The 2006 dissolution of PwC ChuoAoyama significantly changed market share composition of Japanese audit firms which marked the transition from Big 4 period to Big 3 period. This study aims to investigate audit market pricing competitiveness between Big N and non-Big N auditors using a sample of Japanese firms listed in the First Section of Tokyo Stock Exchange during the transition from Big 4 period (2004-2005) to Big 3 period (2006-2011). This paper analyzes audit market pricing competitiveness between Big N and non-Big N auditors by employing panel fixed effects multivariate regression with audit fee as the dependent variable and interaction between audit fee premium and client segment size as variable of interest while controlling for other variables affecting audit fee. The empirical results indicate a non-competitive audit pricing market between Big N and non-Big N auditors where Big N auditors earn increasingly higher audit fee as client segment size becomes larger.
In April 2005, one client of ChuoAoyama―the PwC affiliated audit firm in Japan―committed the then largest accounting fraud in Japan [
A number of influential regulators and organizations have expressed concern over the adverse effect on audit market competition in an event of a hypothetical scenario if one of the Big 4 auditors experienced an unexpected market exit [
Adverse effects of a highly concentrated audit market where a few large auditors have large market power include: limited incentives for auditors to innovate and provide superior audit quality, large audit firms become too-big-to-fail rendering audit regulations to be ineffective and higher audit prices without corresponding increase in audit quality [
Following the market exit of both PwC Misuzu and ChuoAoyama, Japanese Big N audit market has a much higher market share concentration (over 90% market share) for the large public firms’ market segment compared to other developed economies, as illustrated in
This study is interested in investigating audit pricing competitiveness between Big N and non-Big N auditors in Japan. Although market share of Japanese Big N
firms (based on average client numbers per fiscal year) in the Big 3 period (2006-2011) declined by 4.45% compared to the prior Big 4 period (2004-2005); non-Big N firms’ market share (based on average client numbers per fiscal year) increased by 21.42% in the same period (based on author’s calculation). In addition, prior empirical research investigating predictors of audit quality in Japan find that auditor size is not associated with audit quality [
Audit market pricing competitiveness between large (Big N) and small (non-Big N) audit firms is inferred as a function of Big N fee premium and audit clients’ segment size on audit fee. Big N audit fee premium is defined as additional audit fee paid by clients of one of the Big N firms that the clients otherwise would not pay to non-Big N auditors [
The estimation results of the multivariate panel fixed effect regression models indicate a non-competitively priced audit market between Big N and non-Big N auditors as Big N firms receive a disproportionally higher audit fee premium as the client size increases. In addition, this study investigates whether the Big N audit fee premium is differentially affected by the transition from Big 4 period (2004-2006) to Big 3 period (2007-2011) following the demise of PwC ChuoAoyama that significantly changed the Japanese audit market structure (refer to
A number of sensitivity analyses (auditor self-selection control, year-by-year analysis and reduced sample analysis) are performed to ensure the robustness of the audit fee regression models. To the best of my knowledge, there has been no empirical research investigating audit pricing competitiveness between Big N and non-Big N firms during the transition from Big 4 period to Big 3 period in Japan. This paper presents important empirical evidence for the Japanese and international accounting standard setters and regulators to consider when discussing the potential implication future policies regarding regulation or deregulation of competition in the audit market. Although market competition regulators have passed mergers proposal of large auditors in the past; regulators should carefully consider the adverse effects of future merger proposals or potential demise of existing Big N auditors on audit market competition.
The remainder of the paper is organized as follows. In Section 2, prior literatures on international and Japanese audit market structure, market competition, and audit fee premium are discussed. Section 3 develops hypotheses related to audit pricing competition between Big N and Non-Big N Auditors and audit pricing competition among individual firms at the industry level. Section 4 discusses audit fee regression models related to the hypotheses, control variables, industry level audit market concentration measures, and sample selection process. In Section 5, descriptive statistics and estimation results of the multivariate panel fixed effect regressions models related to the hypotheses are evaluated. Section 6 presents the results of the sensitivity analyses and Section 7 concludes the paper.
Audit service market for public companies has three characteristics that differentiate it from other professional services market: capital market transparency, mandated demand, and concentrated supply [
Auditors that have established brands and large scale of operation such as Big N firms are more likely to charge higher audit fee and enjoy positive audit fee premium over non-Big N auditors. Big N audit fee premium is defined as additional audit fee paid by clients of one of the Big N firms that the clients otherwise would not pay to non-Big N auditors [
Japanese audit market changed dramatically in 2006 following the dissolution of PwC ChuoAoyama. The figure tracks the market share trend among the Japanese large audit firms from 2004 to 2011.
