Impact of Parent Companies and Multiple Large Shareholders on Audit Fees in Stakeholder-Oriented Corporate Governance

Abstract

This study aimed to investigate the impact of parent companies and other multiple large shareholders (MLSs) on the audit fees in Japanese firms, where stakeholder-oriented corporate governance is adopted. In such a firm, monitoring by many stakeholders can mitigate conflicts among shareholders. However, because the key stakeholders of these firms tend to resolve information asymmetry problems through insider communication, the level of audit effort is affected not only by the audit risk from principal–principal conflicts, but also by the demands of key stakeholders. Japanese parent companies tend to spin off their departments with high growth potential and provide incentives to lower subsidiaries’ cost of capital through information disclosure. Therefore, parent companies require greater audit efforts, and consequently, audit fees are expected to be higher. However, when MLSs are shareholders of the listed subsidiary, they can obtain relevant information via private communication. Thus, the need for quality accounting information will be smaller, the level of audit effort required will be smaller, and as a result, audit fees will be smaller. The results are consistent with these expectations. This paper contributes to the sustainable growth and economic development of firms and markets and has implications for the development of effective corporate governance mechanisms.

1. Introduction

This study investigates the impact of direct ownership of a parent company and multiple large shareholders (MLSs) on the audit fees paid to the external auditors of listed subsidiaries in stakeholder-oriented corporate governance.

Recent studies on ownership structure and audit fees have focused on the relationship between controlling shareholders and MLSs, not just controlling shareholders alone. The reason for this is that the actual ownership structure of firms in regions around the world, such as the United States, Europe, and East Asia, includes MLSs in many cases, rather than simple controlling shareholders and widely dispersed non-controlling shareholders [1,2,3,4]. Controlling shareholders can appoint board members and allow management to make decisions favorable to them at the expense of, or exploit the wealth of, non-controlling shareholders [5,6]. This can create a problem called principal–principal (PP) conflict between controlling shareholders and widely dispersed non-controlling shareholders. MLSs may collude with controlling shareholders to exacerbate PP conflicts [2,7,8] or mitigate PP conflicts through MLSs’ monitoring of controlling shareholders, competition for corporate control, or search for private information about management [9,10,11]. Transactions and unclear accounting disclosures that could arise from PP conflicts could increase audit risk for external auditors and affect audit fees. Indeed, a previous study that investigated French family firms with MLSs [12] shows that audit fees decrease with the presence of MLSs, under the assumption that family ownership increases audit risk. This suggests that MLSs mitigate PP conflicts and reduce audit risk by enhancing governance through monitoring and competition for controlling shareholders.

The motivation for this study is that recent studies have pointed out that the relationship between firm ownership structure and audit fees may differ significantly depending on the economic environment, including corporate governance orientation and the legal system surrounding the firm [13]. In countries where shareholder-oriented corporate governance is adopted, investor protection is high [14] and external stakeholders demand timely and high-quality public accounting information to mitigate information asymmetries [15]. Given the strong expectations for disclosure to reduce information asymmetry in shareholder-oriented countries [15] and the high litigation risk [16], especially in the US, the relationship between ownership structure and audit fees is considered to reflect the level of agency conflicts associated with the ownership structure [13]. On the other hand, in countries where stakeholder-oriented corporate governance is adopted, accounting information is more susceptible to political interference, information asymmetries among stakeholders tend to be mitigated through private communication, and litigation risks are smaller [15,16]. As a result, in stakeholder-oriented countries, the relationship between ownership structure and audit fees may change depending on the demands of controlling shareholders to the auditor, as the auditor’s interest shifts from litigation to client demands [13,17]. In fact, previous studies that investigated the relationship between corporate governance orientation and audit fees through international comparisons [13] show that the relationship between ownership structure and audit fees differs when corporate governance orientation is different. Meta-analyses have also shown that audit fees depend on different corporate governance and economic environments [18,19].

As previously mentioned, many firms have MLSs along with controlling shareholders. Previous studies dealing with the relationship between ownership structure including MLSs and audit fees [12] assume exploitation by controlling shareholders and PP conflicts and analyze the mitigation of conflicts by MLSs. On the other hand, in firms that adopt stakeholder-oriented corporate governance, it is important to analyze controlling shareholders’ and MLSs’ demands for auditors, because the intensity of their demands for audit effort can be a determinant of audit fees [13]. This paper addresses this research gap.

The value of this study arises from applying the analysis of the impact of controlling shareholders and MLSs on audit fees, already tested in previous studies [12], to Japanese firms adopting stakeholder-oriented corporate governance. The impact of audit fees based on demands for audit services by controlling shareholders can also arise in countries where shareholder-oriented corporate governance is adopted. However, in shareholder-oriented countries, it is difficult to observe the impact of controlling shareholders’ demands for audit services because audit fees are affected by the agency costs arising from PP conflicts between controlling and non-controlling shareholders [12]. On the contrary, we can address this issue by using the unique setting of a listed subsidiary in Japan. In Japan, parent companies tend to establish subsidiaries as part of their business strategy to create flexible organizational forms suitable for their environment [20]. At the same time, in Japanese firms that adopt a stakeholder-oriented corporate governance, monitoring by banks, employees, or between firms mitigate PP conflicts. In particular, ownership by the parent company leads listed subsidiaries to be managed on a long-term horizon, which tends to increase the sales growth rate of listed subsidiaries [21,22]. Conflicts between controlling shareholders (parent companies) and non-controlling shareholders, or PP conflicts, are not severe in Japanese listed subsidiaries because the growth of listed subsidiaries is also distributed to non-controlling shareholders through capital gains [21,22]. In addition, parent companies generally have an incentive to disclose information and reduce the cost of capital [23] for the long-term growth of their listed subsidiaries, and parent companies are likely to require greater audit effort from the auditors of their listed subsidiaries.

However, even for Japanese listed subsidiaries, theoretically a parent company can exploit the wealth of non-controlling shareholders through various transactions with its subsidiaries and may disclose unclear information to non-controlling shareholders. As previous studies have pointed out [24], a positive relationship between parent company control and audit fees can only be established if the financial gains from sending a signal to investors that a quality audit is being performed are large relative to the costs of a quality audit. If the parent company requires inadequate listed subsidiary disclosures, the demand for audit effort from the subsidiary’s auditor may be smaller, resulting in lower audit fees.

Thus, using the Japanese setting, we can analyze the demands for audit services of the controlling shareholder (parent company), and at the same time we can observe the interaction between the parent company and MLSs when MLSs exist as shareholders of the listed subsidiaries.

Using data on Japanese emerging companies from 2010 to 2019 (6653 firm-years), we find that (1) parent company ownership (PO) leads to higher audit fees for listed subsidiaries under stakeholder-oriented corporate governance, and (2) the existence of MLSs weakens the positive relationship between PO and audit fees of listed subsidiaries. These results are robust for controlling shareholders other than the parent company, the choice of MLS proxy variables, and self-selection bias.

This paper contributes to the literature on the relationship between ownership structure and audit, or stakeholder-oriented corporate governance, in the following points.

