1 The Impact of Labor Unions on External Auditor Selection and Audit Scope Ju Ryum Chung School of Business Yonsei University Eun Jung Cho School of Business Yonsei University Ho-Young Lee* School of Business Yonsei University [email protected]+82-2-2123-5484 And Myungsoo Son College of Business and Economics California State University, Fullerton . March 21, 2015 * Please direct all correspondence to Ho-Young Lee, School of Business, Yonsei University, Seoul, S. Korea. TEL/FAX/EMAIL: (+822) 2123-5484/ (+822) 2123-8639/ [email protected]
48
Embed
The Impact of Labor Unions on External ... - kaa-edu.or.kr. Ju Ryum Chung.pdfthan do non-unionized firms. We also find that unionization is negatively (positively) associated We also
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
The Impact of Labor Unions on External Auditor Selection and Audit Scope
Data Availability: data are publicly available from sources identified in the paper.
3
INTRODUCTION
Auditors play a critical role in determining the quality of financial reporting through
external audits. Not surprisingly, high-quality external audits performed by independent auditors
are essential for maintaining efficient capital markets and protecting investors and other
information users. We examine how unions, an important internal user group of financial
statements, influence external audits. While studies on the role of unions have focused on
accounting choices and earnings quality such as income smoothing and conservatism (e.g.,
DeAngelo and DeAngelo 1991; D’Souza et al. 2001; Farber et al. 2012; Leung et al. 2009), no
study has investigated the role of unions in external audits. Using publicly available Korean
union data, we examine whether unions demand high audit quality and thus motivate
management to hire high-quality auditors such as Big 4 and industry specialist auditors. This
study also examines unions’ role as determinants of abnormal audit fees and audit hours.
A union’s main role is to protect employee rights and demand the improvement of
employee welfare. Negotiating wages with management is one of its most important tasks.
Unions rely on financial information when negotiating wage increases with management. For
that purpose, unions make every effort to acquire highly accurate and transparent financial
information representing the reality of the business (Kleiner and Bouillon 1988; Appelbaum and
Hunter 2007; Leung et al. 2010). In addition, unions demand high-quality financial reporting in
order to be able to monitor management effectively and secure their jobs against bankruptcies
resulting from deteriorating financial conditions.
Contrariwise, management has incentives to hide inside information on resources or true
operating income from unions (Bova 2013). Better informed unions will have a better position in
wage negotiations. Management tends to consider unions rent-seekers instead of value creators
4
(Ruback and Zimmerman 1984, Connolly et al. 1986; Hirsch 1991). Thus, the managers of
unionized firms have incentives to manipulate earnings downward as unions’ demand for wage
increases will be high if the firm performs well (Liberty and Zimmerman 1986). Hiding
information on available resources could allow the managers to achieve a more desirable
outcome from wage negotiations or to pursue their own benefits (Hilary 2006; Matsa 2010;
Klasa et al. 2009; Farber et al. 2012). Managers may therefore prefer lower-quality external
auditors, who are more likely to acquiesce to requests for earnings management. However,
management may also have incentives to hire higher-quality external auditors to avoid
unnecessary doubts from unions about the quality of their financial statements or to give the
impression that all relevant information is being disclosed.
This study first examines the role of unions in the selection of external auditors.1
Specifically, this study examines whether the presence of unions increases the likelihood of high-
quality auditors being selected. It is prevalent practice in Korea to external auditors through the
auditor selection committee, where a union representative, an outside member, an audit
committee member or an internal auditor, and management usually participate in and influence
the decision. High-quality auditors are expected to ensure that reported earnings are faithfully
representative of actual performance. We define a high-quality audit as one conducted by a Big 4
or an industry specialist auditor (e.g., Balsam et al. 2003; Behn et al. 2008). We also predict that
this effect of unions, if it exists, is more pronounced in firms with stronger unions. Union
strength is measured by membership: we consider Minju, Hanguk, and unaffiliated unions, where
Minju is considered the most aggressive and unaffiliated unions the least.2
1 There is anecdotal evidence in Korea that unions were involved in auditor selection processes. In December 24,
2002, Yonhap News reported that the “Employee Stock Ownership Association of Hyundai Moto Co. demanded a
change of an external auditor because the auditor has been in position for a long period and provided non-audit
services which may damage their independence as an external auditor. The Association also reported this potential
loss of auditor independence to their labor union, so that the union can exercise the right of veto against the
reappointment of the incumbent auditor.” Yonhap News also reported that “the union of Hyundai Moto Co. has the
right to demand that the firm change its external auditor.” 2 Unions in Korea are classified into three groups: (1) a member of Minju Korean Confederation of Trade Unions
5
For the role of unions on quality or scope of external audits, we test an association
between unionization and abnormal audit fees and audit hours.3 Following prior studies (e.g.,
Hribar et al. 2014), we consider unexplained (i.e., abnormal) audit fees as a proxy conveying
information about firms’ accounting quality. Unexplained (or abnormal) audit fees reflect the
private information collected by auditors, such as unobservable information about firms’
underlying accounting systems. We perform these tests separately by sign of abnormal audit fees
and audit hours because of the different implications of positive and negative abnormal audit fees.
