t - 1
by institutional investors described as dedicated by Bushee (2001) and nonindependent by Brickley et al. (1988);
STD RETk ,t 1 : standard deviation of daily stock returns of firm k over quarter t - 1;
AGEk ,t 1 : age of firm k at the end of quarter t - 1, measured as the number
of years a firm is listed on Compustat;
MVEk ,t 1 : market value of equity (CSHOQ*PRCCQ) of firm k at the end of
quarter t - 1;
MTBk ,t 1 : market-to-book ratio (MVE/CEQQ) of firm k at the end of quarter t - 1;
LEVk ,t 1 : leverage ((DLTTQ + DLCQ)/MVE) of firm k at the end of quarter
t - 1; and
LITk ,t 1 : indicator variable equal to 1 if firm k is in one of the following
industries at the end of quarter t - 1: biotechnology (SIC codes 2833-2836
and 8731-8734), computers (SIC codes 3570-3577 and 7370-7374), electronics
(SIC codes 3600-3674), and retail (SIC codes 5200-5990), and 0 otherwise.
I estimate Equation 6.5 using pooled OLS regression for all firm-quarters from
1993-2011 for the U.S. publicly-traded affiliates in my sample with the required data
available. I winsorize all variables at the 1st and 99th percentiles and include year
and firm fixed effects in the regression. The estimation results are reported in Table
6.8. The coefficient on RETk ,t * NEGk ,t * AFFILIATEk,t is positive (β 7 = 0.112,
t-statistic = 3.830 for model without control variables; β 7 = 0.123, t-statistic = 3.288
for model with control variables), indicating a positive association between equity
method ownership and the incremental timeliness of earnings with respect to bad
75
news. This result is consistent with the view, expressed by Ramalingegowda and Yu
(2012) and others, that large shareholders use conservative financial reporting to assist
them in monitoring/disciplining management by providing more timely indications
of bad news.
76
Table 6.8: Relation between Equity Method Ownership and Accounting Conservatism
of Affiliates
N Ik ,t
β0
N EGk ,t β4 AF F ILIAT Ek ,t β5 N EGk ,t AF F ILIAT Ek ,t
β6 RETk ,t AF F ILIAT Ek ,t β7 RETk ,t N EGk ,t AF F ILIAT Ek ,t β8 17 CON T ROLSk ,t 1
β18 27 N EGk ,t CON T ROLSk ,t 1 β28 37 RETk ,t CON T ROLSk ,t 1
β38 47 RETk ,t N EGk ,t CON T ROLSk ,t 1 εk ,t
β1 N EGk ,t
β2 RETk ,t
β3 RETk ,t
(6.5)
Dependent variable=N Ik,t
Coefficient
NEGk ,t
RETk ,t
RETk ,t *NEGk ,t
AFFILIATEk ,t
NEGk ,t *AFFILIATEk ,t
RETk ,t *AFFILIATEk ,t
RETk ,t *NEGk ,t *AFFILIATEk ,t
t-statistic
0.005
0.966
Intercept
Firm effects
Year effects
0.014
t-statistic
0.915
3.660
0.013
0.434
0.152
0.009
0.000
0.056
0.112
7.964
1.578
0.056
3.696
3.830
0.124
0.005
0.003
0.051
0.123
2.037
0.651
0.276
2.724
3.288
Included
Included
Included
0.002
Included
Included
Firm-quarters
Firms
Adjusted R2
Coefficient
0.035
CONTROLS
NEGk ,t *CONTROLS
RETk ,t *CONTROLS
RETk ,t *NEGk ,t *MONOWNk ,t 1
RETk ,t *NEGk ,t *TRAOWNk ,t 1
RETk ,t *NEGk ,t *QIXOWNk ,t 1
RETk ,t *NEGk ,t *DNIOWNk ,t 1
RETk ,t *NEGk ,t *STD RETk ,t 1
RETk ,t *NEGk ,t *AGEk ,t 1
RETk ,t *NEGk ,t *MVEk ,t 1
RETk ,t *NEGk ,t *MTBk ,t 1
RETk ,t *NEGk ,t *LEVk ,t 1
RETk ,t *NEGk ,t *LITk ,t 1
Dependent variable=N Ik,t
12,749
384
0.159
0.214
0.667
0.387
0.044
0.385
0.971
0.000
0.000
0.009
0.015
0.023
0.021
Included
Included
2.151
1.944
0.336
0.792
1.300
0.139
1.375
3.604
2.201
0.587
0.917
7,398
359
0.363
The table reports results of the estimation of Equation 6.5 using pooled OLS regression with quarterly data from
1993-2011 for the 359 U.S. publicly-traded affiliates in my sample with the required data available. For brevity, I do
not report coefficients for stand-alone control variables or the two-way interactions of control variables with NEG
and RET, but they are included in the estimation. Two-tailed p-values are reported as follows: p
0.10, p 0.05, p 0.01. The variables are defined as follows. All variables (except indicator variables) are winsorized
at the 1st and 99th percentiles. NIk ,t : income before extraordinary items (IBQ) of firm k in quarter t, scaled by
the market value of equity (CSHOQ*PRCCQ) at the end of quarter t - 1. NEGk ,t : indicator variable equal to 1 if
RETk ,t is negative, and 0 otherwise. RETk ,t : buy-and-hold stock returns of firm k over quarter t. AFFILIATEk,t :
an indicator variable equal to 1 if firm k is an affiliate (i.e., has an owner using the equity method to account for its
