employee downsizing and accounting choices: evidence from … · 2017-09-20 · 1 employee...
TRANSCRIPT
DP2017-06
Employee Downsizing and Accounting Choices: Evidence from Japan
Keishi FUJIYAMA Makoto KUROKI
Revised September 20, 2017
1
Employee Downsizing and Accounting Choices: Evidence from Japan
Keishi Fujiyama Kobe University
Makoto Kuroki Yokohama City University [email protected]
2
Employee Downsizing and Accounting Choices: Evidence from Japan
Abstract: While employee downsizing is an important economic phenomenon at least in countries with inflexible employment practices, little is known about accounting practices around downsizing. The literature proposes two competing hypotheses regarding accounting practices around labor negotiations: earnings management hypothesis and informative accounting practice hypothesis. This study investigates accounting choices around employee downsizing, using data from Japan, where employee downsizing is constrained by social norms and case laws. Specifically, it focuses on accounting conservatism, which informs employees of firms’ real economic conditions, and discretionary accruals, which represents opportunistic management behaviors. We find that downsizing firms report more conservative earnings in years during and after downsizing than non-downsizing firms do. However, we find no evidence suggesting that managers engage in income-decreasing earnings management around downsizing. These results are consistent with informative accounting practice hypothesis, in which managers make accounting choices to inform their employees of the firm’s real economic conditions. Keywords: accounting conservatism, earnings management, employee, labor negotiation, downsizing JEL Classifications: G34, J51, M41
INTRODCTION
This study examines accounting choices around employee downsizing. Specifically, it uses
data from Japanese firms whose subsidiaries, if any, operate only in Japan—hereafter,
Japanese domestic firms—and focuses on two earnings attribute measures: discretionary
accruals, which represent an opportunistic accounting practice, and accounting conservatism,
which is assumed to enhance the informativeness of accounting numbers and represents the
3
real economic conditions of a firm.
Employment practices substantially vary around the world. For example, the
“employment at will” norm prevails in the United States, while managers face the “lifetime
employment” norm in Japan. World Economic Forum (2014) reports that hiring/firing
flexibility in the US and Japan is ranked as 11th and 133rd out of 144 countries, respectively.
Moreover, the international variation in labor power affects management decision making on
employment adjustment (Atanassov and Kim 2009). Specifically, they find that firms in
countries with strong labor protection are less likely to downsize their employees. Therefore,
to the extent that financial statements are used during labor negotiations, how accounting
numbers are shaped around employment adjustment, i.e. downsizing, is an important issue
from an international perspective.
In the literature, two competing hypotheses are proposed regarding labor
negotiations, especially wage negotiation. Liberty and Zimmerman (1986) adopt agency
theory to hypothesize that managers manage earnings downward to enhance their bargaining
power when negotiating with employees or labor unions (hereafter, earnings management
hypothesis). Although the earnings management hypothesis is intuitively persuasive, previous
studies show mixed results. Recently, however, García Osma, Mora, and Sabater (2015)
propose an alternative hypothesis that managers make accounting choices to inform
negotiating parties of firms’ real economic conditions (informative accounting practice
4
hypothesis). They show enhanced accounting conservatism during labor negotiations but not
decreased discretionary accruals. These results suggest that negative accruals recognized in
financial statements during employee negotiations are mainly due to economic losses.
Along with this line of research, we investigate the nature of accounting choices
around downsizing by exploiting data from Japanese firms. A Japanese setting provides an
opportunity to investigate accounting practices around downsizing. First, Japanese firms
emphasize relationships with employees. Contrary to the US, where the “employment at will”
norm prevails, stakeholders including managers, shareholders, and employees in Japan share
the “lifetime employment” norm. In addition, firms are required by case laws to demonstrate
the necessity of downsizing. Thus, Japanese firms need to communicate with their
stakeholders and accounting numbers can play a role in doing so and converging stakeholders’
interests. Second, case laws also require that managers talk with labor unions or employee
representatives before implementing downsizings. This results in negotiations with
employees. Third, a certain portion of Japanese firms downsize their employees when they
perform poorly. While the lifetime employment norm prevails in Japan, the two decades of
stagnation in the Japanese economy have caused a deterioration in Japanese firms’ financial
position and downsizing has increased.
The sample for this study consists of 7,217 firm-year observations during 2002-2015,
including 739 observations experiencing a 5 percent or more decrease in employees. The
5
observations are firms that do not have subsidiaries or whose subsidiaries operate only in
Japan. Thus, when managers seek to downsize their employees at any level of a company
group, they face the norms and legal practices of Japanese society.
Consistent with García Osma et al.’s (2015) findings, we find that the degree of
accounting conservatism increases in years during and after downsizing. This result is robust
using several earnings measures. However, we do not find negative discretionary accruals
around downsizing years. Therefore, the results are consistent with the informative
accounting practice hypothesis that managers make accounting choices to inform negotiating
parties of the firm’s real economic conditions. Additionally, we examine whether labor unions
affect accounting practices around downsizing and find that for both the unionized and
non-unionized samples, the degree of accounting conservatism increases in the downsizing
year. Similar to the main results, we do not find a negative relationship between discretionary
accruals and downsizing in the years around downsizing. These results suggest that managers
record economic losses regardless of the existence of labor unions, further supporting the
informative accounting practice hypothesis.
This study contributes to the existing literature in four ways. First, it sheds light on
accounting practices around downsizing. In countries with inflexible employment practices,
personnel reduction impacts significantly on employees’ lives. In addition, managers in such
countries can face stronger resistance from employees. Our results suggest that managers do
6
not deceive employees to obtain better bargaining positions, but inform employees of firms’
real economic conditions to gain concessions and cooperation from employees.
Second, it adds new evidence on accounting practices around labor negotiations.
Prior research proposes two competing hypotheses, the earnings management hypothesis and
informative accounting practice hypothesis. Although García Osma et al. (2015) propose the
informative accounting practice hypothesis and provide consistent evidence, their study is the
first to do so and their sample size is limited to 150 observations including matched
non-negotiation observations. This study complements their findings by focusing on a
different institutional setting, using a larger sample, and investigating a different employee
negotiation. García Osma et al. (2015) acknowledge that relatively small unionized employee
bases of US firms reduce management incentives to depress earnings in a “close-shop”
system, where (re-)negotiation outputs are applied only to unionized workers. In Japan,
managers are often forced to make efforts such as personnel relocation and temporary
assignment before downsizing and to demonstrate the necessity of downsizing, which
suggests that Japanese firms’ financial statements at a firm or business unit level can play a
greater role than those of US counterparts. Nevertheless, we do not find evidence consistent
with the earnings management hypothesis. Therefore, the use of a Japanese setting enhances
the plausibility of the informative accounting practice hypothesis.
Third, this study examines a new aspect of management behavior around downsizing.
7
While prior research emphasizes the importance of trust around downsizing (e.g., Mishra and
Spreitzer 1998; Lee and Teo 2005), little is known about trust destroying behavior in terms of
accounting practices. Hillier, Marshall, McColgan, and Werema (2007) find negative stock
market reaction to layoff announcements and improved employee productivity after
permanent layoffs, which suggests that poor financial conditions drive permanent layoffs and
employee trust is not eroded to a greater extent. Our evidence is consistent with their findings
and suggests that managers do not behave in ways that destroy trust during downsizing.
Finally, this study contributes to prior research on Japanese accounting practices.
While Japan has one of the world’s largest economies and unique institutional characteristics
such as relational banking (main bank system), keiretsu, and stable block holders (e.g.,
Douthett and Jung 2001; Chung, Ho, and Kim 2004; Gramlich, Limpaphayom, and Rhee
2004; Habib, 2004; Baik and Choi, 2010; Shuto and Iwasaki, 2014),1 there is little known
about the context of lifetime employment, especially in the English accounting literature. To
our knowledge, Yamaji (1986) reports the only study examining the influence of Japanese
employees on accounting-related practices. His findings suggest that employee considerations
affect aggressiveness of management forecasts in Japan. This study extends prior research on
Japanese accounting practices by providing new evidence suggesting employee influences on 1 These three characteristics are discussed in management literature (see Yoshikawa and McGuire 2008). Previous studies also investigate accounting practices of Japanese firms and the capital market consequences without analyzing the effects of Japanese institutional characteristics or other factors affecting Japanese accounting practices (Jacobson and Aaker 1993; Inoue and Thomas 1996; Mande, File, and Kwak 2000; Herrmann, Inoue, and Thomas 2003; Skousen, Guan, and Wetzel 2004; Thomas, Herrmann, and Inoue 2004; Shuto 2007; Kato, Skinner, and Kunimura 2009; Shuto 2009; Cho, Hah, and Kim 2011; Higgins 2013; Guo, Huang, Zhang, and Zhou 2015; Goto and Yanase 2016).
8
accounting numbers in financial statements.
The reminder of this paper is as follows. In section 2, we discuss the prior research
and the Japanese institutional setting, and develop our hypothesis. Section 3 describes the
empirical design and sample. We present the results in Section 4. Section 5 presents
conclusions.