Positive abnormal audit fees may create the auditor incentive to compromise
independence, which impairs audit quality (Dye 1991; Choi et al. 2010). Positive abnormally
high audit hours may imply auditor inefficiency or hidden audit risk not captured by publicly
available explanatory variables. If unions effectively perform monitoring over the quality of
external auditors, unions will influence management to reduce the positive abnormal audit fees
and audit hours to the optimal level. Alternatively, if positive abnormal audit fees and audit
hours represent the auditors’ extra audit efforts, unions have no reason to want the high audit
fees and audit hours reduced. Therefore, it is unclear ex-ante how unionization is associated with
positive abnormal high audit fees.
By contrast, we posit a clear relationship between unions and negative abnormal audit
fees and audit hours. We believe that abnormally low audit fees (and hours) imply an inadequate
allocation of audit resources, signaling an “underutilization” of audit services that leads to lower
audit quality and financial statements. Thus, if unions monitor resources allocated to and by
external auditors and block the underutilization of audit services, they are likely positively
associated with negative abnormal audit fees and hours.
(Minju hereafter), (2) a member of Hanguk Federation of Korean Trade Unions (Hanguk hereafter), and (3)
unaffiliated unions. More detailed descriptions of unions in Korea are included in Section 2.1. 3 Unlike in other countries, actual audit hours are publicly disclosed in annual reports in Korea; thus, the effect of
unions on audit hours can be analyzed.
6
This study utilizes publicly available union data unique to Korea to test the effects of
unions on the selection of external auditors and resource allocations to and by auditors. Korean
regulators required publicly traded firms to disclose whether they had a union, their unions’
affiliation, the number of employees who belong to it, and the number of full-time union
administrators until 2008. These data appear to be publicly available only in Korea, and they
cover only this specific time period. Several U.S. studies use estimated unionization or labor
intensity data at the industry level, since they were unable to use real firm-level union data, a
substantial limitation (e.g., Hilary 2006; Matsa 2010; Farber et al. 2013; Chen et al. 2011; Chyz
et al. 2013).
We empirically find that the Minju (stronger union) significantly increases the likelihood
of Big 4 or industry specialist auditors being chosen. We also find that non-affiliated (the
weakest) unions reduce the likelihood of higher-quality auditors being engaged. We further find
that both Minju and Hanguk reduce abnormally positive abnormal audit fees and audit hours and
increase abnormally negative abnormal audit fees and audit hours. On the other hand, non-
affiliated unions have virtually no impact on abnormal audit fees and audit hours.
This study provides several contributions to the extant literature on both unions and
external audits. First, this appears to be the first study to use large public data to empirically
explore the association between labor unions and external audits. We provide evidence that
unions, who are important users of financial statements and have been relatively ignored by the
literature, affect audit-related decisions such as auditor selection and audit scope. This study
extends the literature that unions affect various business decisions (DeAngelo and DeAngelo
1991, Hilary 2006, Matsa 2010).4
4 Prior studies have provided results that unions have effects on earnings management (DeAngelo and DeAngelo
1991; Bova 2013), investments in R&D (Connolly et al. 1986), cash holdings (Klasa et al. 2009), leverage (Matsa
2010), accounting conservatism (Farber et al. 2012), CEO compensation (Banning and Chiles 2007), cost of equity
7
Second, by using the unique Korean data, individual firm-level union measure,5 we can
investigate how the different characteristics of labor unions affect audit selection and resource
allocations to/by auditors. We find that the type of union (i.e., strong versus weak) has
differential effects on firms’ decisions on auditor selection and audit scope. Unlike most other
studies that focus on negative aspects of unions, our study is one of a few studies (e.g., Leung et
al. 2010) that provides evidence of a positive role of unions on audit quality and thus financial
reporting.