investment in firm k) in quarter t, and 0 otherwise. MONOWNk ,t 1 : percentage ownership of firm k at the end
of quarter t - 1 by institutional investors described as monitoring institutions in Ramalingegowda and Yu (2012)
based on classification as dedicated by Bushee (2001) and independent by Brickley et al. (1988). TRAOWNk ,t 1 :
percentage ownership of firm k at the end of quarter t - 1 by institutional investors described as transient by Bushee
(2001). QIXOWNk ,t 1 : percentage ownership of firm k at the end of quarter t - 1 by institutional investors
described as quasi-indexing by Bushee (2001). DNIOWNk ,t 1 : percentage ownership of firm k at the end of
quarter t - 1 by institutional investors described as dedicated by Bushee (2001) and non-independent by Brickley et
al. (1988). STD RETk ,t 1 : standard deviation of daily stock returns of firm k over quarter t - 1. AGEk ,t 1 : age
of firm k at the end of quarter t - 1, measured as the number of years a firm is listed on Compustat. MVEk ,t 1 :
market value of equity (CSHOQ*PRCCQ) of firm k at the end of quarter t - 1. MTBk ,t 1 : market-to-book ratio
(MVE/CEQQ) of firm k at the end of quarter t - 1. LEVk ,t 1 : leverage ((DLTTQ + DLCQ)/MVE) of firm k at
the end of quarter t - 1. LITk ,t 1 : indicator variable equal to 1 if firm k is in one of the following industries at
the end of quarter t - 1: biotechnology (SIC codes 2833-2836 and 8731-8734), computers (SIC codes 3570-3577 and
7370-7374), electronics (SIC codes 3600-3674), and retail (SIC codes 5200-5990), and 0 otherwise.
77
7
Link between Value Relevance and Earnings
Management
At a conceptual level, the value relevance and earnings management tests in my
paper are linked because both address the decision usefulness of the equity method
of accounting as described in Chapter 3, Qualitative Characteristics of Useful Financial Information, of the FASB’s Conceptual Framework. In particular, the value
relevance tests can be thought of as a joint test of the relevance and representational
faithfulness of information provided by measuring investments in affiliates under the
equity method or at fair value. The earnings management tests are an investigation
of the neutrality of information reported by equity method investors, which is one
component of representational faithfulness.
I also investigate whether there is an empirical link between my value relevance
and earnings management tests. In particular, I examine whether the existence of
an incentive for investors to manage income from affiliates affects the valuation coefficients of income reported under the equity method. As shown in Table 7.1, I do
not find evidence of differences in valuation coefficients in my main value relevance
specification depending on the incentives of investors to manage earnings (coefficients
78
on interaction terms all indistinguishable from zero). I then refine the analysis based
on the discussion in Chapter 6.4.2 that it might be more likely for an equity method
investor to manage the earnings of a private affiliate than a public affiliate.1 In particular, if investors believe that a firm with significant influence might manage income
from private affiliates more than income from public affiliates, we might expect the
valuation coefficients on amounts reported under the equity method to be smaller for
private affiliates than public affiliates. As shown in Table 7.2, I do not find significant
differences in the coefficients for the balance sheet and income statement amounts
reported under the equity method for private and public affiliates (p-value of 0.3769
for test of equality of BVprivateaffiliatej ,t = BVpublicaffiliatej ,t ; p-value of 0.4328 for test of
equality of INCprivateaffiliatej ,t = INCpublicaffiliatej ,t ). However, I do find that when investors have an incentive to manage income from affiliates, the valuation coefficient on
income from private affiliates is lower (coefficient on INCprivateaffiliatej ,t * EARNINGS
MGMTj ,t = -6.676, t-statistic = -2.161). One explanation is that investors understand the incentive and ability of firms with significant influence to manage income
from private affiliates and therefore adjust the valuation of that income accordingly.