BACKGROUND AND HYPOTHESIS DEVELOPMENT
Accounting Practices around Labor Negotiations
While the earnings management hypothesis predicts that managers manage earnings
downward in the face of labor negotiations, previous studies of US firms and wage
negotiations show mixed results. Liberty and Zimmerman (1986) propose the earnings
management hypothesis, but do not provide evidence consistent with it. They interpret their
results in three ways: first, managers do not need to manage earnings downward due to firms’
poor real operating performance; second, labor unions’ ability to undo managers’
opportunistic accounting choices prevents downward earnings management; and third, their
research design is inappropriate. Subsequent studies also do not find evidence consistent with
the earnings management hypothesis (Mautz and Richardson 1992; Cullinan and Knoblett
1994). DeAngelo and DeAngelo (1991) report one exception, showing that during the 1980s,
US companies in the steel industry record a large amount of restructuring charges during
9
negotiations with labor unions. As they argue, managers can choose the timing of the
recognition of those charges and the recognition during labor negotiations may be
opportunistic.2
D’Souza, Jacob, and Ramesh (2001) exploit the adoption year of the Statement of
Financial Accounting Standards No. 106, Employer’s Accounting for Postretirement Benefits
Other Than Pensions, and find that highly unionized firms are more likely to choose an
accounting method that reduces earnings. However, they do not find that firms with higher
costs of debt covenant violations do not make such an accounting choice. Bova (2013) shows
that unionized firms are more likely to miss analysts’ earnings expectation partly by
managing earnings downwards; however, the likelihood of this practice does not increase in
the face of wage negotiations with labor unions. In addition, firms with strong labors increase
information asymmetry (Hilary 2006; Chung, Lee, Lee, and Sohn 2016; Ji and Tan 2016).
These studies suggest that labor considerations affect firms’ accounting practices. The
earnings management hypothesis, however, does not sufficiently predict accounting practices
around labor negotiations.
García Osma et al. (2015) offer several criticisms of the assumptions of
agency-based prior research: (1) negotiation outcomes are mainly determined by the
bargaining power of just one negotiating party (management or labors/labor unions) and (2)
2 Note that DeAngelo and DeAngelo (1991) describe that the losses result from managers’ real operating decisions.
10
the bargaining is a one-shot game. The industrial relations literature views the bargaining
relationship between firms and labors as a repeated game (Espinosa and Rhee 1989; Kahn
1993; Sestini 1999). Given the repeated nature of firm-labor interactions, both parties
consider future interactions and cooperate to achieve efficient outcomes. Therefore, both
parties make concessions when firms perform poorly.
From the above explanation, García Osma et al. (2015) propose an alternative
hypothesis, in which managers make accounting choices around labor negotiations to inform
their employees of the firm’s real economic conditions (informative accounting practice
hypothesis). They find that firms with wage renegotiations report more conservative earnings
than non-negotiating firms do, but fail to find decreased discretionary accruals around labor
negotiations. These results suggest that accrual choices around wage renegotiations reflect
managers’ motivation to reflect economic losses in the firm’s financial statements, which can
promote communication between firms and labors.
Employment Practices in Japan
Japanese employment practices include lifetime employment, the seniority wage system, and
enterprise labor union (Yamaji 1986; Hamaaki, Hori, Maeda, and Murata 2012). Lifetime
employment is a norm at the core of those practices, in which an employee works at a same
company until a certain age stipulated by the company, resulting in a long-term relationship.
11
One of the most important aspects of the lifetime employment norm is that it is an implicit
contract, and trust among the related parties is crucial for the practice to work. This also
significantly constrains Japanese firms’ decision making. For example, combined with banks’
influence,3 Japanese firms attempted to maximize sales growth rather than earnings or
shareholder value until only recently (Ahmadjian and Robbins 2005). This enables stable
employment because sales are the first pie for distribution among stakeholders such as
shareholders, banks, and employees. In another example, Japanese firms are less likely to
restructure their businesses and downsize a smaller proportion of employees than their US
counterparts (Kang and Shivdasani 1997). Therefore, managers of Japanese firms may make
accounting choices that avoid undermining employee trust but promote cooperation.
In addition to the lifetime employment norm, judicially created doctrines set four
guidelines for collective dismissals: (1) necessity, (2) duty to implement dismissal avoidance
efforts, (3) selection adequacy, and (4) procedural adequacy. The first and forth guidelines are
relevant to the present study.4 The first guideline requires managers to demonstrate the
necessity of collective dismissal. For example, two consecutive losses are a traditional
indicator of the necessity of collective dismissal, suggesting that losses are one of the ways
3 Similar to employees, banks face asymmetry payout. In other words, while they receive reduced interest in case of borrowers’ insolvency, they do not receive increased interest in case of borrowers’ high performance. Therefore, banks prefer borrowers’ stable growth of sales to maximization of earnings. See Faleye, Mehrotra, and Morck (2006) for incentives of banks and employees. 4 The second guideline requires managers to make efforts such as personnel relocation and temporary assignment before collective dismissals. The third guideline requires reasonable and fair selection of employees dismissed.
12
managers demonstrate the necessity for collective dismissal.5 If accounting numbers play a
role, they can have certain attributes around collective dismissals. According to the forth
guideline, managers must set conversations with labor unions or employee representatives,
indicating that managers negotiate with them before downsizing. Note that most downsizing
activities are not resorted to courts. Those guidelines, nevertheless, are influential in that
firms try to avoid litigation and behave with reference to them.
In sum, Japanese firms are constrained by a trust-based lifetime employment system,
which leads to a long-term relationship between firms and employees, as well as judicially
created doctrines that suggest a role for accounting numbers and labor negotiations around
downsizing. Therefore, a Japanese setting provides an appropriate opportunity to investigate
accounting practices around employee downsizing.
Hypothesis
Repeated game theory assumes that a firm interacts with employees or labor unions
repeatedly, and employees/labor unions incorporate the firms’ behavior in their response. It
also assumes that cooperation among related parties results in efficient outcomes. Thus,
concessions from related parties are predicted when the firm has poor performance. García
Osma et al. (2015) hypothesize, based on repeated game theory, that managers make
5 Note that doctrines change as social norms change. Two-consecutive-year criterion is changing and firms recently tend to downsize their employees regardless of such losses (Noda 2013).
13
accounting choices to enhance the informativeness of the firm’s financial statements instead
of opportunistic earnings management.
In Japan, the lifetime employment norm results in a long-term relationship, and
active, repeated interactions between firms and employees. It is costly for managers to break
and re-establish trust from employees. For example, Japanese managers are viewed as
intermediaries between shareholders and employees (Aoki 1984) and evaluated by both
capital providers (stable shareholders and banks) and employees (Dore 2000). Moreover, the
lifetime employment norm itself is an implicit contract. Thus, trust from employees is
important for managers.
On the other hand, firms need to gain concessions from employees when their
performance is deteriorating. Since the collapse of the economic babble in 1991, Japanese
firms have experienced a two-decade downturn and deteriorating financial positions, causing
a decline in firms’ capacity to protect employment. Employee downsizing is a type of
concession from employees. Thus, managers persuade employees to accept it. At the same
time, they must take the relationship with employees after downsizing into account because
most employees will remain within the firm. If earnings management becomes manifest
ex-post and erode trust in management, managers will not engage in costly opportunistic
accounting choices. Instead, they will enhance the informativeness of accounting numbers
and show their future outlook.
14
Accounting conservatism occurs when managers “anticipate no profits and provide
for all probable losses” (Bliss 1924, 110); and conditional conservatism is “an equivalent bias
conditional on firms experiencing contemporaneous economic losses” (Ball and Shivakumar
2005, 89). Conditionally conservative accounting, by definition, records losses based on
economic losses, which occur in effect but whose related transactions are not yet realized or
whose related accrual accounting standards do not yet require their recognition. Therefore,
conservative earnings are not derived from non-existent losses and can show firms’ future
outlook (García Osma et al. 2015). The above discussion leads to our hypothesis:
H1: Firms report more conservative earnings around employee downsizing.
RESEARCH DESIGN
Model for Conservatism Test
To test our hypothesis, we run the following regression based on Basu (1997) and with
control variables proposed by Khan and Watts (2009):
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 +𝛼𝛼4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 +𝛼𝛼8𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼9𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼10𝑅𝑅𝑖𝑖,𝑘𝑘𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼11𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑘𝑘−1 +𝛼𝛼12𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼13𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼14𝑅𝑅𝑖𝑖,𝑘𝑘𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼15𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘−1 +𝛼𝛼16𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼17𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼18𝑅𝑅𝑖𝑖,𝑘𝑘𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖,𝑘𝑘−1 + 𝛼𝛼19𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖,𝑘𝑘−1 +𝜀𝜀𝑖𝑖,𝑘𝑘, (1)
15
where OIi,k is ordinary income (keijo rieki), deflated by the market value of equity (MVE) at
three months after the fiscal year end k−1; Ri,k is annual stock return ([MVEi,k −
MVEi,k-1]/MVEi,k-1); NEGi,k takes one if Ri,k is negative and zero otherwise; Sizei,k-1 is the
natural logarithm of MVE at the end of fiscal year k−1; MTBi,k-1 is the market-to-book ratio at
the end of fiscal year k−1, defined as MVE/equity capital (jikoshihon); and Levi,k-1 is the debt
ratio at the end of fiscal year k−1, defined as short-term plus long-term debt scaled by the
MVE at three months after the fiscal year end k−1. k represents fiscal year t−1, t, or t+1.
Downi,t is a proxy for downsizing. It takes one if firm i experiences a five-percent
decrease in employees or more in year t and zero otherwise (Ahmadjian and Robinson 2001;
Ahmadjian and Robbins 2005). We hypothesize that managers enhance the informativeness
of accounting numbers in the face of employee downsizing to elicit concession from
employees, and earnings are more conservative around downsizing. Therefore, we expect a
positive coefficient on α7.