The rest of this paper is organized as follows. Section 2 presents the literature review and
hypothesis development. Section 3 describes the research design and sample and provides
descriptive statistics. Section 4 reports the empirical results. Finally, Section 5 concludes the
study.
LITERATURE REVIEW AND HYPOTHESES
Unions in Korea
Most unions in Korea belong to the Hanguk or Minju federations. According to a 2008
Labor White Paper,6 of 5,889 unionized firms (1,559,172 members), 3,429 belonged to Hanguk
(755,234 members), and 1,143 belonged to Minju (627,274 members). The remaining 1,317
firms (176,671 members) were unaffiliated. Thus, about 89 percent of unionized employees
belonged to one of the two major associations, and Hanguk had most. Hanguk and Minju have
very different founding histories, ideologies, and propensities. Not surprisingly, these two
(Chen et al. 2011), and cost of debt (Chen et al. 2012). 5 Existing studies based on U.S. data mostly use industry-level measures (e.g. Chen et al., 2008 and Klasa et al.,
2008). 6 The Labor White Paper is published every year by the Korean Ministry of Employment and Labor to present and
evaluate labor policies. The document, open to the public, includes various statistics on labor markets such as
employment, education, work environment, and labor relations.
8
associations have exhibited very different strategies and approaches in dealing with wage
negotiations and government labor policies (Yoon and Lee 2008).
Hanguk, established in 1954, was sympathetic to government policies and was the only
legal union association until 1996. Under Korea’s authoritarianism regimes, Hanguk followed a
pro-government line and formulated policies that were moderate and management-friendly
(Korea Labor Institute 2003).7 Minju was founded in 1990 by union members who disagreed
with Hanguk’s pro-government policies. Minju became legal in 1996 after five years of
government oppression. Because of this different history, the two organizations pursue different
methods. As frequently reported in major Korean media, Minju has been militant and aggressive
towards government policies and management.8 According to the Labor White Paper, 87.2
percent (565 cases) out of 648 strikes occurred because of Minju, while 11.4 percent (83 cases)
were initiated by Hanguk between 2005 and 2008.9 This suggests that Minju tends to take an
aggressive position and is not reluctant to go on strike if its demands are not satisfied. On the
other hand, Hanguk tends to take a moderate position and will compromise with management as
much as it can.
The number of unaffiliated unions has been increasing since 2000 because some unions
have been unsatisfied with the policies and ideas of the major federations and have not joined or
have disaffiliated from them. These unaffiliated unions are more interested in practical issues
than political issues and prefer moderate solutions over drastic strikes. According to a panel
7 MK Business News (August 25, 2008) reported that some Hanguk leaders were appointed as members of the new
government cabinet and elected as congressmen while none from Minju was appointed or elected. These are
consistent with Hanguk’s status as a government-friendly association. 8 Strikes initiated by Minju are frequently reported in major newspapers. For example, Newsis Media reported on
June 18, 2008 that the union at Hyundai Motors Co., a major Minju member, had begun its fourth strike of the
month. The four strikes had prevented the production of 15,514 cars, representing a loss of about 238 million dollars.
The Kukmin Daily News also reported on August 28, 2007, that strikes at E-land Co. and Newcoa Co., members of
Minju, lasted for 80 days and caused serious damage to the firms. 9 According to the Labor White Paper, the causes of the strike were related to collective bargaining and wage
negotiations (91.2 percent) as well as restructuring and overdue wages (8.8 percent).
9
study of the Korea Labor Institute (2008), the negotiated labor hours per week average 47.6
hours for Minju, 48.7 hours for Hanguk, and 59.6 hours for unaffiliated unions. The average
number of collective bargaining sessions also differs across union confederations (7.94 times for
Minju, 6.12 for Hanguk, and 5.03 for unaffiliated unions). Minju also shows the greatest
difference between wage increase demands and the rate suggested by management during
collective bargaining (with Minju at 5.0 percent, Hanguk at 4.6 percent, and Unaffiliated at 2.7
percent). Thus, the most aggressive union is Minju, followed by Hanguk and unaffiliated unions.