First, we show that in Japanese firms that adopt a stakeholder-oriented corporate governance, direct ownership by the parent company increases audit fees. Previous studies analyzing the relationship between ownership structure and audit fees have shown that as the ownership of a particular shareholder increases, that shareholder can directly monitor management through the appointment of board members [25]. Therefore, controlling shareholders in countries where stakeholder-oriented corporate governance is adopted enjoy rents from privately obtained information, and thus demand less effort from auditors, resulting in lower audit fees [13]. However, given the research context of Japanese spin-offs, audit fees may increase as parent companies have incentives to grow their subsidiaries and have a need to proactively disclose information to widely dispersed non-controlling shareholders to avoid increased cost of capital due to potential PP conflicts. In addition, MLSs have access to relevant information through private channels, which may reduce the demand for high-quality accounting information and decrease audit fees. The results of this paper support the latter view. Our more important contribution is that while previous studies have derived their hypotheses from the relationship between agency costs and audit risk due to PP conflicts [12,13], our study shows that audit fees are also affected by the demands of controlling shareholders and the presence of MLSs in situations where stakeholder-oriented corporate governance is adopted and PP conflicts are not serious. Thus, this paper shows that audit quality is also affected by the demands of controlling shareholders, which is particularly important for firms adopting stakeholder-oriented corporate governance [13], which are relatively influenced by the demands of controlling shareholders and MLSs.

The study also has implications for sustainable economic development and practical aspects of sustainability reporting. The audit system is an alternative monitoring mechanism that reinforces corporate governance by ownership structures [26] and is also a fundamental system that supports securities market disclosure, which is essential to the effectiveness of financial reporting. Therefore, this study’s results have implications for the development of effective corporate governance mechanisms in stakeholder-oriented systems, which will contribute to continued growth and economic development through these systems [27].

Relatedly, the relationship between corporate sustainability reporting and internal auditing has often been discussed in recent years [28,29]. External auditors also provide a variety of non-audit services, including sustainability reporting assurance and sustainability management [30]. The results of this paper suggest that in Japan, where stakeholder-oriented corporate governance is adopted, audit practice is influenced by the demands of controlling shareholders and MLSs when they exist. Therefore, services related to sustainability reporting may also be affected by controlling shareholders and MLSs. Therefore, this study is useful not only for auditors but also for standard setters who consider the audit system and its relationship with non-audit services.

The rest of this paper is organized as follows. In Section 2, relevant studies are discussed. In Section 3, the hypotheses are developed, and Section 4 indicates the analytical methods and data. Section 5 presents the results of our analysis, and Section 6 presents the findings and implications of this study.

2. Research Background

2.1. Listed Subsidiaries in Japan

According to a study on spin-offs of Japanese firms [20], the firms have used spin-offs as a management strategy to create flexible organizational forms suitable to their environment. Interestingly, compared to U.S. firms, where parent companies use spin-offs to focus on their core business and extract benefits for their shareholders [31], many Japanese firms see potential in the spin-offs of the parent company [20]. In fact, in Japan, when a spin-off business unit becomes a subsidiary, it is common for the parent company to not transfer the business unit’s shares to their shareholders but rather retain them [32]. Further, the subsidiaries’ short- and long-term sales and profit growth rates are greater than those of the parent company [32]. These characteristics of Japanese corporate spin-offs can be explained by reduced transaction costs, operation concentration, and employment of qualified personnel of the firm’s internal labor force based on long-term employment and seniority as per the Japanese management style [20]. Surprisingly, there have been no significant changes in the human resource-related priorities and practices of Japanese firms after the bubble economy burst in 1991. In addition, more spin-offs have been established since then. Even today, Japanese firms consider spin-offs as one of their management strategies [33].

Before the 2000s, the market for emerging firms was underdeveloped, and subsidiaries spun off from their parent companies could not be listed on securities markets unless they met certain conditions. For example, only a small number of subsidiaries fulfilling certain requirements, such as having a large size and high liquidity, were allowed to list their shares in the First Section or the Second Section of the Tokyo Stock Exchange. The number of listed subsidiaries in Japan increased dramatically after the establishment of the market for emerging firms around the 2000s [34]. For example, the Tokyo Stock Exchange opened the Mothers market for small and medium-sized firms in 1999. In 2004, JASDAQ, an over-the-counter market for emerging firms, received a license from the Japanese Prime Minister for securities exchanges. The development of these markets for emerging firms has made it possible for even relatively younger and smaller subsidiaries to be listed on stock exchanges. Many subsidiaries listed after 2000 are likely to list on these markets for emerging companies.

2.2. Concentrated Ownership and Audit Quality

When ownership is concentrated, block holders have an incentive to seek information and monitor management directly [35]. Further, as concentrated ownership by block holders increases, block holders can appoint board members to companies and access insider information [13,25,36]. Thus, more concentrated ownership can mitigate principal–agent (PA) conflicts between controlling shareholders and management. However, more concentrated holdings by block holders can lead to PP conflicts, as they can exploit wealth from other shareholders [5,6,37]. Auditing is a monitoring system that is different from ownership structures that can be implemented by shareholders paying additional costs [26]. Audits help reduce information asymmetries by ensuring high-quality accounting information and mitigate the agency problems by ensuring that all shareholders are treated equally [13,25,38]. However, the relationship between the demand and supply of audit quality depends on the strength of investor protection and the corporate governance system adopted by the firm [13].

In countries that adopt shareholder-oriented corporate governance, such as Anglo-American countries, where investor protection is generally strong [14] and concentrated ownership is low [39], information asymmetries are adjusted through the use of accounting information by the market [15]. Further, litigation risk is generally high, especially in the U.S. [16]. Therefore, the higher the business risk, the higher the audit risk, and consequently, the higher the level of audit fees [40,41].

In this context, as concentrated ownership develops and block holders arise, block holders can mitigate information asymmetries with management by appointing board members to mitigate PA conflicts [13,25]. However, in the case of insufficient investor protection, PP conflicts can become serious, as block holders may expropriate private benefits through improper transaction methods such as tunneling and inadequate financial reporting, harming other shareholders’ benefits [5,6]. As these transactions resulting from PP conflicts are subject to audit, the demand for audit effort rises, leading to increases in audit fees [12]. Indeed, although the findings of previous studies on the relationship between concentrated ownership and audit fees in countries where shareholder-oriented corporate governance is adopted are mixed [24,42,43,44,45,46], recent previous studies show that there is a downward convex, non-linear relationship between the concentrated ownership ratio and audit fees [13].

However, the relationship between PP conflicts and audit fees may be mitigated by the presence of MLSs, which can challenge the controlling shareholders. A previous study of these relationships for French family-owned firms finds that the presence of MLSs reduces agency costs and lowers audit fees [12].

By contrast, in countries where stakeholder-oriented corporate governance is adopted (e.g., Germany and Japan), concentrated ownership may mitigate PA conflicts, and PP conflicts may also be mitigated by monitoring stakeholders such as workers and banks [5]. Indeed, previous studies analyzing Japanese family firms, where stakeholder-oriented corporate governance is generally in practice, have shown that controlling shareholders in family firms are consistent with stewardship theory and that dividend payout ratios, profitability, and R&D expenditures are higher in family firms [47,48]. Further, previous studies investigating potential conflicts between parent and non-controlling shareholders in Japanese subsidiaries show that although dividends are reduced in subsidiaries, PP conflicts between parent and non-controlling shareholders are not severe, as the subsidiary’s sales growth is promoted [21,22]. Moreover, information asymmetries between stakeholders and management are mostly resolved through insider communication [15]. Thus, the demand for high-quality accounting information to mitigate information asymmetries among key stakeholders is small, reducing the required audit efforts [13]. Consequently, when considered from the demand side of audit, audit fees are determined by the demands of key stakeholders in countries where stakeholder-oriented corporate governance is adopted [13].