Model for Earnings Management Test
We hypothesize that firms report more conservative earnings around employee downsizing.
While we do not explicitly describe earnings management around downsizing, prior research
proposes the competing earnings management hypothesis. Therefore, in addition to
16
conservative accounting practices, we analyze earnings management around downsizing,
following García Osma et al. (2015).
Discretionary Accrual Measure
We estimate non-discretionary accruals by running the following regression based on
Kasznik’s (1999) model in cross-section for all industry-year combinations (DeFond and
Jiambalvo 1994):6
𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1⁄ = 𝛽𝛽0 + 𝛽𝛽1 1 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1⁄ + 𝛽𝛽2 �∆𝑅𝑅𝑁𝑁𝑅𝑅𝑖𝑖,𝑘𝑘 − ∆𝑅𝑅𝑁𝑁𝐴𝐴𝑖𝑖,𝑘𝑘� 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1�
+𝛽𝛽3 𝑃𝑃𝑃𝑃𝑁𝑁𝑖𝑖,𝑘𝑘 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1⁄ + 𝛽𝛽4 ∆𝐴𝐴𝐶𝐶𝑂𝑂𝑖𝑖,𝑘𝑘 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1⁄ + 𝜀𝜀𝑖𝑖,𝑘𝑘 (2)
where ACCi,k is total accruals in year k, defined as ordinary income after taxes minus cash
flows from operations;7 ΔREVi,k is the change in revenues from year k−1 to k; ΔRECi,k is the
change in accounts receivable from k−1 to k; PPEi,k is gross property, plant, and equipment in
year k; ΔCFOk is the change in cash flows from operations from year k−1 to k; and Assetsi,k-1
is total assets in year k−1.
6 García Osma et al. (2015) examine the relationship of abnormal working capital accruals and cash flows with labor negotiations. We do not employ those measures of opportunistic management behavior because we assume that discretionary total accruals affect ordinary income as a whole and the conservatism measure captures accrual practices. We do not find negative relationships between discretionary accruals and Downi,t from year t−1 to t+1 when we exclude 𝑃𝑃𝑃𝑃𝑁𝑁𝑖𝑖,𝑘𝑘 𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1⁄ from equation (2) based on García Osma et al. (2015). 7 Ordinary income after taxes is the net income after taxes less special profits plus special losses.
17
Empirical Model
To test earnings management around downsizing, we examine the association between
discretionary accruals and downsizing by estimating the following equation similar to Becker,
DeFond, Jiambalvo, and Subramanyam (1998) and García Osma et al. (2015):
𝐷𝐷𝐴𝐴𝑖𝑖,𝑘𝑘 = 𝛾𝛾0 + 𝛾𝛾1𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛾𝛾2𝐿𝐿𝐷𝐷𝐴𝐴𝐴𝐴𝐴𝐴𝑆𝑆𝐴𝐴𝐴𝐴𝑖𝑖,𝑘𝑘−1 + 𝛾𝛾3𝑅𝑅𝑂𝑂𝐴𝐴𝑖𝑖,𝑘𝑘 + 𝛾𝛾4𝐴𝐴𝐶𝐶𝑂𝑂𝑖𝑖,𝑘𝑘 + 𝛾𝛾5𝐷𝐷𝑆𝑆𝐷𝐷𝐴𝐴𝑖𝑖,𝑘𝑘−1 +𝛾𝛾6𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘−1 + 𝛾𝛾7𝑂𝑂𝐴𝐴𝐴𝐴𝐷𝐷𝑁𝑁𝑀𝑀𝑀𝑀𝑖𝑖,𝑘𝑘 + 𝛾𝛾8𝑂𝑂𝐴𝐴𝐴𝐴𝑁𝑁𝐼𝐼𝑖𝑖,𝑘𝑘 + ∑𝐶𝐶𝑆𝑆𝐹𝐹𝑆𝑆𝐹𝐹𝑁𝑁𝐹𝐹𝐹𝐹𝑆𝑆𝐹𝐹𝐴𝐴𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘 (3)
where DAi,k is discretionary accruals computed as total accruals less non-discretionary
accruals estimated from equation (2); LnAssetsi,k-1 is the natural logarithm of total assets in
year k−1; ROAi,k is return on assets in year k−1; Debti,k-1 is the debt ratio; and MTBi,k-1 is
market-to-book ratio at the end of fiscal year. According to García Osma et al. (2015), we
incorporate IssDEBTi,k (IssEQi,k) as a proxy for debt (equity) issuances. IssDEBTi,k (IssEQi,k)
takes one if firm i experiences a 10% increase in outstanding debt (capital stock) or more in
year k and zero otherwise. If managers engage in income-decreasing earnings management
around downsizing, the γ1 will be negative.
Data
Our sample consists of firms operating only in Japan to investigate accounting practices
around downsizing constrained by Japanese employment practices. Our sample period is
18
2002-2015 because Japanese firms disclosed consolidated cash flow statements since 2000.
Table 1 summarizes the sample selection process. First, we use firms with fiscal
years ending in March because approximately 70 percent of Japanese firms use this practice
and this treatment can control for macro-economic effects on financial performance and stock
prices. Second, according to Kimura (2016), we identify Japanese domestic firms by defining
them as firms whose foreign currency translation adjustment account is zero from year t−2 to
t+1, assuming that those firms do not have any subsidiaries or their subsidiaries are located
only in Japan. Thus, the downsizing activities implemented by our sample firms are subject to
Japanese employment practices. Third, we eliminate firms with less than 200 employees. This
process is more likely to prevent us from capturing unintended decreases in employees as
downsizing. Fourth, we eliminate firms with insufficient financial and price data. Fifth, we
exclude financial institutions. Sixth, we exclude firms that downsize their employees in year
t−1 because we are interested in accounting practices in both year t−1 and t, and including
these firms can compound the effects of downsizing in year t on accounting practices in year
t−1. Finally, we include industries comprising 15 firms or more on an annual basis to
compute non-discretionary accruals.
< Insert Table 1 about here >
The result of the above selection process yields a sample of 7,217 firm-year
observations. Table 2 presents the number of firms by industry. The manufacturing industry
19
represents 30.82 percent of the sample. In general, the unionization rate is higher for
manufacturing industries compared to non-manufacturing industries, especially the service
industry (Ministry of Health, Labour and Welfare 2015). Therefore, our sample includes
industries with various unionization rates. A variation in employee influences across firms
enhances the power of empirical test (Bova 2013).
< Insert Table 2 about here >
Table 3 reports the descriptive statistics for the conservatism and earnings
management tests in Panels A and B, respectively. Each continuous variable is winsorized at
its 2nd and 98th percentiles.8 In Panel A, the mean value of Downt is 0.1024 indicating that
on average, 10.24 percent of the sample firms experience downsizing. In Panel B, the median
value of LnAssetst-1 (pre-converting total assets) is 10.260 (JPY 25,323 million9). Our sample
consists of relatively small firms, though does include some large firms. Table 4 presents the
Pearson correlations. Downt is negatively correlated with ROAt, suggesting that downsizing
firms face deteriorating performance.
<Insert Table 3 about here>
RESULTS
8 We set the 2nd and 98th percentiles because the results yield extreme values of annual stock returns in the conservatism test when we winsorize the variables at its 1st and 99th percentiles. The 99th criterion results in an extremely high R-requared for the conservatism test. Ball, Kothari and Robin (2000) employ 2nd and 98th percentile criteria. We obtain similar results of the earnings management tests by winsorizing the variables of the earnings management test at the 1st and 99th percentiles. 9 The amount is US$ 230 million (US$ 1 = JPY 110).
20
Results of Conservatism Test
Table 5 reports the results of the conservatism test. The coefficient of NEGk*Rk*Downt, α7, is
the main interest. The coefficient on year t−1 is insignificant (coefficient = 0.0573; t = 1.32),
while those on years t and t+1 are positive and statistically significant at the 1% and 5%
levels (coefficient = 0.2165 and 0.1234; t = 4.65 and 2.72), respectively. These results suggest
that Japanese domestic firms report more conservative earnings in the years during and after
downsizing, consistent with Hypothesis 1. Interestingly, the coefficients on Rk*Downt in
years t and t+1 are significantly negative (coefficient = -0.0790 and -0.0428; t = -2.76 and
-2.13), indicating that downsizing firms report lower earnings than non-downsizing firms,
even when they earn economic gains. The results are also consistent with the informative
accounting practice hypothesis.
< Insert Table 5 about here >
In Table 5, the coefficient on NEGk*Rk is insignificant. To test the validity of the
empirical model, we run a baseline regression using Basu’s (1997) model. Model (1) in Table
6 presents the baseline results. In Model (1), the coefficient on NEGk*Rk in year t is positive
and statistically significant at the 1% level (coefficient = 0.0645; t = 3.94), confirming the
validity of the models. In Models (2) and (4), the coefficients on NEGk*Rk*Downt remain
statistically significant (coefficient = 0.2519 and 0.2653; t = 5.58 and 5.75), indicating that
the results of the main analysis are robust.
21
< Insert Table 6 about here >
We also confirm the robustness of the main results in two ways. First, we use other
earnings measures: ordinary income plus fixed asset impairment losses10 and net income.
Because Japanese firms disclose financial statements in accordance with Japanese accounting
standards, which require different earnings classifications from US GAAP, it is not possible
to compute the same earnings as employed in US-based studies. The results are qualitatively
the same as in the main analysis except for the coefficient on Rk*Downt in year t
(untabulated), which is insignificant for the model using ordinary income plus fixed asset
impairment losses as the earnings measure.