Related Literature and Hypothesis Development
Unions need information for wage negotiations, but managers tend to hide inside
information for their own benefit (Hilary 2006). Having more information enables unions to
function better during wage negotiations and gain more resources (Kleiner and Bouillon 1988).
Providing more information to unions is likely to increase unions’ bargaining power because it
may eliminate managers’ grounds for refusing wage increases. Experimental results also suggest
that managers are better off negotiating with uninformed unions (Croson 1996). Therefore, the
best option for managers facing wage negotiations is not to share information with unions. Using
Canadian data, Scott (1994) empirically finds that firms facing a higher likelihood of a strike or
operating in an industry with high average salaries tend to reduce the amount of information on
pension issues. Furthermore, unions’ demands for wage increases should be stronger when the
union is stronger, and managers’ incentives to limit information should thus also be stronger
(Klasa et al. 2009; Matsa 2010). As a result, information asymmetry increases with union
strength (Hilary 2006).
Management considers labor unions as rent-seekers (Grout 1984; Connolly et al. 1986;
Hirsch 1992; Klasa et al. 2009; Matsa 2010). Unions have incentives to extract as much quasi-
rent as possible through collective bargaining and strike threats (Grout 1984). Moreover, unions
10
tend to demand wage increases when their firms are performing well, while unions tend to
consent to current wage levels when the firms’ performance is poor (Reynold 1978;
Blanchflower et al. 1996). Management thus has incentives to manipulate earnings downward in
order to minimize rent seeking (Liberty and Zimmerman 1986; Farber et al. 2013). Liberty and
Zimmerman (1986) test the hypothesis that managers have an incentive to manage earnings
downward when they predict wage negotiations but find no evidence of earnings manipulation
prior to wage negotiation. Subsequent studies (e.g., Mautz and Richardson 1992; Cullinan and
Knoblett 1994) fail to provide evidence that managers exercise discretion on earnings through
accounting choices prior to wage negotiations.
However, DeAngelo and DeAngelo (1991) provide evidence of earnings management in
the year of wage negotiation. They find that firms report larger losses in years when wage
negotiations occur. D’Souza, Jacob, and Ramesh (2001) also find evidence consistent with the
earnings management argument that management reduces labor negotiation costs through a
discretionary selection of accounting choices. They document that unionized firms tended to
select the immediate recognition method in the year of SFAS 106 adoption, resulting in lower
earnings in that year. Similarly, Bova (2013) offers evidence supporting the management
incentive to provide a negative outlook to unionized firms. He documents that unionized firms
are more likely to just miss analysts’ forecasts through both expectation and earnings
managements and that this tendency is not restricted to wage negotiation periods.
For unions facing managements with incentives to manipulate information, obtaining
high-quality financial information is critical for successful negotiations (Appelbaum and Hunter
2007; DeAngelo and DeAngelo 1991; Leung et al. 20 10). The high-quality information that
unions demand should faithfully represent the real status of the companies’ business.10 If
10 Studies (e.g., Faleye et al. 2006) also argue that unions prefer conservative earnings because they act like
11
earnings are manipulated by management for opportunistic purposes, the union may not reap
sufficient benefits. Therefore, unions have strong incentives to deter accounting manipulation
(Liberty and Zimmerman 1986). However, unions may not be able to monitor or evaluate the
appropriateness of the financial information management provides because they usually have no
financial expertise. Using high-quality and independent auditors would assure unions that the
information offered is representative of reality.
We use Big 4 and industry specialist auditors as proxies for high-quality auditors, as the
literature suggests that they offer better assurance on the quality of their financial information
than do non-Big 4 and non-industry specialist auditors (DeAngelo 1981; Teoh and Wong 1993;
Balsam et al. 2003; Khurana and Raman 2004; Behn et al. 2008).11 Big 4 auditors also provide
higher insurance coverage to parties that suffer losses through audit failures (Dye 1993; Lennox
1999; Fortin and Pittman 2007), making them more attractive to unions.
Thus, unions are a group of important stakeholders with the ability and incentive to
influence firms’ accounting choices and corporate financial decisions (DeAngelo and DeAngelo
1991; Cullinan and Knoblett 1994; D’Souza et al. 2001; Faleye et al. 2006; Chyz et al. 2013).