Even when considered from the audit supply side, in countries where stakeholder-oriented corporate governance is adopted, the level of audit fees is influenced by the demands of key stakeholders. This is because, in such countries (e.g., Japan), the litigation risk faced by auditors is also generally small [16], and in countries where the litigation risk faced by auditors is small, auditors are likely to be more responsive to their clients’ demands [17].

Therefore, in such countries, a different approach is needed to examine the relationship between audit fees and ownership structure rather than the agency costs of PP conflicts and the associated audit risk. In this regard, although some previous studies have investigated the relationship between concentrated ownership and audit fees by adopting a mechanism different from that of agency costs, no previous study has analyzed the relationship with MLS [13,25,49,50]. In the real world, a company’s capital structure often includes, in addition to controlling and widely distributed non-controlling shareholders, MLSs that may have a significant impact on audit quality [12].

3. Hypotheses Development

Japanese listed subsidiaries can be characterized by the following two points. First, Japanese firms have formed bank-centered or relationship-oriented business groups called keiretsu [51]. Therefore, many Japanese firms have developed strong inter-firm networks with other firms in terms of financial, human, and business relationships. Under these inter-firm networks, Japanese firms are monitored by banks and cross-holding companies in order to maintain strategic and long-term relationships. As a result, Japanese firms adopt so-called stakeholder-oriented corporate governance, which seeks to achieve goals beneficial to various stakeholders, including non-controlling shareholders [22,34,52,53]. Under stakeholder-oriented corporate governance, the demand for accounting information is small, as the interests among stakeholders are resolved through private channels [15].

Second, Japanese firms generally adopt a management strategy of spinning off growing businesses into subsidiaries [20]. Furthermore, Japanese listed subsidiaries distribute wealth to non-controlling shareholders through sales and capital gains because the parent company encourages the subsidiary’s sales growth for a long-term period [21,22]. Thus, PP conflicts between parent companies and non-controlling shareholders are not serious in Japanese listed subsidiaries [21]. Therefore, the controlling shareholder, namely the parent company, has a clear incentive to grow the subsidiary, in addition to not having to consider the serious agency costs arising from PP conflicts, as discussed in previous studies [12].

Based on these discussions, this paper focuses on the key stakeholders, controlling shareholders (parent companies), and MLSs and examines their impact on audit fees in terms of audit demands.

3.1. Parent Company Control and Audit Fees

Japanese firms tend to spin off growing businesses to establish them as subsidiaries as a management strategy [20]. Further, Japanese firms aim at maximizing market share and sales volume rather than seeking short-term profits [54]. Thus, Japanese parent companies are likely to encourage their subsidiaries to manage over a longer time horizon and invest in growth opportunities [21,22]. Moreover, under parent company control, wealth is also returned to stakeholders through sales distributions and capital gains from subsidiary growth, so even if wealth transfers occur from non-controlling shareholders to the parent company, this does not necessarily imply exploitation [21]. From this perspective, the parent company has an incentive to minimize conflicts with non-controlling shareholders while trying to grow the subsidiary in the long term.

It should be noted, however, that the parent company’s investment in the subsidiary and its sales growth is assumed to be associated with a lower dividend payout ratio [21] and a reallocation of wealth within the corporate group [22]. Moreover, previous studies on PP conflicts with respect to control by parent and family firms in Japan have limited the examined variables to dividends and profits, which does not necessarily mean that wealth deprivation by controlling shareholders does not occur [47]. In other words, greater information asymmetry between controlling and non-controlling shareholders (or between management and shareholders) could potentially increase the cost of capital for subsidiaries with an increase in agency costs as a result of potential wealth exploitation. Therefore, the parent company would require high-quality accounting information to mitigate information asymmetries between the parent company (or subsidiary management) and widely dispersed non-controlling shareholders. For example, parent companies are expected to have incentives to proactively disclose private information about the future cash flows of their listed subsidiaries in order to reduce information asymmetry with widely dispersed non-controlling shareholders, increase the liquidity of their shares [55], reduce their cost of capital [56], and improve their subsidiaries’ ability to raise capital [23]. Thus, the parent company tends to require greater audit efforts from the subsidiary’s auditors, which may result in higher audit fees.

However, the above discussion assumes that Japanese parent companies do not exploit the non-controlling shareholders of their listed subsidiaries and promote the long-term growth of the subsidiaries as part of their business strategy. As previous studies pointed out [24], a positive relationship between parent company control and audit fees is only possible if the financial benefits obtained by sending a signal to investors—that a quality audit is being performed—are large compared to the cost of a quality audit.

Inadequate disclosure helps protect the private benefits of the parent company, even at the expense of the non-controlling shareholders of the listed subsidiary [24]. Furthermore, the litigation risk for auditors in Japan is smaller than that in the U.S. [16], and auditors are likely to respond to the demands of the controlling shareholders, namely, the parent company [17]. Indeed, previous studies on PP conflicts between controlling and non-controlling shareholders and the quality of accounting information show that the more serious the PP conflict, the worse the quality of accounting information [57,58]. This information suggests that if the parent company requires inadequate listed subsidiary disclosure, the demand for audit effort from the subsidiary’s auditor is likely to be smaller, resulting in lower audit fees.

Thus, the relationship between control by the parent company and audit effort (audit fees) of listed subsidiaries can be explained from two different perspectives, and these are empirical issues. We, therefore, pose the following two hypotheses:

Hypothesis 1a (H1a).

Compared to other firms, audit fees for listed subsidiaries are high.

Hypothesis 1b (H1b).

Compared to other firms, audit fees for listed subsidiaries are low.

3.2. MLSs Ownership and Audit Fees

The actual capital structure is not binary, comprising controlling shareholders and widely dispersed non-controlling shareholders; rather, MLSs are also often the non-controlling shareholders [12]. MLSs have an incentive to monitor management due to their investment size [35]. They, therefore, will implement direct monitoring by appointing board members and collecting information directly from management [25]. Further, block holders such as MLSs have access to value-related private information [59]. For these reasons, the demand for the high-quality accounting information of MLSs as non-controlling shareholders is small, as the information asymmetry between the parent company and MLSs is eliminated through private channels.

From the perspective of the parent company, MLSs would also reduce the incentive to provide high-quality disclosures. Generally, parent companies hold more than 50% of the listed subsidiary’s shares to retain control of the subsidiary’s management. Thus, when MLSs exist, a large portion of the listed subsidiary’s financing is conducted among a very limited number of investors. Thus, the parent company can raise most of the listed subsidiary’s capital by simply responding to the MLSs’ demands, which reduces the demand for disclosure to other widely dispersed non-controlling shareholders.

As described above, from the viewpoints of both controlling and non-controlling shareholders, MLSs would reduce the parent company’s demand for high-quality audits. In fact, previous studies on firms with governance systems that differ from shareholder-oriented corporate governance showed that audit fees are smaller in firms with more concentrated ownership [13,25,50]. Based on the above, the following hypothesis is raised.

Hypothesis 2 (H2).

Listed subsidiaries with MLSsare restrained from increasing audit fees.