Second, we employ a continuous downsizing measure, CDownt, that takes a positive
percentage change in employees, multiplied by −1, if the pre-multiplication value is −5
percent or less and zero otherwise. 11 The magnitude of downsizing may affect the
conservativeness of downsizing firms’ earnings. Table 7 presents these results, which are
similar to the main analysis.
< Insert Table 7 about here >
Results of Earnings Management Test
10 Accounting for fixed asset impairment is introduced as mandatory in 2006. Impairment losses are reported as a special loss, below ordinary income. We employ ordinary income in the main analysis to use a uniform earnings measure. 11 The variable is winsorized at the 98nd percentile.
22
Table 8 presents the results of the earnings management tests. The coefficient on Downt in
year t−1 is negative but not statistically significant (coefficient = -0.0000; t = -0.04), while
those in year t and t+1 are positive and statistically significant (coefficient = 0.0025 and
0.0016; t = 2.93 and 1.66), suggesting that managers of downsizing firms seek to record
larger profits in the years during and after downsizing. Those results are inconsistent with
earnings management hypothesis.
< Insert Table 8 about here >
Table 9 shows the results using a continuous downsizing measure, which are similar to those
in Table 8. The coefficients on CDownt in years t and t+1 are significantly positive
(coefficient = 0.0246 and 0.0253; t = 2.56 and 2.41), while that on Downt in year t−1 is
negative but insignificant (coefficient = -0.0056; t = -0.57). Firms with a larger downsizing
may have stronger incentives to report higher earnings.
< Insert Table 9 about here >
Finally, we test the conservatism and earnings management hypotheses by dividing
the sample based on the median value of debt ratio at the end of year t−1. In Table 8, the
positive coefficient on Downt in year t is opposite to the earnings management prediction. It
is possible that downsizing firms experience poor underlying performance and do not need to
depress earnings (Liberty and Zimmerman 1986). Consistently, D’Souza et al. (2001) find
that highly unionized firms with higher costs of debt covenant violations do not make an
23
income-decreasing accounting choice. If downsizing firms face poor performance and
financial position, we expect that income-increasing earnings management is observed for
downsizing firms with a high debt ratio. The results are shown in Table 10. The results in
Panel A indicate that downsizing firms with both high and low debt ratios report conservative
earnings, confirming the informative accounting practice hypothesis. The results in Panel B,
however, reveal that downsizing firms with a high debt ratio engage in income-increasing
earnings management, while this is not the case for those with a low debt ratio. These results
are consistent with Liberty and Zimmerman’s (1986) interpretation of their results and our
expectation.
<Insert Table 10 about here>
Additional Analysis
In this section, we report an exploratory investigation of the effect of labor unions on the
earnings attributes around downsizing. The role of labor unions is to protect employees, and
managers of firms with labor unions may face more difficult negotiations than those of firms
without labor unions. If this is the case, managers of downsizing firms without labor unions
face a relatively lower necessity to communicate with employees through accounting
numbers.
Alternatively, managers of downsizing firms both with and without labor unions
24
have incentives to communicate with employees through accounting numbers. In Japan, the
lifetime employment norm and legal doctrines constrain employment practices. Therefore,
Japanese managers, regardless of labor unions, may follow the doctrines and inform their
employees of the firm’s real economic conditions to avoid needless conflicts with them. If
this is the case, managers of downsizing firms both with and without labor unions report
more conservative earnings than those of non-downsizing firms.
We test the conservativeness of earnings, dividing the full sample into samples with
and without labor unions. A firm is unionized if it has a labor union at the end of fiscal year
t−1.12 The results indicate that both unionized and non-unionized downsizing firms report
more conservative earnings in year t than non-downsizing firms do (Panels A and B of Table
11). The coefficient of NEGk*Rk*Downt in year t−1 is positive and statistically significant at
the 10% level for the unionized sample (coefficient = 0.0996; t = 1.72), while that in year t+1
is positive and significant at the 1% level for the non-unionized sample (coefficient = 0.1764;
t = 2.77). This suggests that labor considerations affect the timing of accounting
communication with employees.
<Insert Table 11 about here>
We also test earnings management around downsizing, incorporating a union dummy,
Uniont-1, and its interaction term with Downt. In Table 12, the coefficients on Downt are
12 We obtain labor union data from Nikkei Kihon data database.
25
insignificant in years t−1 and t+1 (coefficient = 0.0003 and -0.0010; t = 0.19 and -0.63). That
in year t is positive and statistically significant at the 10% level (coefficient = 0.0025; t =
1.70). The results suggest that downsizing firms without labor unions do not behave
opportunistically around employee negotiations from the view point of employees. The
coefficients on Downt*Uniont-1 are insignificant in year t−1 and t (coefficient = -0.0006 and
-0.0000; t = -0.32 and -0.01), while that in year t+1 has the positive sign and is statistically
significant at the 5% level (coefficient = 0.0042; t = 2.27). The combinations of Downt and
Downt*Uniont-1 are positive and statistically significant at the 5% and 1% level in years t and
t+1 (coefficient = 0.0025 and 0.0032; t = 2.44 and 2.89), respectively. The result reveals that
downsizing firms with labor unions have incentives to report higher earnings. In summary,
regardless of the existence of labor unions, we do not find evidence suggesting that
downsizing firms engage in opportunistic income-decreasing earnings management before or
during employee negotiations.
< Insert Table 12 about here >
Note, however, that this study does not investigate the difference in conservative
financial reporting around downsizing between unionized and non-unionized firms. However,
we do show that both types of firms report conservative earnings in the downsizing year.
Nevertheless, the analyses in this section is important because the evidence reveals that
downsizing firms make more or less similar accounting choices that incorporate economic
26
losses regardless of labor unions.
CONCLUSIONS
While employee downsizing is an important economic phenomenon at least in countries with
inflexible employment practices, little is known about accounting practices around
downsizing. Prior accounting research proposes two competing hypotheses of accounting
practices around labor negotiations: the earnings management hypothesis and the informative
accounting practice hypothesis. This study exploits the Japanese setting and investigates
earnings attributes around downsizing. We find that downsizing firms report more
conservative earnings in the downsizing year than non-downsizing firms do. On the other
hand, we find no evidence suggesting that downsizing firms engage in opportunistic
income-decreasing earnings management around downsizing. Instead, the results indicate that
downsizing firms make income-increasing accounting choices more often than
non-downsizing firms do when their debt ratio is high. The results are robust using a
continuous downsizing measure. Overall, the evidence is consistent with the informative
accounting practice hypothesis, similar to García Osma et al. (2015).
In addition, we examine the effect of labor unions on accounting choices around
downsizing. The results indicate that both unionized and non-unionized downsizing firms
report more conservative earnings than their counterparts and do not engage in
27
income-decreasing earnings management around downsizing.
This paper contributes to the literature in four ways. First, it sheds light on a new
labor negotiation i.e. employee downsizing. Second, it complements prior research on
accounting choices around labor negotiations by providing evidence consistent with the
informative accounting practice hypothesis and investigating an institutional setting and
negotiations that differ from those in prior research. Third, this study examines a new aspect
of management behavior around downsizing in terms of destruction of employee trust. Our
results suggest that managers do not make accounting choices that destroy trust. Fourth, this
study highlights accounting practices related to employees, another major stakeholder of
Japanese firms, while most previous studies focus on relational banking (main bank system),
keiretsu, and stable block holders.
A caveat of this study is that it excludes many large Japanese firms because it
focuses on firms that operate only in Japan. The lifetime employment norm is common
among large Japanese firms and extends to smaller firms (Cole 1979; Brown, Nakata, Reich,
and Ullman 1997; Ahmadjian and Robinson 2001). Thus, large firms are constrained to a
greater extent by the social norm. Moreover, labor unions of large firms can be more
sophisticated because their larger number of employees gives them more resources to spend
during negotiations. A study including large Japanese firms may provide clearer insights.
28
29
REFERENCES
Ahmadjian, C. L., and G. E. Robbins. 2005. A clash of capitalisms: Foreign shareholders and corporate
restructuring in 1990s Japan. American Sociological Review 70 (3): 451-471.
Ahmadjian, C. L., and P. Robinson. 2001. Safety in numbers: Downsizing and the deinstitutionalization of
permanent employment in Japan. Administrative Science Quarterly 46 (4): 622-654.
Aoki, M. 1984. The Co-operative Game Theory of the Firm (Oxford, UK: Oxford University Press).
Atanassov, J., and E. Kim. 2009. Labor and corporate governance: International evidence from
restructuring decisions. The Journal of Finance 64 (1): 341-374.
Baik, B., and W. Choi. 2010. Managing earnings surprises in Japan: Perspectives from main bank
relationships and institutional ownership. Journal of Business Finance & Accounting 37 (5‐6):
495-517.
Ball, R., S. P. Kothari, and A. Robin. 2000. The effect of international institutional factors on properties of
accounting earnings. Journal of Accounting and Economics 29 (1): 1-51.
Ball, R., and L. Shivakumar. 2005. Earnings quality in UK private firms: comparative loss recognition
timeliness. Journal of Accounting and Economics 39 (1): 83-128.
Basu, S. 1997. The conservatism principle and the asymmetric timeliness of earnings. Journal of
accounting and economics 24 (1): 3-37.
Becker, C. L., M. L. DeFond, J. Jiambalvo, and K. R. Subramanyam. 1998. The effect of audit quality on
earnings management. Contemporary Accounting Research 15 (1): 1-24.