We predict that unions influence audit issues as well. We hypothesize that unions demand high-
quality audits and request that managers hire Big 4 or industry specialists, ceteris paribus,12
because unions equipped with high quality financial information are in a better position in wage
bondholders in that they claim fixed amounts from companies in the form of fixed wages and benefits. To secure
these fixed claims, unions tend to demand conservative accounting. For example, firms with inflated earnings may
pay dividends to shareholders, increasing the risk that unions, as fixed claimers, will lose guaranteed wages and
benefits. In addition, conservative accounting is less vulnerable to litigation and bankruptcy risks. 11 Lawrence et al. (2011) suggest that the audit quality of Big 4 and industry specialist auditors is not significantly
higher after controlling for client characteristics. However, DeFond et al. (2014) cast doubt on their findings that
differences in client characteristics cause the appearance that Big N auditors provide higher audit quality. 12 Faleye et al. (2006) note that unions can influence firms’ investment activities directly or indirectly through at
least three avenues: (1) unionized employee activism, (2) the introduction of cooperation agreements between
unions and management, and (3) shareholder activism. We believe that unions can also use these three avenues to
influence audit issues. For anecdotal evidence of union activism, see footnote 3.
12
negotiations with management. Motivated by the above discussion, our first hypothesis, stated in
the alternative form, is as follows:
H1a: Unionization is positively associated with the probability of Big 4 or industry
specialist auditors being selected, ceteris paribus.
We now turn to the research question of whether union type (i.e., strong versus mild) has
a differential effect on audit-related issues. We believe that the union request to hire Big 4 or
specialist auditors should be stronger when unions are stronger and/or more active. Strong unions
possess greater bargaining power through the threat of strike and therefore have greater influence
on management decisions (Farber et al. 2013). Managers could also be proactive in selecting
high-quality auditors to satisfy the demand from strong unions if their best interest is to minimize
unnecessary uneasiness and uncertainty due to the threat of strikes and other harmful actions.
As mentioned, Minju is considered the more aggressive and active association because of
its frequent and aggressive strikes. Therefore, we expect that Minju unions possess more
bargaining power than other unions and thus have greater influence on management when
demanding high-quality audits. Managements facing strong unions may also have greater
incentives to hire high-quality auditors in the hope of avoiding unnecessary disputes with unions
by signaling that they are trying to provide high-quality information and have no intention of
hiding anything. We therefore posit that Minju unions are more likely to hire Big 4 or industry
specialist auditors. Thus, we propose the following hypothesis, in the alternative form:
H1b: Firms whose unions are affiliated with Minju (stronger union) are more likely to
hire Big 4 or industry specialist auditors.
Next, we examine whether unionization is associated with abnormal audit fees and audit
hours. Hribar et al. (2014) argue that abnormal audit fees (residuals from the audit fee model)
should contain information about accounting quality, as unexplained audit fees reflect
13
unobservable information assessed and priced by auditors. We conduct these tests separately
according to the sign of the abnormal audit fees. As positive and negative abnormal audit fees
have different implications, tests using both samples together may produce spurious results
(Picconi and Reynolds 2013).
Abnormally high audit fees create auditor incentives to compromise independence, which
impairs audit quality (Choi et al. 2010; Asthana and Boone 2012; Dye 1991). The normal audit
fee level is determined by factors suggested by prior studies, such as firm characteristics,
complexity, or risk. Abnormal fees are residuals that cannot be explained by those factors. Choi
et al. (2010) suggest that positive abnormal fees imply that the auditor has bonded economically
to the client. Abnormally high audit hours indicate the unnecessary allocation of audit resources
(Caramanis and Lennox 2008), signaling a type of audit inefficiency. An unnecessarily high
audit time could help build personal relationships between auditors and management, increasing
doubts about auditor independence (Tackett et al. 2004). Moreover, extra high audit fees (and
hours) could be bribes delivered by management to auditors in exchange for weak monitoring.
These discussions lead us to predict that, if labor unions successfully monitor audit quality,
positive abnormal audit fees and audit hours will decrease in unionized firms.
However, positive abnormal fees and audit hours can also be interpreted as extra effort
(Eshleman and Guo 2014; Blankley et al. 2012).13 Auditors exert extra effort when assessing
high audit risk in clients. To maintain a certain level of audit risk, auditors expand their audit
scope by increasing substantive tests. Positive abnormal audit fees (or hours) reflect extra efforts
beyond the controlled risk factors that are included in the audit fee (or audit hour) model. These
extra efforts may lead to high audit quality. Auditors also exert extra effort when firms with
13 Still others (e.g., Picconi and Reynolds 2013) argue that abnormally high audit fees represent a risk premium
charged by auditors for firms with high audit risk. See DeFond and Zhang (2014) for a review of audit fees as a
proxy of audit quality.