4. Methods and Data

4.1. Analysis Methodology

To investigate the impact of PO and MLSs on audit fees for Japanese firms, we relied on a previous study [12] to estimate the following model using ordinary least squares (OLS). Among the studies that have investigated the influence of MLS on audit fees, the study by Ben Ali, Boubaker, and Magnan [12] is the most recent and most impactful. Their model is based on a common understanding developed from previous audit fee studies. Therefore, by referring to their model, we can avoid estimation problems, such as the omitted variable bias. However, they focus only on family firms (firms with controlling shareholders) and do not examine the impact of controlling shareholders on audit fees and their interaction with MLSs. Therefore, we rearranged their model and used the following estimation model in this study.

$$\begin{array}{l}\mathrm{LNFEE}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{DPO}+{\mathsf{\alpha }}_2\mathrm{DPO}\times \mathrm{VARMLS}+{\mathsf{\alpha }}_3\mathrm{VARMLS}\\ +{\mathsf{\alpha }}_4\mathrm{OWN}1+{\mathsf{\alpha }}_5\mathrm{DMBANK}+{\mathsf{\alpha }}_6\mathrm{DFOREIGN}\\ +{\mathsf{\alpha }}_7\mathrm{SIZE}+{\mathsf{\alpha }}_8\mathrm{INTSALES}+{\mathsf{\alpha }}_9\mathrm{INVREC}+{\mathsf{\alpha }}_{10}\mathrm{NBS}\\ +{\mathsf{\alpha }}_{11}\mathrm{ROA}+{\mathsf{\alpha }}_{12}\mathrm{LOSS}+{\mathsf{\alpha }}_{13}\mathrm{BTM}+{\mathsf{\alpha }}_{14}\mathrm{LEVERAGE}+{\mathsf{\alpha }}_{15}\mathrm{CURRATIO}\\ +{\mathsf{\alpha }}_{16}\mathrm{BIG}4+{\mathsf{\alpha }}_{17}\mathrm{BUSY}+{\mathsf{\alpha }}_{18}\mathrm{OPINION}\\ +\mathrm{Industry}\mathrm{Dummies}+\mathrm{Year}\mathrm{Dummies}+\mathsf{\epsilon }\end{array}$$

(1)

4.1.1. Dependent Variable

LNFEE is the natural logarithm of audit fees.

4.1.2. Main Independent Variables

DPO is a dummy variable that indicates that the firm is a listed subsidiary. DPO is set to 1 if the firm-year is owned and effectively controlled by the parent company, such as through the appointment of management, and otherwise 0. Data on the presence of a parent company are obtained from NIKKEI Cges. DPO is 1 if there is a parent company that directly owns the subsidiary and 0 otherwise. It should be noted that, by the definition of this study, DPO does not indicate the existence of an ultimate parent company through indirect ownership. This is because we observe the impact of parent companies and MLSs simultaneously. It is impossible in our database to investigate the ultimate ownership of MLSs, and we define parent companies and MLSs by direct ownership. For H1a, α₁ is estimated as a positive value if the parent company requires a high audit effort from the subsidiary’s auditors. Conversely, for H1b, if the parent company demands a low audit effort from the subsidiary’s auditors, then α₁ is estimated as a negative value.

VARMLS is a set of variables indicating the presence of MLSs. As in previous studies [12,60,61,62], the DMLS is used primarily in this study. DMLS is a dummy variable that is 1 if at least one or more of the top two to four largest shareholders own 10% or more of the firm-year; otherwise, 0. In our analysis, we also use O234 for the percentage of shares owned by the top two to four largest shareholders in order to indicate the impact of the percentage of shares owned by MLSs. If the presence of MLSs reduces the demand of audit effort by the parent company from the subsidiary’s auditor, then α₂ is estimated with a negative value.

4.1.3. Control Variables

The estimation model includes control variables related to the following four issues: (1) Ownership variables other than parent companies and MLSs (OWN1, DMBANK, DFOREIGN), (2) Variables related to firm size and audit complexity (SIZE, INTSALES, INVREC, NBS), (3) Variables indicating the risk of potential losses to the auditor (ROA, LOSS, BTM, LEVERAGE, CURRATIO), (4) Control variables for other factors (BIG4, BUSY, OPINION).

OWN1 represents the percentage of the largest shareholder’s ownership of the firm-year. While this analysis focuses specifically on listed subsidiaries, it may not only be the presence of a parent company but also the percentage of ownership of the parent company that makes a difference in the degree of commitment to the subsidiary’s management [21]. The study sample also includes firms with controlling shareholders that are not parent companies. OWN1 aims to control for these effects. DMBANK and DFOREIGN are dummy variables set to 1 if the main bank and foreign investor own shares in the firm-year, respectively; 0 otherwise. The main bank’s monitoring is one of the features of corporate governance in Japan, and previous studies have shown that such monitoring reduces earnings management and improves accounting information quality [63]. Foreign investors also put pressure on information disclosure for their own benefit, increasing market liquidity [64]. In sum, monitoring by main banks and foreign investors may reduce audit fees by improving the quality of accounting information and lowering audit risk.

Previous studies on the drivers of audit fees from a supply-side perspective show that audit fees are related to client size, complexity, risk, and other factors [18,65]. We also control for these factors in this study. First, SIZE is a variable that controls for firm size and is the natural logarithm of total assets of the firm-year. Generally, the level of audit fees is mostly dependent on size [18]. Second, INTSALES is the ratio of foreign sales to total sales; INBREC is the ratio of inventories and accounts receivable to total assets; and NBS is the natural logarithm of the business segment plus 1. These are proxy variables for audit complexity. These variables are created by the firm-year. Third, ROA, LOSS, BTM, LEVERAGE, and CURRATIO are variables related to the auditor’s potential risk of loss. These variables are created for each firm-year. ROA is a variable that divides EBIT (earnings before extraordinary items before interest and taxes) by assets at the beginning of the year; LOSS is a dummy variable set to 1 if net income is negative; otherwise, 0. We also use BTM, the ratio of book to the market value of net assets, following a previous study [12]. LEVERAGE is the ratio of liabilities to total assets; CURRATIO is the ratio of current assets to current liabilities. Finally, we include BIG4, BUSY, and OPINION as variables that are systematically related to the level of audit fees. BIG4 is a variable that is set to 1 if the firm uses one of the Big 4 audit firms; otherwise, 0, and also controls for the audit fee premium due to the Big 4 [66]. BUSY is a dummy variable that is set to 1 for firms whose fiscal year ends in March when audits are performed during the auditor’s busy season, and 0 for all others [12,18]. Finally, OPINION is a dummy variable indicating the previous period’s audit opinion. OPINION is 0 for an unqualified opinion and 1 for otherwise [19].

4.2. Sample Selection and Descriptive Statistics

This study focuses on the Japanese securities market for emerging firms to investigate the impact of a parent company and MLSs ownership on audit fees. Financial and corporate governance data are obtained from Nikkei NEEDS Financial Quest (Nikkei Media Marketing, Inc.) and Nikkei NEEDS Cges (Nikkei Media Marketing, Inc.), respectively, while audit fees and auditors are obtained from the eol database (Pronexus, Inc.). The selection criteria for the data used in the analysis are as follows:

(1)

The data period is from 2010 to 2019.

(2)

The firm is listed on the Tokyo Stock Exchange, Mothers, or JASDAQ.