Bliss, J. H. 1924. Management through Accounts (New York, NY: The Ronald Press Co.).
Bova, F. 2013. Labor unions and management’s incentive to signal a negative outlook. Contemporary
Accounting Research 30 (1): 14-41.
Brown, C., Y. Nakata, M. Reich, and L. Ullman. 1997. Work and Pay in the United States and Japan (New
York, NY: Oxford University Press).
Cho, M., Y. D. Hah, and O. Kim. 2011. Optimistic bias in management forecasts by Japanese firms to
avoid forecasting losses. The International Journal of Accounting 46 (1): 79-101.
Chung, R., S. Ho, and J. B. Kim. 2004. Ownership structure and the pricing of discretionary accruals in
Japan. Journal of International Accounting, Auditing and Taxation 13 (1): 1-20.
Chung, R., B. B. H. Lee, W. J. Lee, and B. C. Sohn. 2016. Do managers withhold good news from labor
unions? Management Science 62 (1): 46-68.
Cole, R. E. 1979. Work, Mobility and Participation: A Comparative Study of American and Japanese
Industry (Berkeley, CA: University of California Press).
Cullinan, C. P., and J. A. Knoblett. 1994. Unionization and accounting policy choices: An empirical
examination. Journal of Accounting and Public Policy 13 (1): 49-78.
DeAngelo, H., and L. DeAngelo. 1991. Union negotiations and corporate policy: A study of labor
concessions in the domestic steel industry during the 1980s. Journal of Financial Economics 30 (1):
3-43.
30
DeFond, M. L., and J. Jiambalvo. 1994. Debt covenant violation and manipulation of accruals. Journal of
Accounting and Economics 17 (1-2): 145-176.
Dore, R. 2000. Stock Market Capitalism: Welfare Capitalism: Japan and Germany Versus the
Anglo-Saxons (Oxford University Press on Demand).
Douthett Jr, E. B., and K. Jung. 2001. Japanese corporate groupings (keiretsu) and the informativeness of
earnings. Journal of International Financial Management & Accounting 12 (2): 133-159.
D'Souza, J., J. Jacob, and K. Ramesh. 2000. The use of accounting flexibility to reduce labor renegotiation
costs and manage earnings. Journal of Accounting and Economics 30 (2): 187-208.
Espinosa, M. P., and C. Rhee. 1989. Efficient Wage Bargaining as a Repeated Game. The Quarterly
Journal of Economics 104 (3): 565-588.
Faleye, O., V. Mehrotra, and R. Morck. 2006. When labor has a voice in corporate governance. Journal of
Financial and Quantitative Analysis 41 (3): 489-510.
García Osma, B. G., A. Mora, and A. M. Sabater. 2015. Strategic Accounting Choice Around Firm-Level
Labor Negotiations. Journal of Accounting Auditing and Finance 30 (2): 246-277.
Goto, S., and N. Yanase. 2016. The Information Content of Corporate Pension Funding Status in Japan.
Journal of Business Finance & Accounting 43 (7-8): 903-949.
Gramlich, J. D., P. Limpaphayom, and S. G. Rhee. 2004. Taxes, keiretsu affiliation, and income shifting.
Journal of Accounting and Economics 37 (2): 203-228.
Guo, J., P. Huang, Y. Zhang, and N. Zhou. 2015. Foreign ownership and real earnings management:
Evidence from Japan. Journal of International Accounting Research 14 (2): 185-213.
Habib, A. 2004. Impact of earnings management on value-relevance of accounting information: Empirical
evidence from Japan. Managerial Finance 30 (11): 1-15.
Hamaaki, J., M. Hori, S. Maeda, and K. Murata. 2012. Changes in the Japanese employment system in the
two lost decades. Industrial & Labor Relations Review 65 (4): 810-846.
Herrmann, D., T. Inoue, and W. B. Thomas. 2003. The sale of assets to manage earnings in Japan. Journal
of Accounting Research 41 (1): 89-108.
Higgins, H. N. 2013. Do stock-for-stock merger acquirers manage earnings? Evidence from Japan. Journal
of Accounting and Public Policy 32 (1): 44-70.
Hilary, G. 2006. Organized labor and information asymmetry in the financial markets. Review of
Accounting Studies 11 (4): 525-548.
Hillier, D., A. Marshall, P. McColgan, and S. Werema. 2007. Employee layoffs, shareholder wealth and
firm performance: Evidence from the UK. Journal of Business Finance & Accounting 34 (3‐4):
467-494.
Inoue, T., and W. B. Thomas. 1996. The choice of accounting policy in Japan. Journal of International
Financial Management & Accounting 7 (1): 1-23.
Jacobson, R., and D. Aaker. 1993. Myopic management behavior with efficient, but imperfect, financial
markets: A comparison of information asymmetries in the US and Japan. Journal of Accounting and
Economics 16 (4): 383-405.
31
Ji, Y., and L. Tan. 2016. Labor Unemployment Concern and Corporate Discretionary Disclosure. Journal
of Business Finance & Accounting 43 (9-10): 1244-1279.
Kahn, L. M. 1993. Managerial quality, team success, and individual player performance in major league
baseball. Industrial & Labor Relations Review 46 (3): 531-547.
Kang, J. K., and A. Shivdasani. 1997. Corporate restructuring during performance declines in Japan.
Journal of Financial Economics 46 (1): 29-65.
Kasznik, R. 1999. On the Association between Voluntary Disclosure and Earnings Management. Journal of
Accounting Research 37 (1): 57-81.
Kato, K., D. J. Skinner, and M. Kunimura. 2009. Management forecasts in Japan: An empirical study of
forecasts that are effectively mandated. The Accounting Review 84 (5): 1575-1606.
Khan, M., and R. L. Watts. 2009. Estimation and empirical properties of a firm-year measure of accounting
conservatism. Journal of Accounting and Economics 48 (2): 132-150.
Kimura, F. 2016. Consolidated Earnings Management in Consolidated Subsidiaries. Discussion Paper No.
120: Tohoku Management & Accounting Research Group, Tohoku University. In Japanese.
Lee, G., and A. Teo. 2005. Organizational restructuring: Impact on trust and work satisfaction. Asia Pacific
Journal of Management 22 (1): 23-39.
Liberty, S. E., and J. L. Zimmerman. 1986. Labor union contract negotiations and accounting choices. The
Accounting Review 61 (4): 692-712.
Mande, V., R. G. File, and W. Kwak. 2000. Income smoothing and discretionary R&D expenditures of
Japanese firms. Contemporary Accounting Research 17 (2): 263-302.
Mautz, R. D., and F. Richardson. 1992. Employer financial information and wage bargaining: Issues and
evidence. Labor Studies Journal 17 (3): 35-52.
Ministry of Health, Labour and Welfare. 2015. Heisei 27 nen roudoukumiai kiso chosa no gaikyo (The
briefing of substructure investigation of labor unions in 2015). In Japanese.
Mishra, A. K., and G. M. Spreitzer. 1998. Explaining how survivors respond to downsizing: The roles of
trust, empowerment, justice, and work redesign. Academy of Management Review 23 (3): 567-588.
Noda, T. 2013. Determinants of the Timing of Downsizing Among Large Japanese Firms: Long‐Term
Employment Practices and Corporate Governance. Japanese Economic Review 64 (3): 363-398.
Sestini, R. 1999. Union–Firm Bargaining as a Repeated Game and the Behaviour of Wages over the
Business Cycle. Labour 13 (4): 821-857.
Shuto, A. 2007. Executive compensation and earnings management: Empirical evidence from Japan.
Journal of International Accounting, Auditing and Taxation 16 (1): 1-26.
Shuto, A. 2009. Earnings Management to Exceed the Threshold: A Comparative Analysis of Consolidated
and Parent‐only Earnings. Journal of International Financial Management & Accounting 20 (3):
199-239.
Shuto, A., and T. Iwasaki. 2014. Stable shareholdings, the decision horizon problem and earnings
smoothing. Journal of Business Finance & Accounting 41 (9-10): 1212-1242.
Skousen, C. J., L. Guan, and T. S. Wetzel. 2004. Anomalies and unusual patterns in reported earnings:
32
Japanese managers round earnings. Journal of International Financial Management & Accounting
15 (3): 212-234.
Thomas, W. B., D. R. Herrmann, and T. Inoue. 2004. Earnings management through affiliated transactions.
Journal of International Accounting Research 3 (2): 1-25.
World Economic Forum. 2014. The Global Competitiveness Report 2014-2015.
Yamaji, H. 1986. Collective bargaining and accounting disclosure: An inquiry into the changes in
accounting policy. International Journal of Accounting Education and Research 22 (1): 11-23.
Yoshikawa, T., and J. McGuire. 2008. Change and continuity in Japanese corporate governance. Asia
Pacific Journal of Management 25 (1): 5-24.
33
Table 1. Sample Selection Process 1. Fiscal year end is March, with 12 months in the fiscal year; 2. Foreign currency translation adjustment is zero from year t-2 to t+1; 3. The number of employees in year t-1 is 200 or more; 4. Necessary financial and price data are available; 5. Firms in the financial industry are excluded; 6. Firms in industries with less than 15 peers on an annual basis are excluded; 7. Firms implementing a downsizing in year t-1 are excluded.