1
I also investigate whether there are differences in valuation coefficients based on the length
of holding period since I find that equity method investors with longer holding periods also have
stronger earnings management results (see Chapter 6.4.1). In an untabulated result, I do not find
differences in any valuation coefficients based on length of holding period.
79
Table 7.1: Effect of Earnings Management by Investors on the Relation between
Investors’ Stock Prices and Balance Sheet and Income Measures of Investments in
Affiliates
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β5 IN Cpublicaffiliatej ,t
β6 F V IN C
β3 F V BVpublicaffiliatej ,t
IN Cpublicaffiliatej ,t εj ,t
Dependent variable = MVEj ,t
Coefficient
BVotherj ,t
BVotherj ,t * EARNINGS MGMTj ,t
BVpublicaffiliatej ,t
BVpublicaffiliatej ,t * EARNINGS MGMTj ,t
FV-BVpublicaffiliatej ,t
FV-BVpublicaffiliatej ,t * EARNINGS MGMTj ,t
INCotherj ,t
INCotherj ,t * EARNINGS MGMTj ,t
INCpublicaffiliatej ,t
INCpublicaffiliatej ,t * EARNINGS MGMTj ,t
FVINC-INCpublicaffiliatej ,t
FVINC-INCpublicaffiliatej ,t * EARNINGS MGMTj ,t i
EARNINGS MGMTj ,t
Intercept
0.774
t-value
10.092
1.113
3.673
0.712
3.726
0.811
4.285
2.132
1.591
0.396
2.186
14.408
5.560
Year fixed effects
Firm fixed effects
Firm-years
Adjusted R2
Included
Included
857
0.828
β4 IN Cotherj ,t
(5.1)
Dependent variable = MVEj ,t
Coefficient
t-value
0.746
0.115
1.107
0.391
0.750
0.085
0.824
0.112
1.066
3.577
0.264
0.679
0.028
14.640
9.516
1.100
3.572
0.687
3.705
0.225
4.131
0.200
0.629
1.040
1.335
1.355
0.013
5.586
Included
Included
857
0.828
The table incorporates into Equation 5.1 the effect of investors’ incentives to manage income from affiliates. EARNINGS
MGMTj ,t : indicator variable equal to 1 if income from affiliates causes firm j to meet one of three earnings targets
(non-negative income, non-negative change in income, analysts’ forecasted earnings) in year t, and 0 otherwise. Additional
variables are defined in Table 5.1. Two-tailed p-values are reported as follows: p
0.10, p
0.05, p
0.01.
80
Table 7.2: Effect of Separating Publicly-Traded and Private Affiliates on the Relation between Investors’ Stock Prices and Balance Sheet and Income Measures of
Investments in Affiliates
M V Ej ,t
β0
β1 BVotherj ,t
β2 BVpublicaffiliatej ,t
β5 IN Cpublicaffiliatej ,t
β6 F V IN C
β3 F V BVpublicaffiliatej ,t
IN Cpublicaffiliatej ,t εj ,t
Dependent variable = MVEj ,t
Coefficient
BVnonaffiliatej ,t
BVnonaffiliatej ,t * EARNINGS MGMTj ,t
BVpublicaffiliatej ,t
BVpublicaffiliatej ,t * EARNINGS MGMTj ,t
BVprivateaffiliatej ,t
BVprivateaffiliatej ,t * EARNINGS MGMTj ,t i
FV-BVpublicaffiliatej ,t
FV-BVpublicaffiliatej ,t * EARNINGS MGMTj ,t
INCnonaffiliatej ,t
INCnonaffiliatej ,t * EARNINGS MGMTj ,t
INCpublicaffiliatej ,t
INCpublicaffiliatej ,t * EARNINGS MGMTj ,t
INCprivateaffiliatej ,t
INCprivateaffiliatej ,t * EARNINGS MGMTj ,t
FVINC-INCpublicaffiliatej ,t
FVINC-INCpublicaffiliatej ,t * EARNINGS MGMTj ,t
EARNINGS MGMTj ,t
Intercept
9.373
1.195
3.926
0.857
3.798
0.599
3.055
0.818
4.036
1.877
1.369
0.678
0.647
0.415
2.284
13.467
5.098
Year fixed effects
Firm fixed effects
Firm-years
Adjusted R2
Tests of Equality of Coefficients:
t-value
0.770
Included
Included
784
0.834
β4 IN Cotherj ,t
(5.1)
Dependent variable = MVEj ,t
Coefficient
0.756
t-value
9.134
0.192
1.372
0.225
1.504
4.383
0.372
0.644
1.678
0.600
0.036
0.786
0.820
0.951
1.253
2.022
6.676
0.264
0.626
1.072
13.406
2.749
3.286
2.919
0.097
3.719
1.299
0.561
0.302
1.772
2.161
1.342
1.255
0.516
5.046
Included
Included
784
0.837
p-value
BVpublicaffiliatej ,t = BVprivateaffiliatej ,t
INCpublicaffiliatej ,t = INCprivateaffiliatej ,t
.3769
.4328
The table incorporates into Equation (1) the effect of separately identifying publicly-traded affiliates and private affiliates
and the effect of investors’ incentives to manage income from affiliates. BVnonaffiliatej ,t : book value (CEQ) of firm j less
the book value of its investments in affiliates (IVAEQ) at the end of year t. INCnonaffiliatej ,t : income before extraordinary