PwC ChuoAoyama’s audit failure in the Kanebo fraud in 2006 forced the Japanese Financial Services Agency (FSA) to suspend ChuoAoyama’s operation for two months in May 2006. Following the suspension, PwC International divided the firm into two separate entities to salvage the reputation of its Japanese affiliate following the scandal. The first firm, Misuzu, is a rebranded ChuoAoyama. Eventually, PwC Misuzu is disbanded in July 2007 due to another accounting fraud committed by one of its major client, Nikko Cordial [
Skinner and Srinivasan (2012) examine the effect of the scandal on Japanese auditor reputation by examining client switching following the uncovering of the fraud [
to make the transition. Approximately 75 percent of the former CPAs, staffs and clients of PwC ChuoAoyama transferred to the other Big 3 firms [
The size and industry diversity of audit clients requires auditors to tailor their audit services to meet the varied demand of their clients. Audit market pricing competitiveness between large (Big N) and small (non-Big N) audit firms is a function of the audit clients’ segment size. Audit market serves two different market segments with specific market characteristics: audit services for listed and large companies and audit services for small and medium-sized companies [
Although market share of Japanese Big N firms (based on average client numbers per fiscal year) in the Big 3 period (2006-2011) declined by 4.45% compared to the prior Big 4 period (2004-2005); non-Big N firms’ market share (based on average client numbers per fiscal year) increased by 21.42% in the same period (based on author’s calculation). Yoshida (2008) argues that audit quality measured by discretionary accruals is not strongly associated with auditors’ size in Japan due to low litigation risk and inadequate internal control of the Big N auditors [
Empirical evidence on the state of competition in the audit market can be further inferred from the difference in the “average cost residuals” or audit fee premium between Big N and non-Big N for both small and large client market segment [
When client segment size is taken into consideration, pricing between Big N and non-Big N auditors is regarded to be competitive if Big N firms earn consistent fee premium through audit service differentiation that does not vary with the size of clients [
Thus, the argument for a non-competitive audit market between Big N and non-Big N auditors can be expressed in the following alternative H1:
H1: Audit fee is positively associated with the interaction variable between Big N fee premium and client segment size, other things being equal.
The transition from Big 4 to the Big 3 period in the Japanese market is indicated with the dissolution of PwC ChuoAoyama into a smaller PwC Aarata. Around a quarter of former ChuoAoyama’s clients switched to new auditors [
Large clients market segment, such as multinational companies with many international subsidiaries, demand audit service which can only be provided by large auditors with experience auditing large clients and extensive international network of affiliated firms. Thus, the unexpected market exit of PwC ChuoAoyama is more likely to disproportionately affect the large audit clients segment due to the short term disruption of Big N market supply. Chen et al. (2007) argue that the audit fee premium provides a measure of market power or competition in a market where the dynamics of audit supply and demand have not reached equilibrium, such as during the transition from Big 4 to Big 3 period [
H2: Audit fee is positively associated with the interaction variable between Big N fee premium and audit market transition from Big 4 period (2004-2005) to Big 3 period (2006-2011), other things being equal.
Audit fee premium is defined as the difference between what a client with an incumbent Big N auditor would pay for an equivalent non-Big N audit [
Audit fee is a function of audit unit price multiplied by the quantity of audit services, thus an audit fee model that explain audit competition should control for the determinants of audit quantity and price [
This study measures audit market pricing competitiveness across markets segmented by client size and industry groups. Following Hamilton et al. (2008), clients market segment is measured by the ClientSeg dummy variable, where it takes the value of one if the client’s median total assets belong to the upper half (>50th percentile) of the industry-year sample, and zero otherwise [
Audit fee regression model in Equation (1) is a function of audit fee premium (FeePrem) while controlling for clients’, auditors’ and audit engagements’ attributes. Panel data regression estimate that considers both cross sectional (companies) and time series (fiscal year) dimensions of the data is employed. Panel data estimated model can better cope with the problem of unobserved time-invariant heterogeneity in cross-sectional models [
Hypothesis 1 described in Section 3.1 argues that pricing between Big N and non-Big N auditors is regarded to be competitive if Big N firms earn consistent fee premium through audit service differentiation that does not vary with the size of clients [
Year and industry dummy variables are employed so that the regression estimates results are less likely to be affected by contemporaneous changes in regulatory measures and other omitted time and industry level variables that affect audit pricing [
A F i , t = α 1 F e e P r e m i , t + α 2 B i g 3 P e r i , t + α 3 C l i e n t S e g i , t + α 4 F e e P r e m i , t × C l i e n t S e g i , t + ∑ α j C o n t r o l s j , i , t + ε i , t (1)
where:
A F i , t = natural log of total audit fee paid by client i at time t, which consists of fee paid to the client’s external auditor for financial statement audit of parent company and consolidated subsidiaries.