(3)

The firm does not belong to the banking, securities, insurance, or other financial industry sectors.

(4)

The data necessary for estimation are not missing.

To avoid the impact of COVID-19, data until 2019 are used in this paper. The number of firm-years meeting the above four criteria is 6653.

Table 1. Descriptive statistics.

	Mean	Std. Dev.	5% Perc.	Median	95% Perc.
LNFEE	3.039	0.394	2.398	3.045	3.664
DPO	0.093	0.290	0.000	0.000	1.000
DMLS	0.439	0.496	0.000	0.000	1.000
O234	0.209	0.091	0.073	0.200	0.375
OWN1	0.257	0.146	0.074	0.230	0.536
DMBOWN	0.451	0.498	0.000	0.000	1.000
DFOREIGN	0.245	0.430	0.000	0.000	1.000
SIZE	8.992	1.069	7.287	8.984	10.702
INTSALES	0.082	0.190	0.000	0.000	0.556
INVREC	0.287	0.169	0.030	0.275	0.590
NBS	1.215	0.461	0.693	1.386	1.946
ROA	0.035	0.137	−0.162	0.042	0.189
LOSS	0.215	0.411	0.000	0.000	1.000
BTM	1.162	0.994	0.115	0.960	2.902
LEVERAGE	0.461	0.227	0.119	0.461	0.822
CURRATIO	3.167	4.729	0.750	1.982	8.720
BIG4	0.602	0.489	0.000	1.000	1.000
BUSY	0.567	0.495	0.000	1.000	1.000
OPINION	0.178	0.383	0.000	0.000	1.000

Note: N = 6653.

summarizes descriptive statistics. First, for the main variables, the proportion of listed subsidiaries, or DPOs, averages at about 9.3% of the sample under analysis. The observed presence of major shareholders holding more than 10% of shares other than the controlling shareholder, i.e., DMLS, averages about 43.9% of the sample. The percentage of shares held by MLSs (mean value of O234) is about 20.9%. The percentage of shares owned by the largest shareholder, including the parent company, the mean value of OWN1, is about 25.7%. The other variables in Table 1 show that many firms have a main bank (the mean value of DMBOWN is 0.451). This is characteristic of the stakeholder-oriented corporate governance system adopted in Japan. However, as previous studies [52,63] pointed out, corporate governance in Japan is in a transition period, and the share of foreign investors has increased with the disposal of bad loans and bank restructuring since 2000. Indeed, in our sample, firms with foreign investors as shareholders, or DFOREIGN, averaged about 24.5%.

Table 2. Percentage of the parent company and MLS ownership in listed subsidiaries.

	Mean	Std. Dev.	5% Perc.	Median	95% Perc.
PO	0.558	0.106	0.410	0.535	0.761
DMLS	0.320	0.467	0.000	0.000	1.000
O234	0.156	0.091	0.044	0.131	0.329

Note: N = 618.

summarizes the percentage of PO, presence of MLSs (DMLS), and percentage of MLS ownership (O234) for only the listed subsidiaries in our sample (a total of 618 firm-years). The average percentage of PO is about 55.8% among the listed subsidiaries. Further, the average DMLS in listed subsidiaries is about 32.0%, and the average O234 is about 15.6%. Thus, on average, the top four shareholders, including the controlling shareholder, own about 71.4% (55.8% + 15.6%) of the shares in listed subsidiaries, and most of the funds obtained from the shares come from the parent company or MLS.

Table 3. Correlation matrix.

		(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1)	LNFEE	0.020	−0.036	0.019	−0.057	−0.302	0.107	0.549	0.169	0.160
(2)	DPO	1.000	−0.076	−0.184	0.584	0.015	−0.073	0.019	−0.053	0.036
(3)	DMLS		1.000	0.745	0.058	0.007	−0.105	−0.013	−0.124	−0.008
(4)	O234			1.000	−0.058	0.020	−0.039	0.071	−0.115	−0.001
(5)	OWN1				1.000	0.072	−0.142	−0.095	−0.158	−0.044
(6)	DMBOWN					1.000	−0.027	−0.178	−0.156	−0.063
(7)	DFOREIGN						1.000	0.337	0.080	0.148
(8)	SIZE							1.000	0.127	0.164
(9)	INTSALES								1.000	0.037
(10)	INVREC									1.000
		(11)	(12)	(13)	(14)	(15)	(16)	(17)	(18)	(19)
(1)	LNFEE	0.129	0.040	−0.034	0.078	0.183	−0.153	0.282	0.110	−0.009
(2)	DPO	−0.014	0.028	0.007	−0.093	0.009	−0.027	−0.033	0.005	0.034
(3)	DMLS	−0.014	0.094	−0.053	−0.047	0.021	−0.021	−0.023	−0.053	−0.032
(4)	O234	−0.033	0.159	−0.127	0.020	0.039	−0.087	0.038	−0.006	−0.075
(5)	OWN1	−0.058	0.158	−0.051	−0.165	0.032	−0.075	0.028	−0.063	−0.010
(6)	DMBOWN	−0.259	0.035	−0.016	−0.033	−0.096	0.080	0.071	−0.092	−0.049
(7)	DFOREIGN	0.007	0.061	−0.133	0.335	0.099	−0.178	0.121	0.167	−0.120
(8)	SIZE	0.104	0.200	−0.231	0.292	0.116	−0.121	0.116	0.173	−0.095
(9)	INTSALES	−0.113	−0.074	0.087	0.003	−0.107	0.091	0.107	0.047	−0.002
(10)	INVREC	0.018	0.081	−0.115	0.129	0.317	−0.247	−0.060	0.062	−0.043
(11)	NBS	1.000	−0.003	−0.003	0.007	0.159	−0.102	−0.126	0.083	0.082
(12)	ROA		1.000	−0.574	0.012	−0.087	−0.156	0.034	0.044	−0.149
(13)	LOSS			1.000	−0.070	0.055	0.173	−0.020	−0.055	0.162
(14)	BTM				1.000	−0.133	−0.077	0.061	0.143	−0.106
(15)	LEVERAGE					1.000	−0.477	−0.149	0.027	0.103
(16)	CURRATIO						1.000	0.075	−0.042	0.030
(17)	BIG4							1.000	0.017	−0.190
(18)	BUSY								1.000	−0.043
(19)	OPINION									1.000

Note: N = 6653.

summarizes Pearson’s correlation coefficients. Similar to previous studies [16,65], a strong positive correlation is observed between audit fees (LNFEE) and firm size (SIZE). The correlation between MLS-related variables (DMLS and O234) is high (0.745), suggesting that they capture the same factors. Besides this, no strong correlations are observed between the variables used in Equation (1), which would lead to multicollinearity.

5. Results

5.1. Main Results

Table 4. Main results.