items (IB) less reported income from affiliates (ESUB) of firm j in year t. EARNINGS MGMTj ,t : indicator variable
equal to 1 if income from affiliates causes firm j to meet one of three earnings targets (non-negative income, non-negative
change in income, analysts’ forecasted earnings) in year t, and 0 otherwise. Additional regression details and variables are
defined in Table 4. Two-tailed p-values are reported as follows: p
0.10, p
0.05, p
0.01.
81
8
Conclusion
This paper examines the decision usefulness of the equity method of accounting.
For a sample of 221 U.S. investors with publicly-traded affiliates during 1993-2011,
I find that fair value balance sheet and income measures of investments in publiclytraded affiliates are incrementally associated with investors’ stock prices after controlling for information provided under the equity method. This result suggests that fair
value provides incremental relevant information, relative to the equity method, for investments in affiliates. In addition, I find that the incremental value relevance of fair
value measures exists for both investments in affiliates classified as held for sale and
those classified as strategic, with no evidence that the incremental value relevance
is higher (lower) for investments in affiliates identified as held for sale (strategic).
The result speaks to the ongoing debate at the FASB about whether measurement
attributes should be based on management’s intended method of value realization for
the items in question or whether the same measurement attribute should be applied
to similar items regardless of management’s intent. In particular, the result suggests
that significant influence might not be a basis for measuring investments in affiliates
that an entity does not intend to sell differently from other investments in marketable
82
equity securities (which are required to be measured at fair value under U.S. GAAP).
I also find evidence that the equity method does not achieve its objective of
reducing earnings management by investors with significant influence. For a sample
of 202 publicly-traded U.S. firms that were affiliates for a portion of the period 19932011, I find that signed discretionary accruals of affiliates are higher when income
from affiliates allows investors to meet earnings targets. This result indicates that the
equity method does not eliminate concerns about earnings management by investors
with significant influence (such as those expressed about the timing of dividends
under the cost method); instead, it changes the mechanism through which earnings
are managed.
83
Appendix A
Description of the Cost Method
APB 18 (para. 6) describes the cost method as follows: “An investor records an
investment in the stock of an investee at cost, and recognizes as income dividends
received that are distributed from net accumulated earnings of the investee since the
date of acquisition by the investor. The net accumulated earnings of an investee subsequent to the date of investment are recognized by the investor only to the extent
distributed by the investee as dividends.” In contrast, under the equity method, dividends received reduce the carrying amount of the investment and income recognition
by the investor is based on the investor’s share of the net income of the investee (adjusted as described in Chapter 2) and also results in a change in the carrying amount
of the investment. Impairment requirements are similar under the cost and equity
methods.
The cost method is distinct from amortized cost used to measure investments in
held-to-maturity debt securities under SFAS 115. Under the amortized cost method
(as described on page 570 of Stickney et al. (2010)), “A firm initially records these
debt securities at acquisition cost. This acquisition cost will differ from the maturity
84
value of the debt if the coupon rate on the bonds differs from the required market
yield on the bonds at the time the firm acquired them. The firm must use the
effective interest method to amortize any difference between acquisition cost and
maturity value over the life of the debt as an adjustment to interest revenue. ...