34
Table 2. Number of Observations by Industry # of Obs. Proportion Cumulative
Food 406 5.63% 5.63% Chemicals 258 3.57% 9.20% Pharmaceuticals 82 1.14% 10.34% Ceramics 65 0.90% 11.24% Steel 79 1.09% 12.33% Non-ferrous metal 338 4.68% 17.02% Machinery 459 6.36% 23.38% Electronics 264 3.66% 27.03% Other manufacturing 273 3.78% 30.82% Construction 924 12.80% 43.62% Trading 1072 14.85% 58.47% Retail 680 9.42% 67.90% Real estate 31 0.43% 68.32% Rail road and bus 261 3.62% 71.94% Land transportation 184 2.55% 74.49% Services 1,841 25.51% 100.00%
35
Table 3. Descriptive Statistics (n = 7,217) Panel A: Variables for the Conservatism Test in year t Mean SD Min 25 %ile Median 75 %ile Max
OIt 0.1274 0.1140 -0.3581 0.0647 0.1175 0.1834 0.7511
NEGt 0.4959 0.5000 0.0000 0.0000 0.0000 1.0000 1.0000
Rt 0.0594 0.3810 -0.6487 -0.1533 0.0029 0.1800 3.3169
Downt 0.1024 0.3032 0.0000 0.0000 0.0000 0.0000 1.0000
Sizet-1 9.2356 1.2623 6.5212 8.3517 9.0619 9.9755 13.0341
MTBt-1 1.0680 0.8382 0.2016 0.5386 0.7992 1.2731 5.8207
Levt-1 0.7948 1.0976 0.0000 0.0410 0.3622 1.0914 7.0912
Panel B: Variables for the Earnings Management Test in Year t Mean SD Min 25 %ile Median 75 %ile Max
DAt 0.0001 0.0269 -0.0737 -0.0163 0.0004 0.0167 0.0755
LnAssetst-1 10.2604 1.1015 7.8694 9.4688 10.1395 10.9058 13.5125
ROAt 0.0304 0.0320 -0.0713 0.0123 0.0259 0.0458 0.1411
CFOk 0.0533 0.0547 -0.1100 0.0213 0.0512 0.0836 0.2354
Debtt-1 0.1834 0.1729 0.0000 0.0220 0.1445 0.2990 0.6588
MTBt-1 1.0555 0.8502 0.1949 0.5268 0.7864 1.2553 6.7251
IssEQt 0.3410 0.4741 0.0000 0.0000 0.0000 1.0000 1.0000
IssDEt 0.0317 0.1753 0.0000 0.0000 0.0000 0.0000 1.0000
Note: Panels A and B report descriptive statistics of variables for the conservatism and earnings
management tests, respectively. OI is ordinary income, scaled by lagged market capitalization. NEG takes
one if a firm has a negative annual stock return and zero otherwise. R is the annual stock return. Down is a
dummy variable that takes one if a firm experiences a 5% decrease in employees or more and zero
otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book ratio. Lev is
total debt, scaled by market capitalization. DA is discretionary accruals. LnAssets is the natural logarithm
of total assets. ROA is the return on assets. CFO is cash flows, scaled by lagged assets. Debt is the debt
ratio. IssEQ (IssDE) takes one if the positive change in capital stock (total debt) is more than 10% and
zero otherwise.
36
Table 4. Peason Correlation for the Earnings Management Test in Year t (n = 7,217) (1) (2) (3) (4) (5) (6) (7) (8) (1) DAt 1.0000 (2) Downt 0.0047 1.0000 (3) LnAssetsk-1 0.0085 -0.0459 1.0000 (4) ROAk 0.1681 -0.2109 -0.0581 1.0000 (5) CFOk -0.4176 -0.1290 -0.0704 0.5359 1.0000 (6) Debtk-1 -0.0280 0.1297 0.1158 -0.2594 -0.0359 1.0000 (7) MTBk-1 -0.0148 -0.0360 0.0238 0.2978 0.1927 0.1992 1.0000 (8) IssDEk-1 0.1231 -0.0636 -0.1118 0.1263 -0.0965 -0.3472 0.0323 1.0000 (9) IssEQt-1 0.0167 -0.0090 0.0127 0.0764 0.0297 0.0893 0.1782 -0.0085
Note: This table reports Pearson correlation coefficients of the variables for the earnings management test
in year t. DA is discretionary accruals. Down is a dummy variable that takes one if a firm experience a 5%
decrease in employees or more, zero otherwise. LnAssets is the natural logarithm of total assets. ROA is
the return on assets. CFO is cash flows, scaled by lagged assets. MTB is the market-to-book ratio. Debt is
the debt ratio. IssEQ (IssDE) takes one if the positive change in capital stock (total debt) is more than 10%,
zero otherwise. Bold represents statistical significance at the 5% level.
37
Table 5. Conservatism Test Results Year t-1 Year t Year t+1
Coef. t-value Coef. t-value Coef. t-value Intercept 0.2190 *** 8.24 0.1731 *** 6.43 0.1763 *** 6.70 NEGk -0.0458 -1.79 -0.0443 -1.60 -0.0741 ** -2.58 NEGk * Rk 0.0997 1.02 0.0979 0.88 0.1306 1.08 Rk 0.1338 * 2.84 0.2134 *** 4.67 0.1165 ** 2.45
Downt -0.0598 *** -6.15 -0.0298 *** -2.93 -0.0370 *** -4.29 NEGk * Downt 0.0133 1.06 0.0004 0.03 0.0187 1.52 Rk * Downt -0.0247 -0.83 -0.0790 *** -2.76 -0.0428 ** -2.13 NEGk * Rk * Downt 0.0573 1.32 0.2165 *** 4.65 0.1234 *** 2.72
Sizek-1 -0.0093 *** -3.36 -0.0048 * -1.74 -0.0060 ** -2.19 NEGk * Sizek-1 0.0031 1.08 0.0022 0.73 0.0055 * 1.77 Rk * Sizek-1 -0.0025 -0.48 -0.0104 ** -2.11 0.0005 0.11 NEGk * Rk * Sizek-1 -0.0105 -1.05 -0.0086 -0.76 -0.0163 -1.31 MTBk-1 -0.0146 *** -4.80 -0.0185 *** -5.04 -0.0169 *** -4.31 NEGk * MTBk-1 0.0063 * 1.66 0.0106 ** 2.37 0.0035 0.81 Rk * MTBk-1 -0.0251 *** -4.46 -0.0263 *** -3.74 -0.0336 *** -5.31 NEGk * Rk * MTBk-1 0.0120 1.16 0.0092 0.75 0.0061 0.43 Levk-1 0.0094 *** 2.77 0.0069 * 1.87 0.0027 0.75 NEGk * Levk-1 -0.0050 -1.23 -0.0022 -0.48 0.0047 1.06 Rk * Levk-1 -0.0018 -0.35 0.0017 0.40 0.0061 1.60 NEGk * Rk * Levk-1 0.0180 1.04 0.0353 * 1.94 0.0510 *** 3.11
Year Yes Yes Yes Industry Yes Yes Yes
# of Obs. 7,217 7,217 7,217 adjR2 0.2999 0.3139 0.2628
Note: This table reports estimation results of the conservatism tests.
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡
+𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘
The dependent variable, OI, is ordinary income, scaled by lagged market capitalization. NEG takes one if
a firm has a negative annual stock return and zero otherwise. R is the annual stock return. Down is a
dummy variable that takes one if a firm experiences a 5% decrease in employees or more and zero
otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book ratio. Lev is
total debt, scaled by market capitalization. ***, **, and * represent significance at the 1%, 5%, and 10%
levels (two-tailed), respectively. Standard errors are clustered at the firm level.
38
Table 6. Model Validity Test Using Data in Year t Model (1) Model (2) Model (3) Model (4)
Coef. t-value Coef. t-value Coef. t-value Coef. t-value
Intercept 0.1025 *** 10.73 0.1092 *** 11.68 0.1382 *** 6.39 0.1554 *** 7.44
NEGk -0.0120 *** -3.41 -0.0137 *** -3.88 -0.0126 *** -3.59 -0.0148 *** -4.21
NEGk * Rk 0.0645 *** 3.94 0.0206 1.31 0.0461 *** 2.72 -0.0007 -0.05
Rk 0.1069 *** 13.85 0.1105 *** 14.21 0.1049 *** 14.13 0.1070 *** 14.57
Downt -0.0169 * -1.68 -0.0232 ** -2.35 NEGk * Downt -0.0097 -0.75 -0.0051 -0.39 Rk * Downt -0.0798 *** -2.97 -0.0781 *** -2.95 NEGk * Rk * Downt 0.2519 *** 5.58 0.2653 *** 5.75
Sizek-1 -0.0029 -1.39 -0.0042 ** -2.07 MTBk-1 -0.0113 *** -4.73 -0.0117 *** -5.06 Levk-1 0.0015 0.54 0.0036 1.35
Year Yes Yes Yes Yes Industry Yes Yes Yes Yes
# of Obs. 7,217 7,217 7,217 7,217 adjR2 0.2515 0.276 0.2594 0.2876
Note: This table reports estimation results of model validity tests of the main conservatism tests, using data
in year t.
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡
+𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑘𝑘
The dependent variable, OI, is ordinary income, scaled by lagged market capitalization. NEG takes one if
a firm has a negative annual stock return and zero otherwise. R is the annual stock return. Down is a
dummy variable that takes one if a firm experiences a 5% decrease in employees or more and zero
otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book ratio. Lev is
total debt, scaled by market capitalization. ***, **, and * represent significance at the 1%, 5%, and 10%
levels (two-tailed), respectively. Standard errors are clustered at the firm level.