14
high-quality governance request more thorough audits to minimize litigation risks (Carcello et al.
2002; Abbott et al. 2003). In these cases, positive abnormal audit fees and hours are considered
desirable and consistent with labor unions’ interests. The above arguments may lead to a
prediction that unions encourage extra audit effort, suggesting a positive association between
unionization and positive abnormal audit fees and hours. Due to this conflict between predictions,
we state the hypothesis non-directionally:
H2a-1: Unionization affects positive abnormal audit fees and audit hours.
Regarding negative abnormal audit fees and hours, we predict a positive association with
unions. Normal audit fee and hour levels should be secured to ensure audit quality, at the
minimum. Thus, negative abnormal audit fees and audit hours are interpreted as an
underutilization of audit services (Picconi and Reynolds 2013) or something that confers greater
bargaining power onto audit clients (Asthana and Boone 2012), both of which lead to lower-
quality audits. Higher-quality auditors can charge clients higher audit fees (e.g., the Big N
premium), which suggests higher audit quality (Simunic 1980, Francis 1994). Audit fees lower
than the norm may therefore be related to low audit quality. Unions that prefer higher-quality
audits will make the effort to deter this underutilization of audit services (i.e., lower-quality
audits). Consequently, we hypothesize that negative abnormal audit fees and audit hours will be
positively associated with labor unions:
H2a-2: Unionization is positively associated with negative abnormal audit fees and audit
hours.
Consistent with the rationale for H1b, we argue that strong unions have greater influence
on company decisions and that their demand for higher audit quality should be stronger. This
leads us to predict that unions’ effect on abnormal audit fees and audit hours are more
pronounced in stronger unions (i.e., Minju):
15
H2b. The effects of unionization on abnormal audit fees and audit hours, if they exist, are
more pronounced in unions affiliated with Minju (stronger union).
RESEARCH DESIGN AND SAMPLE SELECTION
Research Design
To investigate whether unionization is associated with auditor choice, we estimate logistic
regressions in equation (1) and (2) as baseline models.
ABFEE: audit fees minus the normal level of audit fees;
ABHOUR: audit hours minus the normal level of audit hours;
Ini: 1 for the first year of audit, 0 otherwise; and
See equations (1) and (2) for the definitions of other variables.
The variables of interest are Union in model (4) and Minju, Hanguk, and Nonaffiliated in
model (5). As mentioned, we estimate models (4) and (5) separately according to the sign of the
abnormal audit fees and hours. If unionization increases (reduces) the abnormal audit fees or
audit hours, coefficient β1 will be significantly positive (negative) in model (4). Furthermore, we
investigate the impact of union type in model (5). If abnormal audit fees increase (decrease) in
firms with Minju unions relative to those in non-union firms, for example, coefficient β1 will be
significantly positive (negative). We include the same control variables used in models (1) and (2)
and add two more indicator variables: one for the first year of audit (Ini) to control for the
19
possibility of lowballing (Simon and Francis 1988) and one for Big 4 (BIG) to control for the fee
difference between Big 4 and non-Big 4 auditors (Craswell et al. 1995). Following the literature,
we expect the variable of Ini to negatively affect and the variable of Big to positively affect
abnormal audit fees and audit hours. The definitions of the variables are summarized in Table 1.
[Insert Table 1 here]
Sample Selection
The initial sample comprised 6,594 non-financial firms listed on the Korean stock market
between 2005 and 2008. We impose 2008 as the limit because union data were no longer
mandated in companies’ annual reports after 2009. Union data such as membership and
federation status are hand collected from the companies’ annual reports.15 The financial data are
obtained from KIS-Value III, which is equivalent to Compustat in the U.S. We exclude firms
without December fiscal year-ends to control for potential effects resulting from the difference in
fiscal year-ends. We also exclude firms whose financial data are not available from the KIS-
Value III database. Finally, we exclude firms affiliated with both Minju and Hanguk because
they have the characteristics of both associations. This process yields a final sample of 4,568
companies, the union sample of which is 1,751 (38 percent) and the non-union sample 2,817 (62
percent). Of the union sample, Minju and Hanguk firms account for 457 (26 percent) and 1,150
(66 percent) of the total, respectively; the remaining 144 firms (8 percent) are not affiliated with
Minju or Hanguk.16 The sample selection process is summarized in Table 2.