The amortization procedure involves the following steps: (1) The holder of the debt
securities (the investor) records interest revenue each period at an amount equal to
the carrying value of the debt at the start of the period multiplied by the market rate
of interest applicable to that debt on the day the firm acquired the debt. It debits
the Marketable Securities account and credits Interest Revenue, which after closing
entries increases Retained Earnings. (2) If the investor receives cash each period, it
debits Cash and credits the Marketable Securities account. The result of this process
is a new carrying value (called the amortized cost) for use in the computations during
the next period.”
The key difference between the amortized cost method and the equity method is
the basis for income recognition. Under the amortized cost method, income recognition is based on the amortization of any difference between acquisition cost and
maturity value over the life of the debt security held as interest revenue. The amortized cost method cannot be applied to investments in equity securities because equity
securities do not have maturity dates.
The impairment guidance for debt securities differs from that for equity securities. After it has been determined that an other-than-temporary impairment exists
for equity securities, the entire difference between the investment’s cost and its fair
value is recognized in earnings (ASC 320-10-35-34). For debt securities (ASC 32010-35-34D), the other-than-temporary impairment is separated into (a) the amount
representing the credit loss (exists when present value of cash flows expected to be
collected is less than the amortized cost basis of the security), which is recognized in
earnings, and (b) the amount related to all other factors, which is recognized in other
85
comprehensive income. The guidance for other-than-temporary impairments for debt
securities changed during my sample period. Prior to reporting periods ending after
June 15, 2009, once it was determined that an other-than-temporary impairment existed, the entire amount of the impairment was recognized in earnings. In addition,
the previous requirement to avoid recognizing an other-than-temporary impairment
was for an investor to assert that it has both the intent and the ability to hold a
security for a period of time sufficient to allow for an anticipated recovery in its fair
value to its amortized cost basis. Now, an entity must assess whether it (a) has the
intent to sell the debt security or (b) more likely than not will be required to sell the
debt security before its anticipated recovery. If either of these conditions is met, the
investor must recognize an other-than-temporary impairment. (FASB Staff Position
No. FAS 115-2 and FAS 124-2, para. 7)
86
Bibliography
Ahmed, A. and Duellman, S. (2007), “Accounting conservatism and board of director
characteristics: An empirical analysis,” Journal of Accounting and Economics, 43,
411–437.
Ahmed, A. and Duellman, S. (2011), “Evidence on the role of accounting conservatism
in monitoring managers investment decisions,” Accounting & Finance, 51, 609–633.
Ahmed, A. S. and Takeda, C. (1995), “Stock market valuation of gains and losses
on commercial banks’ investment securities An empirical analysis,” Journal of Accounting and Economics, 20, 207 – 225.
Ahmed, A. S., Kilic, E., and Lobo, G. J. (2006), “Does recognition versus disclosure matter? Evidence from value-relevance of banks’ recognized and disclosed
derivative financial instruments.” Accounting Review, 81, 567 – 588.
Baber, W. R., Kang, S., and Li, Y. (2011), “Modeling discretionary accrual reversal
and the balance sheet as an earnings management constraint,” The Accounting
Review, 86, 1189–1212.
Barrett, M. (1971), “Accounting for intercorporate investments: a behavior field experiment,” Journal of Accounting Research, pp. 50–65.
Barth, M. and Clinch, G. (2009), “Scale effects in capital markets-based accounting
research,” Journal of Business Finance & Accounting, 36, 253–288.
Barth, M. E. (1994), “Fair value accounting: evidence from investment securities and
the market valuation of banks,” Accounting Review, 69, I – 25.
Barth, M. E., Beaver, W. H., and Landsman, W. R. (1996), “Value-relevance of
banks’ fair value disclosures under SFAS No. 107,” Accounting Review, 71, 513 –
537.
Basu, S. (1997), “The conservatism principle and the asymmetric timeliness of earnings,” Journal of Accounting and Economics, 24, 3–37.
Bauman, M. (2003), “The impact and valuation of off-balance-sheet activities concealed by equity method accounting,” Accounting Horizons, 17, 303–314.
87
Bohren, O. and Haug, J. (2006), “Managing earnings with intercorporate investments,” Journal of Business Finance & Accounting, 33, 671–695.
Bohren, O., Haug, J., and Michalsen, D. (2004), “Compliance with flexible accounting
standards,” The International Journal of Accounting, 39, 1–19.
Bona-Sanchez, C., Perez-Aleman, J., and Santana-Martin, D. (2011), “Ultimate ownership and earnings conservatism,” European Accounting Review, 20, 57–80.