39
Table 7. Conservatism Test Results Using a Continuous Downsizing Measure Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept 0.2188 *** 8.24 0.1732 *** 6.44 0.1760 *** 6.68 NEGk -0.0461 * -1.81 -0.0434 -1.57 -0.0754 *** -2.63 NEGk * Rk 0.0901 0.92 0.1112 0.99 0.1305 1.08 Rk 0.1349 *** 2.87 0.2112 *** 4.59 0.1140 ** 2.40
CDownt -0.7395 *** -6.74 -0.3584 *** -2.81 -0.3563 *** -3.46 NEGk * CDownt 0.1798 1.28 -0.0583 -0.37 0.1894 1.34 Rk * CDownt -0.0484 -0.17 -0.8147 ** -2.42 -0.4192 * -1.95 NEGk * Rk * CDownt 0.0770 0.18 1.5752 *** 3.24 1.3565 *** 3.05
Sizek-1 -0.0093 *** -3.36 -0.0048 * -1.74 -0.0061 ** -2.21 NEGk * Sizek-1 0.0032 1.12 0.0021 0.71 0.0057 * 1.83 Rk * Sizek-1 -0.0024 -0.46 -0.0103 ** -2.09 0.0007 0.15 NEGk * Rk * Sizek-1 -0.0093 -0.93 -0.0091 -0.80 -0.0164 -1.32 MTBk-1 -0.0142 *** -4.69 -0.0183 *** -4.99 -0.0163 *** -4.13 NEGk * MTBk-1 0.0062 1.64 0.0104 ** 2.34 0.0030 0.69 Rk * MTBk-1 -0.0258 *** -4.53 -0.0252 *** -3.56 -0.0334 *** -5.25 NEGk * Rk * MTBk-1 0.0124 1.21 0.0075 0.60 0.0062 0.43 Levk-1 0.0102 *** 3.00 0.0071 * 1.95 0.0026 0.73 NEGk * Levk-1 -0.0058 -1.43 -0.0018 -0.39 0.0047 1.07 Rk * Levk-1 -0.0035 -0.67 0.0022 0.52 0.0064 * 1.66 NEGk * Rk * Levk-1 0.0196 1.15 0.0379 ** 2.10 0.0502 *** 3.07
Year Yes Yes Yes Industry Yes Yes Yes
# of Obs. 7,217 7,217 7,217 adjR2 0.3015 0.3138 0.2618
Note: This table reports estimation results of the conservatism tests, using a continuous downsizing
measure.
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼4𝐴𝐴𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐴𝐴𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡
+𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐴𝐴𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐴𝐴𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘
The dependent variable, OI, is ordinary income, scaled by lagged market capitalization. NEG takes one if
a firm has a negative annual stock return and zero otherwise. R is the annual stock return. CDown takes a
positive percentage change in employees, multiplied by −1, if the pre-multiplication value is −5 percent or
less and zero otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book
ratio. Lev is total debt, scaled by market capitalization. ***, **, and * represent significance at the 1%, 5%,
and 10% levels (two-tailed), respectively. Standard errors are clustered at the firm level.
40
Table 8. Earnings Management Test Results
Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept -0.0065 * -1.65 0.0031 0.82 0.0040 1.05 Downt -0.0000 -0.04 0.0025 *** 2.93 0.0016 * 1.66 LnAssetsk-1 0.0007 ** 1.99 0.0001 0.41 0.0002 0.60 ROAk 0.5481 *** 36.74 0.5401 *** 37.32 0.5164 *** 36.83 CFOk -0.3944 *** -51.02 -0.3815 *** -47.17 -0.3820 *** -46.52 Debtk-1 0.0176 *** 7.60 0.0186 *** 7.92 0.0179 *** 7.68 MTBk-1 -0.0024 *** -5.50 -0.0026 *** -5.34 -0.0026 *** -5.16 IssDEk 0.0015 ** 2.40 0.0009 1.42 0.0014 ** 2.13 IssEQk 0.0015 1.02 0.0007 0.40 0.0000 -0.03
Industry Yes Yes Yes Year Yes Yes Yes
# of Obs. 7,217 7,217 7,217 adjR2 0.4677 0.4559 0.4494
Note: This table reports estimation results of the earnings management tests.
𝐷𝐷𝐴𝐴𝑖𝑖,𝑘𝑘 = 𝛾𝛾0 + 𝛾𝛾1𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘 The dependent variable, DA, is discretionary accruals. Down is a dummy variable that takes one if a firm
experience a 5% decrease in employees or more, zero otherwise. LnAssets is the natural logarithm of total
assets. ROA is the return on assets. CFO is cash flows, scaled by lagged assets. Debt is the debt ratio.
MTB is the market-to-book ratio. IssEQ (IssDE) takes one if the positive change in capital stock (total
debt) is more than 10% and zero otherwise. ***, **, and * represent significance at the 1%, 5%, and 10%
levels (two-tailed), respectively. Standard errors are clustered at the firm level.
41
Table 9. Earnings Management Test Results Using a Continuous Downsizing Measure
Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept -0.0064 -1.61 0.0031 0.84 0.0036 0.96 CDownt -0.0056 -0.57 0.0246 ** 2.56 0.0253 ** 2.41 LnAssetsk-1 0.0007 ** 1.96 0.0001 0.41 0.0002 0.67 ROAk 0.5472 *** 36.62 0.5397 *** 37.24 0.5171 *** 36.94 CFOk -0.3945 *** -50.97 -0.3814 *** -47.15 -0.3820 *** -46.45 Debtk-1 0.0176 *** 7.63 0.0185 *** 7.91 0.0177 *** 7.62 MTBk-1 -0.0024 *** -5.48 -0.0026 *** -5.35 -0.0026 *** -5.2 IssDEk 0.0015 ** 2.41 0.0009 1.40 0.0014 ** 2.16 IssEQk 0.0015 1.03 0.0006 0.38 -0.0001 -0.03
Industry Yes Yes Yes Year Yes Yes Yes
# of Obs. 7,217 7,217 7,217 adjR2 0.4414 0.4292 0.4259
Note: This table reports estimation results of the earnings management tests using an alternative
downsizing variable.
𝐷𝐷𝐴𝐴𝑖𝑖,𝑘𝑘 = 𝛾𝛾0 + 𝛾𝛾1𝐴𝐴𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘 The dependent variable, DA, is discretionary accruals. CDown takes a positive percentage change in
employees, multiplied by −1, if the pre-multiplication value is −5 percent or less and zero otherwise.
LnAssets is the natural logarithm of total assets. ROA is the return on assets. CFO is cash flows, scaled by
lagged assets. Debt is the debt ratio. MTB is the market-to-book ratio. IssEQ (IssDE) takes one if the
positive change in capital stock (total debt) is more than 10% and zero otherwise. *** and ** represent
significance at the 1% and 5% levels (two-tailed), respectively. Standard errors are clustered at the firm
level.
42
Table 10: Results of conservatism and earnings management tests dividing the sample based on the median value of debt ratio at the end of year t−1 Panel A: Conservatism test in year t Debtt-1 >= median Debtt-1 < median Coef. t-value Coef. t-value Intercept 0.1347 *** 3.50 0.2037 *** 5.84 NEGt -0.0607 -1.56 -0.0474 -1.20 NEGt * Rt 0.1545 1.01 -0.0303 -0.20 Rt 0.1708 ** 2.41 0.2204 *** 4.29
Downt -0.0446 *** -3.22 -0.0154 -1.23 NEGt * Downt 0.0129 0.72 -0.0086 -0.50 Rt * Downt -0.0213 -0.57 -0.1322 *** -3.55 NEGt * Rt * Downt 0.1541 *** 2.63 0.2717 *** 4.12
Sizek-1 0.0001 0.03 -0.0097 *** -2.76 NEGk * Sizek-1 0.0011 0.27 0.0046 1.10 Rk * Sizek-1 -0.0090 -1.22 -0.0089 -1.52 NEGk * Rk * Sizek-1 -0.0170 -1.10 0.0056 0.36 MTBk-1 -0.0234 *** -5.01 -0.0078 -1.37 NEGk * MTBk-1 0.0196 *** 3.08 0.0016 0.28 Rk * MTBk-1 -0.0139 -1.64 -0.0389 *** -3.11 NEGk * Rk * MTBk-1 0.0148 0.86 0.0106 0.58 Levk-1 0.0069 1.41 0.0558 *** 3.29 NEGk * Levk-1 0.0047 0.78 -0.0481 * -1.91 Rk * Levk-1 0.0058 0.80 -0.0127 ** -2.00 NEGk * Rk * Levk-1 0.0400 * 1.69 0.0440 0.36
Year Yes Yes Industry Yes Yes
# of Obs. 3,609 3,608 adjR2 0.3088 0.3402
43
Panel B: Earnings management test in year t
Debtt-1 >= median Debtt-1 < median
Coef. t-value Coef. t-value
Intercept 0.0034 0.64 -0.0001 -0.01
Downt 0.0024 ** 2.28 0.0023 1.61 LnAssetsk-1 0.0003 0.66 0.0000 0.07 ROAk 0.5473 *** 25.40 0.5285 *** 28.75 CFOk -0.3924 *** -33.67 -0.3708 *** -35.64 Debtk-1 0.0216 *** 5.45 0.0223 ** 2.17 MTBk-1 -0.0033 *** -5.19 -0.0017 ** -2.20 IssDEk-1 0.0002 0.23 0.0014 * 1.64 IssEQk-1 0.0029 1.46 -0.0030 -1.04
Industry Yes Yes Year Yes Yes
# of Obs. 3,609 3,608 adjR2 0.439 0.4212
Note: Panels A and B report estimation results of the conservatism and earnings management tests, using
samples with high and low debt ratio firms, respectively.