F e e P r e m i , t = dummy variable equals to 1 if client i at time t is audited by one of the Japanese auditors affiliated with the global Big 4 audit firms networks (Deloitte Touche Tohmatsu, E&Y Shin Nihon, KPMG AZSA, PwC Aarata, PwC Chuo Aoyama, and PwC Misuzu), and 0 otherwise.
B i g 3 P e r i , t = dummy variable equals to 1 if the audit client i at time t took place during the Big 3 period (2006-2011), and 0 otherwise.
C l i e n t S e g i , t = dummy variable equals to 1 if the median total assets of client i at time t belong to the upper half (>50th percentile) of the industry-year sample, and 0 otherwise.
C o n t r o l s j , i , t = audit fee control variable j for client i at time t as listed in
The pre and post treatment approach is a subset of difference-in-difference (DiD) analysis commonly used in empirical research to estimate the effects of certain policy interventions and policy changes that affect the population groups in a different way [
The treatment and control group is denoted by the auditor size dummy variable (FeePrem). The control group is represented by client of non-Big N auditors that is assigned with dummy value of 0 for the FeePrem variable. Accordingly, the treatment group consists of client of Big N auditors whose audit market is directly affected by the transition from Big 4 to Big 3 period where its FeePrem variable is assigned with the value of 1.
The pre and post treatment variable is defined as the Big N period dummy variable (Big3Per). Pre-treatment period is represented by the Big3Per value of 0 where it represents fiscal year 2004 and 2005 during which the audit market is dominated by Big 4 firms before the audit market transition. Post-treatment period represents time period after Big 3 audit firms dominate the audit market from fiscal year 2006 to 2011 where it is denoted with the Big3Per value of 1. The interaction between FeePrem and Big3Per measures the differences between audit fee paid by clients of non-Big N firms (control group) and clients of Big N firms (treatment group) following the audit market transition from Big 4 period (pre-treatment) to Big 3 period (post-treatment).
The regression estimate of the difference-in-difference interaction variable between FeePrem and Big3Per (α4 coefficient) in the following Equation (2) captures how the association between audit fee and Big N fee premium is differentially affected by the audit market transition from Big 4 to Big 3 period. Hypothesis 2 described in Section 3.2 argues that audit market becomes less competitively priced if audit fee is positively associated with Big N audit fee premium following the transition from Big 4 to Big 3 period. Thus, a statistically significant positive difference-in-difference α4 coefficient indicates a less competitive audit pricing market following Big 4 to Big 3 transition where Big N fee premium is increasing as Big N audit market becomes more concentrated in the Big 3 period.
We modify prior Equation (1) by replacing the ClientSeg interaction variable with the treatment period dummy variable (Big3Per) to estimate the difference-in-difference coefficient (α4 coefficient), as shown in the following Equation 2 (variables are defined in
A F i , t = α 1 F e e P r e m i , t + α 2 B i g 3 P e r i , t + α 3 C l i e n t S e g i , t + α 4 F e e P r e m i , t × B i g 3 P e r i , t + ∑ α j C o n t r o l s j , i , t + ε i , t (2)
where:
A F i , t = natural log of total audit fee paid by client i at time t, which consists of fee paid to the client’s external auditor for financial statement audit of parent company and consolidated subsidiaries.
F e e P r e m i , t = dummy variable equals to 1 if client i at time t is audited by one of the Japanese auditors affiliated with the global Big 4 audit firms networks (Deloitte Touche Tohmatsu, E&Y Shin Nihon, KPMG AZSA, PwC Aarata, PwC Chuo Aoyama, and PwC Misuzu), and 0 otherwise.
B i g 3 P e r i , t = dummy variable equals to 1 if the audit client i at time t took place during the Big 3 period (2006-2011), and 0 otherwise.
C l i e n t S e g i , t = dummy variable equals to 1 if the median total assets of client i at time t belong to the upper half (>50th percentile) of the industry-year sample, and 0 otherwise.