[Insert Table 2 here]
15 We also collect data on the number of unionized employees and fulltime union staff members and test whether
the result differs by including these variables. The un-tabulated results indicate that these variables are not
statistically significant, while the variables of interest remain unchanged. This result suggests that in Korea, the
affiliation with Minju or Hanguk represents the union’s strength or negotiation power better than the unionized ratio
or number of full time union members. 16 During the sample period, 49 firms established new unions, and 25 changed their associations; these account for
about 1.6 percent of the total sample. Thus, firms rarely change their associations.
20
EMPIRICAL RESULTS
Descriptive Statistics
Table 3 summarizes the basic statistics for all variables used in this study. Table 3 Panel
A presents the mean and median for the full sample (N=4,568) as well as the union sample
(N=1,751) and the non-union sample (N=2,817). We focus on mean values in this discussion, as
a discussion on median values would be virtually identical (with the exception of Invrec). The
statistical significance of the mean differences between the Union and non-Union samples is
reported in the Union column. Union firms are more likely to engage Big 4 audit firms (BIG)
than are non-Union firms (65.3 percent versus 49.4 percent). The auditor specialist variable,
ISPE_client, is significantly higher in the Union sample (32.6 percent versus 23.3 percent). We
find similar results for other industry specialist measures for assets and sales. These results are
consistent with the prediction that unionized firms tend to engage high-quality auditors.
Abnormal audit fees (ABFEE) and audit hours (ABHOUR) are significantly higher in the
non-Union group. The Union sample has a higher mean value for firm size (Size), ratio of export
sales to total sales (Export), audit complexity (Invrev), leverage ratio (Lev), return on asset
(ROA), and consolidated financial statements (Consol), but it has a lower mean value for loss
(Loss). These results suggest that unionized firms are, on average, bigger, more highly leveraged,
more complex, and more profitable.17
In Table 3 Panel B, we use only unionized firms to compare the mean and median of the
variables for each union group (Minju, Hanguk, and Nonaffiliated). The Hanguk column contains
the statistical significance of the mean and median differences between the Minju and Hanguk
samples, while the Nonaffiliated column includes the statistical significances of the mean and
17 To control for the possibility that the differences in firm characteristics between the Union and non-Union
samples are driving the main results, we match the sample by the propensity score and conduct the tests again. The
results are reported in Section 4.4
21
median differences between the Minju and Nonaffiliated samples. We find that 73.3 percent of
the Minju sample is audited by a Big 4 audit firm (BIG), while 63.1 percent of the Hanguk and
57.6 percent of the Nonaffiliated samples are audited by those firms. Minju firms are also more
likely to engage an industry specialist auditor (ISPE_Client) than are Hanguk and Nonaffiliated
firms. The Minju samples have, on average, greater abnormal audit fees (ABFEE), audit hours
(ABHOUR), firm size (Size), leverage (Lev), and complexity (Invrev, Consol) than the other
groups.
[Insert Table 3 in here]
Table 4 reports Pearson correlations for the variables used in our regression analyses. We
find significant and positive correlations between union variables (Union, Minju, and Hanguk)
and higher-quality auditors (BIG, ISPE), while non-affiliated unions (Nonaffiliated) are not
significantly correlated with BIG and ISPE. These results suggest that firms with affiliated
unions are more likely to hire a Big 4 or an industry specialist auditor. In addition, the abnormal
audit fees and audit hours are negatively correlated with the union variables (Union, Hanguk, and
Nonaffiliated).
[Insert Table 4 here]
Regression Result
Table 5 presents the results of the tests regressing audit firm choice on Union. Columns
(1) and (2) report the union impact on the choice of a Big 4 auditor (BIG). The coefficient on
Union in column (1) is positive but not significant. We next assess the impact of each union
group by replacing Union with Minju, Hanguk, and Nonaffiliated in column (2), where we find a
positive coefficient of Minju that is statistically significant at the five percent level, an
insignificant coefficient of Hanguk, and a negative coefficient of Nonaffiliated significant at the
22
five present level. These results suggest that firms affiliated with Minju are more likely to choose
a Big 4 audit firm than non-union firms, while those affiliated with Hanguk are not. The firms
with nonaffiliated unions are less likely than firms without unions to hire a Big 4 auditor, which
may indicate that the nonaffiliated union lacks the power to successfully demand higher-quality
auditors.