	Full Sample
	Model 1			Model 2			Model 3
	Coeff.	t		Coeff.	t		Coeff.	t
const	1.125	36.740	***	1.160	33.570	***	1.163	36.890	***
DPO				0.086	5.860	***	0.089	3.057	***
DPO_DMLS				−0.151	−7.427	***
DPO_O234							−0.375	−2.809	***
DMLS				−0.012	−2.552	**
O234							−0.096	−3.207	***
OWN1	−0.041	−2.208	**	−0.109	−5.968	***	−0.099	−5.250	***
DFOREIGN	−0.157	−30.780	***	−0.158	−29.920	***	−0.157	−30.690	***
DMBOWN	−0.086	−15.530	***	−0.090	−16.190	***	−0.089	−16.570	***
SIZE	0.204	54.870	***	0.203	48.300	***	0.203	47.900	***
INTSALES	0.188	8.284	***	0.183	8.111	***	0.183	8.062	***
INVREC	0.198	11.340	***	0.191	10.280	***	0.193	10.540	***
NBS	0.036	7.099	***	0.035	6.723	***	0.035	6.740	***
ROA	−0.102	−2.751	***	−0.094	−2.466	**	−0.090	−2.340	**
LOSS	0.049	3.995	***	0.049	3.822	***	0.048	3.768	***
BTM	−0.025	−3.112	***	−0.026	−3.158	***	−0.025	−3.186	***
LEVERAGE	0.160	9.263	***	0.161	9.384	***	0.162	9.064	***
CURRATIO	−0.006	−6.554	***	−0.006	−6.653	***	−0.006	−6.643	***
BIG4	0.216	37.480	***	0.217	37.430	***	0.218	38.780	***
BUSY	0.017	1.908	*	0.017	1.880	*	0.017	1.817	*
OPINION	0.040	2.236	**	0.037	2.062	**	0.037	2.058	**
Adj. R2	0.484			0.488			0.486
IND/YEARDUM	yes			yes			yes
N	6653			6653			6653

Note: ***, **, and * indicate the results of robustness tests with one-way cluster-robust standard errors (adjusted for degrees of freedom) with Year as the dimension and are statistically significant at the 1%, 5%, and 10% levels, respectively. All models in the table are estimated by Pooled OLS.

summarizes the estimation results for all observations in our analysis. First, the coefficient on DPO, the main variable, is estimated with positive values and is statistically significant at the 1% level for Models 2 and 3. These results are consistent with H1a, which suggests that parent companies require a high level of audit effort from their subsidiaries’ auditors to improve their subsidiaries’ accounting information quality and prevent increasing the cost of capital due to agency costs. Furthermore, the coefficient of the cross term between DPO and MLS (DMLS, O234) is −0.151 and −0.375 for Models 2 and 3, respectively, which are both statistically significant at the 5% level and better. These results are consistent with H2. The demand for high-quality accounting information is smaller because MLSs mitigate information asymmetries between the parent or the management of the listed subsidiary through private channels.

Next, looking at DMLS, O234, and OWN1 as variables related to ownership structure, the coefficients for each are estimated with negative values and are statistically significant for Models 1 through 3. These suggest that concentrated ownership by a small number of shareholders reduces audit fees. The coefficients on DFOREIGN, indicating ownership by foreign investors, and DMBOWN, indicating ownership by main banks, are estimated with negative values in Models 1 through 3.

SIZE is a variable indicating firm size. Size is the most important determinant of audit fees and is expected to be positively related to audit fees [65]. Among all of Models 1 through 3, the coefficients are estimated to be positive and statistically significant at the 1% level.

In general, the more complex the client’s business, the more difficult and time-consuming the audit is expected to be [65,67]. Therefore, the higher the complexity of the client’s business, the higher the level of audit fees is expected to be. The ratio of foreign sales (INTSALES), the ratio of trade receivables and inventories to total assets (INVREC), and the number of business segments (NBS) are all variables that indicate audit complexity. In Model 1, the coefficients on INTSALES, INVREC, and NBS are estimated at positive values of 0.188, 0.198, and 0.036, respectively, which are all significant at the 1% level. These are also observed for Models 2 and 3.

ROA, LOSS, BTM, LEVERAGE, and CURRATIO represent the degree of potential loss the auditor may incur if the client fails. In general, the worse the client’s financial condition and performance, the greater the auditor’s risk and the higher the expected audit fees [18,65]. The coefficient of ROA, which indicates performance, is −0.102 and is statistically significant at the 1% level in Model 1. Further, the coefficient of LOSS is 0.049 in Model 1, which is statistically significant at the 1% level. The coefficient of BTM, which indicates growth potential, is −0.025 in Model 1, statistically significant at the 1% level. These results indicate that audit fees are higher for firms with poorer performance and growth potential. The coefficient of LEVERAGE, which indicates the debt ratio, is 0.160 in Model 1 and is statistically significant at the 1% level. Additionally, the coefficient of CURRATIO, which indicates the current ratio, is −0.006 in Model 1, which is statistically significant at the 1% level. These results imply that audit fees are higher for firms in worse financial conditions. These estimation results are similar for Models 2 and 3.

Finally, the coefficient on BIG4 indicating the audit premium by big audit firms is estimated positively, as expected, and is statistically significant at the 1% level for Models 1 through 3. The coefficient of BUSY, indicating the busy audit season, is estimated with a positive value, as expected, and is statistically significant at the 10% level in Models 1 through 3. The coefficient of OPINION, indicating that it is not an unqualified opinion, is estimated with a positive value, as expected, and is statistically significant at the 5% level in Models 1 to 3.

In addition, we also perform the following tests to check the robustness of these results. First, the impact may differ depending on the percentage of PO. Therefore, we performed the same tests for the percentage of PO. Second, 2011 was when Japan experienced a major earthquake and a nuclear power plant accident. Therefore, we estimated Equation (1) annually to check the stability of the coefficients. In addition, although we did not address outliers in the main analysis, we also estimated following [68] to check whether outliers affect the results. These results are consistent with the main results.

5.2. Additional Tests

Prior research suggests that audited firms are not randomly assigned to auditors and may be influenced by client characteristics [69,70]. In particular, the firms analyzed in this paper have two distinctive characteristics: (1) the presence of controlling shareholders and (2) the presence of MLSs. These characteristics of the client firms may have led to the selection of auditors. This section addresses these endogeneity issues by testing only firms with controlling shareholders and using propensity scores for MLSs.

5.2.1. Controlling Shareholders and Auditor Selection

Although the main analysis tested the relationship between PO and audit fees, there may be differences in corporate management strategies between firms with controlling shareholders and those without. For example, a previous study [71] on the relationship between institutional investor ownership and performance in Japanese firms found that the larger the institutional investor ownership ratio, the greater the growth potential and profitability of the firm. Therefore, to address the endogeneity issue due to controlling shareholders, we checked the robustness of the results by testing additionally only for firms with controlling shareholders.

Table 5. Firms with controlling shareholders.