Brickley, J., Lease, R., and Smith, C. (1988), “Ownership structure and voting on
antitakeover amendments,” Journal of Financial Economics, 20, 267–291.
Bushee, B. (2001), “Do institutional investors prefer near-term earnings over long-run
value?” Contemporary Accounting Research, 18, 207–246.
Chambers, D., Linsmeier, T., Shakespeare, C., and Sougiannis, T. (2007), “An evaluation of SFAS No. 130 comprehensive income disclosures.” Review of Accounting
Studies, 12, 557 – 593.
Chen, S., Chen, X., Cheng, Q., and Hutton, A. (2009), “Accounting conservatism and
large shareholders: an empirical analysis,” Tech. rep., Working Paper, University
of Texas at Austin.
Chung, R., Firth, M., and Kim, J. (2002), “Institutional monitoring and opportunistic
earnings management,” Journal of Corporate Finance, 8, 29–48.
Comiskey, E. and Mulford, C. (1986), “Investment decisions and the equity accounting
standard,” Accounting Review, pp. 519–525.
Comiskey, E. and Mulford, C. (1988), “The influence of accounting principles on
management investment decisions: an illustration,” Accounting Horizons, 2, 67 –
72.
Cronqvist, H. and Fahlenbrach, R. (2009), “Large shareholders and corporate policies,” Review of Financial Studies, 22, 3941–3976.
Cullinan, C., Wang, F., Wang, P., and Zhang, J. (2012), “Ownership structure and
accounting conservatism in China,” Journal of International Accounting, Auditing
and Taxation, 21, 1–16.
Davis, M. and Largay III, J. (1999), “Financial reporting of “significant-influence”
equity investments: analysis and managerial issues,” Journal of Managerial Issues,
pp. 280–298.
Dechow, P. (1994), “Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals,” Journal of Accounting and Economics,
18, 3–42.
88
Dechow, P. and Dichev, I. (2002), “The quality of accruals and earnings: The role of
accrual estimation errors,” The Accounting Review, 77, 35–59.
Dechow, P., Sloan, R., and Sweeney, A. (1995), “Detecting earnings management,”
Accounting Review, pp. 193–225.
Dechow, P., Sloan, R., and Sweeney, A. (2010a), “Causes and consequences of earnings
manipulation: An analysis of firms subject to enforcement actions by the SEC,”
Contemporary Accounting Research, 13, 1–36.
Dechow, P., Ge, W., and Schrand, C. (2010b), “Understanding earnings quality:
A review of the proxies, their determinants and their consequences,” Journal of
Accounting and Economics, 50, 344–401.
Dhaliwal, D., Subramanyam, K., and Trezevant, R. (1999), “Is comprehensive income
superior to net income as a measure of firm performance?” Journal of Accounting
and Economics, 26, 43 – 67.
Easley, D., Hvidkjaer, S., and Ohara, M. (2002), “Is information risk a determinant
of asset returns?” The Journal of Finance, 57, 2185–2221.
Easton, P. (1999), “Security returns and the value relevance of accounting data,”
Accounting Horizons, 13, 399–412.
Easton, P. and Harris, T. (1991), “Earnings as an explanatory variable for returns,”
Journal of Accounting Research, pp. 19–36.
Eccher, E. A., Ramesh, K., and Thiagarajan, S. (1996), “Fair value disclosures by
bank holding companies,” Journal of Accounting and Economics, 22, 79 – 117.
Edwards, E. O. and Bell, P. W. (1961), The theory and measurement of business
income, Univ of California Press.
Fan, J. and Wong, T. (2002), “Corporate ownership structure and the informativeness
of accounting earnings in East Asia,” Journal of Accounting and Economics, 33,
401–425.
Francis, J. and Schipper, K. (1999), “Have financial statements lost their relevance?”
Journal of Accounting Research, 37, 319–352.
Givoly, D. and Hayn, C. (2000), “The changing time-series properties of earnings, cash
flows and accruals: Has financial reporting become more conservative?” Journal of
Accounting and Economics, 29, 287–320.
Gopalan, R. and Jayaraman, S. (2012), “Private control benefits and earnings management: Evidence from insider controlled firms,” Journal of Accounting Research,
50, 117–157.
89
Gordon, I. and Morris, R. (1996), “The equity accounting saga in Australia: cyclical
standard setting,” Abacus, 32, 153–177.
Graham, J. R., Harvey, C. R., and Rajgopal, S. (2005), “The economic implications
of corporate financial reporting,” Journal of Accounting and Economics, 40, 3–73.