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡
+𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘
𝐷𝐷𝐴𝐴𝑖𝑖,𝑘𝑘 = 𝛾𝛾0 + 𝛾𝛾1𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘 The dependent variable, OI, is ordinary income, scaled by lagged market capitalization. NEG takes one if
a firm has a negative annual stock return and zero otherwise. R is the annual stock return. Down is a
dummy variable that takes one if a firm experiences a 5% decrease in employees or more and zero
otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book ratio. Lev is
total debt, scaled by market capitalization. The dependent variable, DA, is discretionary accruals. LnAssets
is the natural logarithm of total assets. ROA is the return on assets. CFO is cash flows, scaled by lagged
assets. Debt is the debt ratio. IssEQ (IssDE) takes one if the positive change in capital stock (total debt) is
more than 10% and zero otherwise. ***, **, and * represent significance at the 1%, 5%, and 10% levels
(two-tailed), respectively. Standard errors are clustered at the firm level.
44
Table 11: Effect of Labor Unions on Conservatism around Downsizing Panel A: Unionized Sample Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept 0.2092 *** 5.97 0.1785 *** 4.64 0.1730 *** 4.79 NEGk -0.0515 -1.45 -0.0592 -1.48 -0.0684 * -1.75 NEGk * Rk 0.0479 0.31 0.1187 0.69 0.1476 0.92 Rk 0.1314 ** 2.20 0.1727 ** 2.20 0.0959 1.22
Downt -0.0563 *** -5.29 -0.0226 * -1.78 -0.0267 ** -2.33 NEGk * Downt 0.0114 0.74 0.0040 0.25 -0.0014 -0.09 Rk * Downt -0.0390 -1.29 -0.0866 ** -2.29 -0.0400 -1.23 NEGk * Rk * Downt 0.0996 * 1.72 0.3153 *** 4.51 0.0684 1.03
Sizek-1 -0.0102 *** -2.81 -0.0076 * -1.92 -0.0085 ** -2.24 NEGk * Sizek-1 0.0048 1.25 0.0055 1.26 0.0057 1.35 Rk * Sizek-1 -0.0017 -0.25 -0.0050 -0.62 0.0020 0.24 NEGk * Rk * Sizek-1 -0.0020 -0.13 -0.0058 -0.33 -0.0150 -0.92 MTBk-1 -0.0186 *** -4.65 -0.0181 *** -3.38 -0.0167 *** -2.96 NEGk * MTBk-1 0.0057 1.02 0.0030 0.43 0.0035 0.55 Rk * MTBk-1 -0.0171 ** -2.16 -0.0297 ** -2.47 -0.0268 ** -2.57 NEGk * Rk * MTBk-1 -0.0164 -0.87 -0.0211 -0.86 -0.0044 -0.19 Levk-1 0.0111 ** 2.46 0.0053 1.02 -0.0004 -0.07 NEGk * Levk-1 -0.0123 ** -2.40 -0.0060 -0.95 0.0073 1.14 Rk * Levk-1 -0.0133 * -1.90 0.0040 0.59 0.0105 * 1.68 NEGk * Rk * Levk-1 -0.0054 -0.21 -0.0037 -0.16 0.0383 1.48
Year Yes Yes Yes Industry Yes Yes Yes
# of Obs. 3,906 3,906 3,906 adjR2 0.264 0.283 0.2345
45
Panel B: Non-unionized Sample Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept 0.2174 *** 6.00 0.1501 *** 4.09 0.1638 *** 4.45 NEGk -0.0268 -0.72 -0.0093 -0.24 -0.0629 -1.39 NEGk * Rk 0.0929 0.74 0.0276 0.19 0.0920 0.47 Rk 0.1548 ** 2.48 0.2813 ** 4.84 0.1574 ** 2.54
Downt -0.0562 *** -3.45 -0.0380 ** -2.43 -0.0469 *** -3.47 NEGk * Downt 0.0151 0.73 0.0046 0.22 0.0414 ** 2.23 Rk * Downt -0.0174 -0.38 -0.0669 -1.60 -0.0457 * -1.73 NEGk * Rk * Downt 0.0370 0.59 0.1467 ** 2.45 0.1764 *** 2.77
Sizek-1 -0.0056 -1.44 0.0014 0.37 0.0000 0.01 NEGk * Sizek-1 0.0000 -0.01 -0.0034 -0.78 0.0032 0.66 Rk * Sizek-1 -0.0050 -0.73 -0.0194 *** -2.99 -0.0040 -0.59 NEGk * Rk * Sizek-1 -0.0118 -0.89 -0.0044 -0.28 -0.0147 -0.70 MTBk-1 -0.0134 *** -3.32 -0.0216 *** -4.26 -0.0211 *** -3.86 NEGk * MTBk-1 0.0069 1.44 0.0177 *** 2.97 0.0063 1.07 Rk * MTBk-1 -0.0296 *** -4.59 -0.0203 ** -2.28 -0.0331 *** -4.02 NEGk * Rk * MTBk-1 0.0264 ** 2.18 0.0239 1.59 0.0123 0.61 Levk-1 0.0138 *** 3.08 0.0119 ** 2.56 0.0083 * 1.86 NEGk * Levk-1 -0.0008 -0.13 -0.0001 -0.02 -0.0013 -0.24 Rk * Levk-1 0.0010 0.16 -0.0020 -0.38 0.0013 0.27 NEGk * Rk * Levk-1 0.0515 ** 2.21 0.0748 *** 3.14 0.0606 *** 2.91
Year Yes Yes Yes Industry Yes Yes Yes
# of Obs. 3,311 3,311 3,311 adjR2 0.3534 0.3579 0.3029
Note: Panels A and B report estimation results of the conservatism tests, using samples with unionized and
non-unionized firms, respectively.
𝑂𝑂𝑂𝑂𝑖𝑖,𝑘𝑘 = 𝛼𝛼0 + 𝛼𝛼1𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘 + 𝛼𝛼2𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼3𝑅𝑅𝑖𝑖,𝑘𝑘 + 𝛼𝛼4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼5𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡
+𝛼𝛼6𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛼𝛼7𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑘𝑘𝑅𝑅𝑖𝑖,𝑘𝑘𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝐴𝐴𝐷𝐷𝐷𝐷𝐴𝐴𝐶𝐶𝐷𝐷𝐶𝐶𝐴𝐴 + 𝜀𝜀𝑖𝑖,𝑘𝑘
The dependent variable, OI, is ordinary income, scaled by lagged market capitalization. NEG takes one if
a firm has a negative annual stock return and zero otherwise. R is the annual stock return. Down is a
dummy variable that takes one if a firm experiences a 5% decrease in employees or more and zero
otherwise. Size is the natural logarithm of market capitalization. MTB is the market-to-book ratio. Lev is
total debt, scaled by market capitalization. ***, **, and * represent significance at the 1%, 5%, and 10%
levels (two-tailed), respectively. Standard errors are clustered at the firm level.
46
Table 12: Effect of Labor Unions on Earnings Management around Downsizing Year t-1 Year t Year t+1 Coef. t-value Coef. t-value Coef. t-value Intercept -0.0064 -1.61 0.0033 0.89 0.0044 1.18 Downt 0.0003 0.19 0.0025 * 1.70 -0.0010 -0.63 Downt*Uniont-1 -0.0006 -0.32 0.0000 -0.01 0.0042 ** 2.27 Uniont-1 0.0023 *** 3.10 0.0021 *** 2.87 0.0025 *** 3.61 LnAssetsk-1 0.0005 1.45 0.0000 -0.14 -0.0001 -0.16 ROAk 0.5523 *** 37.28 0.5438 *** 37.68 0.5208 *** 37.19 CFOk -0.3944 *** -50.84 -0.3814 *** -46.99 -0.3814 *** -46.44 Debtk-1 0.0177 *** 7.65 0.0188 *** 7.96 0.0183 *** 7.83 MTBk-1 -0.0023 *** -5.45 -0.0026 *** -5.27 -0.0026 *** -5.16 IssDEk 0.0017 *** 2.68 0.0011 * 1.68 0.0016 ** 2.49 IssEQk 0.0015 1.04 0.0008 0.45 0.0000 0.01
Year Yes Yes Yes Industry Yes Yes Yes
# of Obs. 7,217 7,217 7,217 adjR2 0.4425 0.4304 0.4283
Downt + Downt*Uniont-1 -0.0003 -0.29 0.0025 ** 2.44 0.0032 *** 2.89
Note: This table reports estimation results of the earnings management tests, incorporating a union dummy
and its interaction with Down. The dependent variable, DA, is discretionary accruals. Down is a dummy
variable that takes one if a firm experience a 5% decrease in employees or more, zero otherwise. Union is
a dummy variable that takes one if a firm has a labor union at the end of fiscal year t−1 and zero otherwise.
LnAssets is the natural logarithm of total assets. ROA is the return on assets. CFO is cash flows, scaled by
lagged assets. Debt is the debt ratio. MTB is the market-to-book ratio. IssEQ (IssDE) takes one if the
positive change in capital stock (total debt) is more than 10% and zero otherwise. ***, **, and * represent
significance at the 1%, 5%, and 10% levels (two-tailed), respectively. Standard errors are clustered at the
firm level.