In columns (3) to (6), we report the impact of Union on the choice of industry specialist
auditor (ISPE). Columns (3) and (4) show the regression results based on the industry specialist
classification defined by the number of audit clients. The variable of Union in column (3) is
significantly positive (p<0.05), suggesting that unionization positively affects the choice of
industry expert. In column (4), the coefficient of Minju is significantly positive (p<0.01), that of
Hanguk is positive but not significant, and that of Nonaffiliated is negatively significant (p<0.1).
We apply other definitions of industry specialist, based on client asset and sales, in
columns (5) to (6). The results are similar to those reported in columns (3) to (4). Overall, we
find strong evidence that firms affiliated with Minju are more likely to hire a Big 4 or an industry
specialist auditor, and only limited evidence that firms with Hanguk are associated with auditor
choice. Interestingly, firms with nonaffiliated unions (Nonaffiliated), relative to non-union firms,
are less likely to engage a Big 4 or an industry specialist auditor. Regarding the control variables,
we find that the probability of hiring a high-quality auditor is generally high in firms that are
large and less risky in terms of leverage, as predicted.
[Insert Table 5 in here]
Table 6 shows the results of testing the effects of unions on abnormal audit fees and audit
hours. In Panel A, we conduct separate tests according to the sign of the abnormal audit fees and
hours. The impact of unionization on abnormal audit fees is reported in columns (1) to (4) in
Table 6. When actual audit fees are higher than normal (i.e., predicted) audit fees (i.e.,
23
ABFEE>0), Union decreases the abnormal audit fee (-0.047, t value=-4.25) in column (1), and
this negative effect of Union probably comes from both Minju (-0.068, t value -4.14) and
Hanguk (-0.041, t-value -3.31), as shown in column (2). Nonaffiliated firms have no significant
impact on the audit fees. These results suggest that Union (Minju and Hanguk) effectively
reduces the potential for a loss of auditor independence, while unaffiliated unions do not. When
the actual audit fees are lower than normal (i.e., ABFEE<0), Union increases the negative
abnormal audit fees in column (3), and this positive effect of Union also comes from both Minju
and Hanguk, as shown in column (4) This result suggests that both Minju and Hanguk effectively
prevent the deterioration of audit quality, while Nonaffiliated does not. We also notice that, in
both positive and negative cases, the coefficient of Minju is significantly larger (at p<0.01 from
F-test) than that of Hanguk, which suggests that the Minju, the stronger union, reduces abnormal
audit fees more than does Hanguk.
The impact of unionization on abnormal audit hours is reported in columns (5) to (8) in
Panel A. When actual audit hours are higher than normal or necessary (i.e., AHOUR>0), Union
can effectively reduce the abnormal hours (-0.041, t value=-3.03), as seen in column (5). This
reduction happens in all union groups, as shown in column (6). When AHOUR<0, Union
increases the abnormal audit hour, as indicated by the positive coefficient of Union (p<0.01) in
column (7). This effect probably comes from both Minju (0.103, t-value=3.22) and Hanguk
(0.053, t-value=2.27) but not from Nonaffiliated, as shown in column (8) Abnormally lower
audit hours likely lead to lower audit quality; therefore, Minju and Hanguk effectively deter the
audit quality impairment resulting from low audit effort. Overall, we find that both Minju and
Hanguk effectively increase the negative abnormal audit fees/hours and decrease positive
abnormal audit fees/hours, which suggests that they effectively monitor audit quality. By contrast,
nonaffiliated unions generally lack the power to monitor audit quality.
24
In Panel A of Table 6, we conduct tests separately according to the sign of the abnormal
audit fees and hours. As a sensitivity test, we also test whether unionized firms reduce the
absolute values of abnormal audit fees and audit hours. Asthana and Boone (2012) contend that
Note: This table reports the Pearson correlations among the variables used in regression tests. See Table 1 for the definition of all variables. ISPE is reported based on
ISPE_Client, where industry specialists are defined as the audit firm that has the largest number of clients in the industry.