Graham, R. and Lefanowicz, C. (1996), “Evidence of the relation between accounting
for equity investments and equity valuation,” Journal of Accounting, Auditing &
Finance, 11, 587–605.
Graham, R. and Lefanowicz, C. (1997), “Parent and subsidiary earnings announcements and parent and subsidiary valuation,” Accounting and Business Research,
28, 3–17.
Graham, R. and Lefanowicz, C. (1999), “Majority and minority ownership of publiclytraded firms: a test of the value of control using market multiples,” Journal of
Business Finance & Accounting, 26, 171–198.
Graham, R., King, R., and Morrill, C. (2003a), “Decision usefulness of alternative
joint venture reporting methods,” Accounting Horizons, 17, 123–137.
Graham, R., Lefanowicz, C., and Petroni, K. (2003b), “The value relevance of equity
method fair value disclosures,” Journal of Business Finance & Accounting, 30,
1065–1088.
Hadani, M., Goranova, M., and Khan, R. (2011), “Institutional investors, shareholder
activism, and earnings management,” Journal of Business Research, 64, 1352–1360.
Haw, I., Hu, B., Hwang, L., and Wu, W. (2004), “Ultimate ownership, income management, and legal and extra-legal institutions,” Journal of Accounting Research,
42, 423–462.
Hirst, D. E. and Hopkins, P. E. (1998), “Comprehensive income reporting and analysts’ valuation judgments,” Journal of Accounting Research, 36, 47–75.
Hirst, D. E., Hopkins, P. E., and Wahlen, J. M. (2004), “Fair values, income measurement, and bank analysts’ risk and valuation judgments,” The Accounting Review,
79, 453–472.
Hope, O. (2013), “Large shareholders and accounting research,” China Journal of
Accounting Research, 6, 3–20.
Hribar, P. and Craig Nichols, D. (2007), “The use of unsigned earnings quality measures in tests of earnings management,” Journal of Accounting Research, 45, 1017–
1053.
90
Jensen, M. and Meckling, W. (1976), “Theory of the firm: Managerial behavior,
agency costs and ownership structure,” Journal of Financial Economics, 3, 305–
360.
Jiambalvo, J., Rajgopal, S., and Venkatachalam, M. (2002), “Institutional ownership and the extent to which stock prices reflect future earnings,” Contemporary
Accounting Research, 19, 117–145.
Jones, J. (1991), “Earnings management during import relief investigations,” Journal
of Accounting Research, pp. 193–228.
Khan, M. and Watts, R. (2009), “Estimation and empirical properties of a firm-year
measure of accounting conservatism,” Journal of Accounting and Economics, 48,
132–150.
Koh, P. (2003), “On the association between institutional ownership and aggressive
corporate earnings management in Australia,” The British Accounting Review, 35,
105–128.
Koh, P. (2007), “Institutional investor type, earnings management and benchmark
beaters,” Journal of Accounting and Public Policy, 26, 267–299.
Kothavala, K. (2003), “Proportional consolidation versus the equity method: A risk
measurement perspective on reporting interests in joint ventures,” Journal of Accounting and Public Policy, 22, 517–538.
La Porta, R., López de Silanes, F., and Shleifer, A. (1999), “Corporate ownership
around the world,” Journal of Finance, 54, 471–517.
Lafond, R. and Roychowdhury, S. (2008), “Managerial ownership and accounting
conservatism,” Journal of Accounting Research, 46, 101–135.
LaFond, R. and Watts, R. (2008), “The information role of conservatism,” The Accounting Review, 83, 447–478.
Lambert, C. and Lambert, C. (2011), “An investigation of earnings management
through investments in associated companies: An Australian perspective,” Journal
of Business & Economics Research, 1.
Lee, S., Pandit, S., and Willis, R. (2013), “Equity method investments and sell-side
analysts’ information environment,” The Accounting Review, forthcoming.
Leisenring, J., Linsmeier, T., Schipper, K., and Trott, E. (2012), “Business-model
(intent)-based accounting,” Accounting and Business Research, 42, 329–344.
Leuz, C., Nanda, D., and Wysocki, P. (2003), “Earnings management and investor
protection: an international comparison,” Journal of Financial Economics, 69,
505–527.