	With Control Shareholders
	Model 4			Model 5			Model 6
	Coeff.	t		Coeff.	t		Coeff.	t
const	0.985	22.530	***	1.031	18.620	***	1.014	19.550	***
DPO				0.080	5.033	***	0.102	3.071	***
DPO_DMLS				−0.160	−7.922	***
DPO_O234							−0.496	−3.342	***
DMLS				0.002	0.268
O234							0.011	0.314
OWN1	0.062	2.602	***	−0.040	−1.092		−0.015	−0.390
DFOREIGN	−0.174	−22.480	***	−0.175	−21.860	***	−0.174	−22.460	***
DMBOWN	−0.108	−17.430	***	−0.112	−17.640	***	−0.110	−16.770	***
SIZE	0.216	50.890	***	0.215	44.550	***	0.215	44.940	***
INTSALES	0.023	1.261		0.020	1.090		0.020	1.105
INVREC	0.159	4.712	***	0.151	4.212	***	0.156	4.468	***
NBS	0.056	6.994	***	0.055	6.070	***	0.057	6.859	***
ROA	−0.071	−1.167		−0.073	−1.169		−0.069	−1.108
LOSS	0.041	2.387	**	0.042	2.376	**	0.042	2.392	**
BTM	−0.028	−4.228	***	−0.030	−4.519	***	−0.029	−4.438	***
LEVERAGE	0.117	3.528	***	0.113	3.406	***	0.113	3.339	***
CURRATIO	−0.008	−3.857	***	−0.007	−3.735	***	−0.008	−3.766	***
BIG4	0.247	33.770	***	0.247	34.260	***	0.248	36.930	***
BUSY	0.028	2.282	**	0.030	2.407	**	0.029	2.279	**
OPINION	0.052	2.595	***	0.050	2.446	**	0.051	2.493	**
Adj. R2	0.514			0.519			0.516
IND/YEARDUM	yes			yes			yes
N	3821			3821			3821

Note: *** and ** indicate the results of robustness tests with one-way cluster-robust standard errors (adjusted for degrees of freedom) with Year as the dimension and are statistically significant at the 1%, 5%, and 10% levels, respectively. All models in the table are estimated by Pooled OLS.

summarizes the estimation results for firms with controlling shareholders only. We identify firms with a controlling shareholder as those where the largest shareholder owns more than 20% of the shares (3203 firm-years), besides those with a parent company (618 firm-years).

Looking at Model 5 in Table 5, the coefficient of DPO is 0.080, a statistically significant positive value at the 1% level. This result is similar for Model 6 and consistent with H1a, suggesting that parent companies require a higher audit effort from auditors to improve the quality of accounting information of their listed subsidiaries, resulting in higher audit fees. What differentiates parent companies from other controlling shareholders is that Japanese parent companies do not merely own shares in their subsidiaries, but as part of their management strategy, they spin off growing businesses and establish them as subsidiaries [20]. The results in Table 5 suggest that parent companies demand higher audit efforts from their subsidiaries’ auditors than other controlling shareholders.

Moreover, the coefficient of the cross term between DPO and DMLS is −0.160, a statistically significant negative value at the 1% level. This is also similar for Model 6 and consistent with H2, suggesting that MLSs mitigate information asymmetries through private channels and interest alignment, thus reducing the parent company’s demand for higher-quality audits.

We also analyzed firms with the largest shareholder holding more than 10% and 30%. The results are the same as in Table 5.

5.2.2. MLS and Auditor Selection

The ownership structure of a firm is influenced by various characteristics, such as the firm’s size, the proportion of fixed assets, and profitability [61,72]. These various characteristics of firms may influence auditor selection. Therefore, this section addresses the issue of endogeneity arising from the presence of MLSs.

Previous studies on concentrated ownership and audits often use the instrumental variable method [12,62]. However, in this paper, we could not obtain sufficient observations by year and industry to create a valid industry average to be used in creating instrumental variables for the estimation [12,62]. Indeed, 11 of the 24 industries in our sample have an average number of firms of 10 or less for the period.

On the other hand, endogeneity due to selection bias can be reduced by propensity score matching (PSM), which estimates propensity scores and matches observations from the treatment and control groups with the smallest difference in estimated propensity scores [73]. Therefore, prior studies on the relationship between MLS and audit fees estimate propensity scores for firms with MLSs and match observations by year and industry [12]. However, because this paper focuses on emerging firms, it is not possible to obtain sufficient observations for each year and every industry sector, making it impossible to perform appropriate matching. Therefore, this paper addresses the endogeneity problem by performing inverse probability weighting estimation (IPW) [74], which can be applied even when there are no appropriate pairs to be matched.

IPW is conducted in the following two steps. First, a propensity score regarding the MLS is estimated by using a probit model. Next, Equation (1) is estimated by a weighted least squares method using weights that are the inverse of the propensity score estimated by the probit model for firms with an MLS and the inverse of (1–propensity score) for firms without an MLS.

To estimate the propensity score, we first estimate the following probit model that identifies the determinants of the presence of MLS following previous studies [12,61].

$$\begin{array}{l}\mathrm{DMLS}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{SIZE}+{\mathsf{\beta }}_2\mathrm{AGE}+{\mathsf{\beta }}_3\mathrm{LEVERAGE}+{\mathsf{\beta }}_4\mathrm{FCF}+{\mathsf{\beta }}_5\mathrm{PPE}\\ +\mathrm{Industry}\mathrm{Dummies}+\mathrm{Year}\mathrm{Dummies}+\mathsf{\mu }\end{array}$$

(2)

where AGE is the natural logarithm of the number of years the firm has been in existence, FCF is the variable of the firm’s free cash flow divided by total assets at the beginning of the year, and PPE is the variable of tangible fixed assets divided by total assets. The estimation results for Equation (2) are summarized in

Table 6. Estimation of propensity score.

	Full Sample			With Control Shareholders
	Model 7			Model 8
	Coeff.	z		Coeff.	z
const	0.260	1.526		0.078	0.362
SIZE	0.030	1.765	*	0.059	2.751	***
LEVERAGE	−0.061	−0.800		−0.242	−2.460	**
AGE	−0.098	−2.964	***	−0.049	−1.190
FCF	0.200	1.911	*	0.046	0.451
PPE	−0.149	−1.587		−0.117	−0.997
McFadden R2	0.034			0.039
IND/YEARDUM	yes			yes
N	6653			3821

Note: ***, **, and * indicate the results of robustness tests using the Quasi-Maximum Likelihood Estimator, which are statistically significant at the 1%, 5%, and 10% levels, respectively. All models in the table are estimated by Probit model.

. The estimation results are generally similar to previous studies [61], with smaller AGE and larger FCF tending to be the present MLSs in the full sample. Note that coefficient of SIZE is estimated with positive values in this paper. Furthermore, in the subsample of firms with controlling shareholders only (with control shareholders), the larger the SIZE, the smaller the LEVERAGE, and the more MLS tends to be present.

Table 7. Estimation results using IPW.

	Full Sample						With Control Shareholders
	Model 9			Model 10			Model 11			Model 12
	Coeff.	t		Coeff.	t		Coeff.	t		Coeff.	t
const	1.139	23.300	***	1.145	23.080	***	1.038	15.890	***	1.029	15.270	***
DPO	0.076	3.795	***	0.070	2.255	**	0.066	2.957	***	0.096	2.848	***
DPO_DMLS	−0.158	−5.081	***				−0.163	−5.142	***
DPO_O234				−0.360	−2.317	**				−0.530	−3.279	***
DMLS	−0.016	−2.214	**				−0.002	−0.154
O234				−0.133	−2.903	***				−0.019	−0.309
OWN1	−0.102	−2.708	***	−0.091	−2.377	**	−0.042	−0.632		−0.028	−0.403
DFOREIGN	−0.158	−17.910	***	−0.156	−17.710	***	−0.178	−15.050	***	−0.176	−15.010	***
DMBOWN	−0.094	−10.900	***	−0.094	−10.890	***	−0.118	−10.430	***	−0.117	−10.260	***
SIZE	0.205	39.170	***	0.206	39.220	***	0.215	33.530	***	0.216	33.450	***
INTSALES	0.174	6.144	***	0.176	6.140	***	0.002	0.056		0.002	0.064
INVREC	0.206	7.339	***	0.208	7.384	***	0.136	3.587	***	0.142	3.750	***
NBS	0.036	3.798	***	0.035	3.769	***	0.057	4.076	***	0.058	4.176	***
ROA	−0.090	−2.164	**	−0.087	−2.084	**	−0.079	−1.246		−0.076	−1.206
LOSS	0.049	4.107	***	0.048	3.994	***	0.052	3.279	***	0.052	3.244	***
BTM	−0.026	−4.712	***	−0.026	−4.655	***	−0.026	−3.447	***	−0.025	−3.284	***
LEVERAGE	0.160	6.613	***	0.162	6.689	***	0.105	2.805	***	0.104	2.759	***
CURRATIO	−0.006	−5.455	***	−0.006	−5.545	***	−0.008	−3.792	***	−0.008	−3.796	***
BIG4	0.221	27.010	***	0.222	27.100	***	0.249	22.220	***	0.250	22.220	***
BUSY	0.013	1.671	*	0.013	1.705	*	0.027	2.749	***	0.027	2.694	***
OPINION	0.037	3.364	***	0.037	3.322	***	0.049	3.461	***	0.049	3.487	***
Adj. R2	0.495			0.493			0.516			0.513
IND/YEARDUM	yes			yes			yes			yes
N	6652			6652			3821			3821

Note: ***, **, and * indicate the results of robustness tests with White’s standard errors (adjusted for degrees of freedom), which are statistically significant at the 1%, 5%, and 10% levels, respectively. IPW performs a weighted least squares (WLS) method using the inverse of the propensity score (PS) estimated in Table 6 ([DMLS/PS] + {[1-DMLS]/[1-PS]}) as the weight. In Models 9 and 10, one observation is excluded from the analysis because its propensity score is estimated to be 0 and its weight cannot be calculated.

summarizes the results of Equation (1) by IPW using the propensity score estimated in Equation (2). In Model 9, using the full sample, the coefficient on DPO is estimated to be positive and statistically significant. This trend is also observed in Model 10, consistent with H1a, which predicts that audit fees are higher for listed subsidiaries. The coefficient of DPO is positive and statistically significant at the 1% level in Models 11 and 12, which are estimated for the subsample of firms with a controlling shareholder. These results indicate that the audit fees are higher when a parent company is present.

Furthermore, in Model 9, the coefficient on the intersection term between DPO and DMLS is estimated at a negative value and is statistically significant at the 1% level. In Model 10, the coefficient on the intersection term between DPO and O234 is also statistically significant and negative at the 5% level. From here, the increase in audit fees of listed subsidiaries due to the parent company is mitigated by the presence of the MLS, which is consistent with H2. This trend is also observed in Models 11 and 12.

6. Discussion and Conclusions

Previous studies on ownership structure (controlling shareholders and MLSs) and audit fees [12] stand on the agency costs caused by PP conflicts and the audit risks faced by auditors. However, in countries where stakeholder-oriented corporate governance is adopted, accounting information is more susceptible to political interference, information asymmetries among stakeholders tend to be mitigated through private communication, and the litigation risk is smaller [15]. As a result, in stakeholder-oriented countries, the relationship between ownership structure and audit fees may change depending on the demands of controlling shareholders for auditors [13], as auditors’ attention shifts from litigation to client demands [17]. Therefore, it is necessary to examine controlling shareholders’ demands for audit services and audit fees in order to understand the characteristics of external audits, especially in firms that adopt stakeholder-oriented corporate governance. To address this research gap, this paper focuses on Japanese-listed subsidiaries that adopt stakeholder-oriented corporate governance and analyzes the impact of direct ownership by the parent company and MLSs on audit fees.

In Japan, parent companies tend to spin off growth businesses as subsidiaries [20] and to manage listed subsidiaries on a long-term horizon to maximize sales growth rates [21,22]. As a result, PP conflicts between parent and noncontrolling shareholders are less severe in listed subsidiaries [21], and parent companies may have incentives to reduce their cost of capital by disclosing information to widely dispersed noncontrolling shareholders [23]. Therefore, audit fees are likely to be higher for listed subsidiaries. On the other hand, if the costs associated with disclosure exceed the benefits derived from it, the parent company may be able to enjoy private benefits from insufficient information. Therefore, in theory, audit fees can be lower for listed subsidiaries. As discussed above, the demands for audit services by parent companies as controlling shareholders of listed subsidiaries is an empirical issue.

The results of this paper indicate that audit fees are higher for listed subsidiaries (i.e., when owned by the parent company). It is also shown that audit fees are smaller when MLSs are present. This is consistent with expectations from previous studies [25,59].

Our research contributes to both research on auditing or stakeholder-oriented corporate governance and practice on sustainable economic development. First, for the audit and corporate governance studies, we showed that controlling shareholders’ demands of auditors affect audit fees. We also show that MLSs affect audit fees even under stakeholder-oriented corporate governance. The relationship between ownership structure and audit fees is influenced by corporate governance orientation [13]. Moreover, the ownership structure of a firm does not necessarily consist of controlling shareholders and widely dispersed non-controlling shareholders; in fact, MLSs exist, and the existence of MLSs affects audit fees [1,2,3,4,13]. The results of this paper reveal the impact of controlling shareholder parent company demands and MLSs on audits in firms that adopt stakeholder-oriented corporate governance, indicating that controlling shareholders’ demands for audit services and the existence of MLSs are one of the determinants of audit fees in firms that adopt stakeholder-oriented corporate governance.

Auditing is considered to be a monitoring mechanism different from the corporate ownership structure [26], reducing information asymmetry by guaranteeing higher accounting information quality and ensuring that all shareholders are treated equally and agency problems can be mitigated [13,25,38]. Because key stakeholders tend to resolve information asymmetry problems through insider communication, the level of audit effort is affected not only by the audit risk from PP conflicts [12], but also by the demands of key stakeholders where stakeholder-oriented corporate governance is adopted. Therefore, by considering audit systems that are useful to other stakeholders, such as debt holders, it may be possible to design more effective corporate governance systems that combine ownership structures and audit systems. In conclusion, this paper provides suggestions for making effective corporate governance in a stakeholder-oriented system, which will contribute to sustainable growth and economic development [27]. In this regard, external auditors also provide various non-audit services such as sustainability reporting assurance and sustainability management [30]. The results of this paper suggest that in Japan, where stakeholder-oriented corporate governance is adopted, audit practice is influenced by the demands of controlling shareholders and MLSs when they exist. Therefore, services related to sustainability reporting may also be affected by controlling shareholders and MLSs. Therefore, this study is useful not only for auditors but also for standard setters who consider the audit system and its relationship with non-audit services.

Finally, we identify the remaining challenges to our study. First, our study leaves room for improvement in the estimation model used to test our hypotheses. For example, a recent study has shown that client sentiment [75] and other factors affect audit fees. In addition, because this paper focuses on the existence of parent firms, it is not possible to use a fixed-effects model. These may have influenced the estimation results in this paper by omitted variables bias. Second, recent research on auditing often discusses audit fees not only in terms of audit work, but also in conjunction with non-audit work, auditor rotation, and other social relationships [76,77,78,79,80,81]. Third, our study does not consider ultimate ownership. Due to the limitations of the database, our variable on ownership is a very simplified variable. This paper does not consider factors other than these ownership structures that affect audit fees, which is a topic for future work.