91
Libby, R. and Seybert, N. (2009), “Behavioral studies of the effects of regulation on
earnings management and accounting choice,” in Accounting, Organizations, and
Institutions: Essays for Anthony Hopwood, eds. C. S. Chapman, D. J. Cooper, and
P. B. Miller, Oxford University Press, Oxford.
Lloyd, B. and Weygandt, J. (1971), “Market value information for non-subsidiary
investments,” The Accounting Review, 46, 756–764.
Maines, L. A. and McDaniel, L. S. (2000), “Effects of comprehensive-income characteristics on nonprofessional investors’ judgments: The role of financial-statement
presentation format,” The Accounting Review, 75, 179–207.
Mazay, V., Wilkins, T., and Zimmer, I. (1993), “Determinants of the choice of accounting for investments in associated companies,” Contemporary Accounting Research, 10, 31–59.
Miller, P. and Bahnson, P. (2007), “It’s time to throw the equity method overboard,”
Accounting Today.
Morris, R. and Gordon, I. (2006), “Equity accounting adoption in regulated and
unregulated settings: an empirical study,” Abacus, 42, 22–42.
Nelson, K. K. (1996), “Fair value accounting for commercial banks: an empirical
analysis of SFAS No. 107,” Accounting Review, 71, 161 – 182.
Nobes, C. (2002), “An analysis of the international development of the equity
method,” Abacus, 38, 16–45.
Ohlson, J. (1995), “Earnings, book values, and dividends in equity valuation,” Contemporary Accounting Research, 11, 661–687.
Peasnell, K. V. (1982), “Some formal connections between economic values and yields
and accounting numbers,” Journal of Business Finance & Accounting, 9, 361–381.
Petersen, M. (2009), “Estimating standard errors in finance panel data sets: Comparing approaches,” Review of Financial Studies, 22, 435–480.
Ramalingegowda, S. and Yu, Y. (2011), “Institutional ownership and conservatism,”
Journal of Accounting and Economics, 53, 98–114.
Rees, L. L. and Shane, P. B. (2012), “Academic Research and Standard-Setting: The
Case of Other Comprehensive Income,” Accounting Horizons, 26, 789–815.
Ricks, W. and Hughes, J. (1985), “Market reactions to a non-discretionary accounting
change: the case of long-term investments,” The Accounting Review, 60, 33–52.
Sacristan-Navarro, M. and Gomez-Anson, S. (2007), “Family ownership and pyramids
in the Spanish market,” Family Business Review, 20, 247–265.
92
Shleifer, A. and Vishny, R. (1997), “A survey of corporate governance,” The Journal
of Finance, 52, 737–783.
Song, C. J., Thomas, W. B., and Yi, H. (2010), “Value relevance of FAS No. 157 fair
value hierarchy information and the impact of corporate governance mechanisms,”
Accounting Review, 85, 1375 – 1410.
Stickney, C., Weil, R., Schipper, K., and Francis, J. (2010), “Financial Accounting
(13e),” South-Western Cengage Learning, Mason, OH.
Velury, U. and Jenkins, D. (2006), “Institutional ownership and the quality of earnings,” Journal of Business Research, 59, 1043–1051.
Venkatachalam, M. (1996), “Value-relevance of banks’ derivatives disclosures,” Journal of Accounting and Economics, 22, 327 – 355.
Wang, D. (2006), “Founding family ownership and earnings quality,” Journal of Accounting Research, 44, 619–656.
Watts, R. (2003a), “Conservatism in accounting part I: Explanations and implications,” Accounting Horizons, 17, 207–221.
Watts, R. (2003b), “Conservatism in accounting part II: Evidence and research opportunities,” Accounting Horizons, 17, 287–301.
Yeo, G., Tan, P., Ho, K., and Chen, S. (2002), “Corporate ownership structure and
the informativeness of earnings,” Journal of Business Finance & Accounting, 29,
1023–1046.
Zang, A. Y. (2012), “Evidence on the trade-off between real activities manipulation
and accrual-based earnings management,” The Accounting Review, 87, 675–703.
93
Biography
Amanda Lee (Quiring) Gonzales was born on April 24, 1980, in Aurora, Nebraska. She
earned a Bachelor of Arts degree in Professional Accounting and Mathematics from
Hastings College (Hastings, Nebraska) in 2002, a Master’s degree in Professional Accountancy from the University of Nebraska—Lincoln in 2003, and a Ph.D. in Business
Administration from Duke University in 2013. Prior to attending Duke University,
she worked as a project manager at the International Accounting Standards Board
and as a postgraduate technical assistant at the U.S. Financial Accounting Standards
Board. At Duke University, she received recognition as a James B. Duke Fellow and
University Scholar.
94