C o n t r o l s j , i , t = audit fee control variable j for client i at time t as listed in
Consistent with prior audit fee studies; client’s attributes (client size, business complexity, risk, and accounting standards), auditor’s attributes (audit staff number, audit tenure period, non-audit fee and industry specialization), and audit engagement’s attributes (audit opinion, audit quality, client’s bargaining power, auditor industry dominance, competitor distance, and exogenous events) are controlled in the audit fee regression models [
The following client’s attributes variables are controlled: client size, business complexity, risk, and accounting standards. Client size effect explains most of the variation in audit fees between clients [
The following auditor’s attributes determinants on audit fee are controlled: audit staff number, audit tenure period, non-audit fee and industry specialization. The number of audit staff working on the audit engagement is employed to control for one of the major determinants of audit fee and audit effort. Carson et al. (2014) argue that the observed increase of audit fees in Australian from 2000-2011 might be driven by higher audit effort that is driven by the global financial crisis and more stringent regulations [
AISpec variable is included as a continuous measure of audit industry specialization that is calculated as the auditor market share within the industry-year (based on client number). Following Cahan et al. (2008), the auditor market dominance (DOMN variable) measures auditors’ dispersion in each of the industries classification of the Tokyo Stock Exchange (TSE) New Industry Code [
Audit engagement’s attributes are controlled by including audit opinion (AOP variable), audit quality (ACC variable), the client’s bargaining power (POW variable) and competitor distance (DIST variable) variables in the audit fee model. AOP represents the audit opinion dummy variable where a value of one if an auditee receives a modified audit opinion or worse, and zero if the auditee receives an unqualified audit opinion with additional notes or better. To complement AOP, discretionary accruals (ACC) are employed as a continuous measure of audit quality. ACC represents absolute value of total discretionary accruals estimated using the following modified Jones (1991) model that is also adopted by Skinner and Srinivasan (2012) [
POW variable is associated with the relative client’s bargaining power by measuring the relative size of the client’s audit fee relative to the sum of the auditor’s total audit fee received from all its clients in the industry [
Numan and Willekens (2012) find that audit fee is higher when the industry market share distance to its next closest competitor increases [
Lastly, the effect of exogenous global financial crisis and major change in regulations that might affect audit fee are controlled. A dummy variable GFC takes the value of 1 to denote fiscal year 2008 as the period of the global financial crisis. Major regulation changes during the observation periods that might increase audit fee in Japan. The amendment of the Financial Instruments and Exchange Act (FIEA) introduced two major changes to financial reporting that have the potential to increase audit fee [
The variables used in audit fee premium (Equation (1) and Equation (2)) regression models are summarized in
Japanese companies publicly listed in the First Section of the Tokyo Stock Exchange
Description | Variable | Definition |
---|---|---|
Dependent variable | AF | natural log of total audit fee, which consists of fee paid to external auditors for financial statement audit of the parent company and consolidated subsidiaries. |
Variables of Interest | ||
Audit Fee Premium Model: Equations ((1) and (2)) | FeePrem | dummy variable equals to 1 if the client is audited by one of the Japanese auditors affiliated with the global Big 4 audit firms networks (Deloitte Touche Tohmatsu, E&Y Shin Nihon, KPMG AZSA, PwC Aarata, PwC Chuo Aoyama, and PwC Misuzu), and 0 otherwise. |
ClientSeg | dummy variable equals to 1 if the median total assets of the client belong to the upper half (>50th percentile) of the industry-year sample, and 0 otherwise. | |
Big3Per | dummy variable equals to 1 if the audit took place during the Big 3 period (2006-2011), and 0 otherwise. | |
Control Variables | ||
Description | Variable | Definition |
Client’s attributes―size | TA | natural log of clients’ total assets. |
Client’s attributes―size | IndPTA | ratio of the client’s total assets to total assets of companies within the industry-year. |
Client’s attributes―complexity | SUBS | natural log of number of consolidated subsidiaries (if a company has zero subsidiaries, it is re-coded as 1 before taking the natural log). |
Client’s attributes―complexity | FORN | ratio of the client’s overseas sales to net sales. |
Client’s attributes―risk | ROI | ratio of the client’s net income to total assets. |
Client’s attributes―risk | LIQ | ratio of the client’s current assets (less inventories) to current liabilities. |
Client’s attributes―risk | LEV | ratio of the client’s total liabilities to total equity. |
Client’s attributes―risk | LOSS | dummy variable equal to 1 if the client incurred a net loss in the previous fiscal year, and 0 otherwise. |
Client’s attributes―accounting standards | GAAP | dummy variable equal to 1 if the client is a SEC registrant or an IFRS adopter, and 0 otherwise. |
Auditor’s attributes | TEAM | natural log of number of CPAs, junior accountants and other staffs employed in the audit engagement (excluding engagement partners). |
Auditor’s attributes | TENR | number of years an auditee has hired its current auditor. |
Auditor’s attributes | NAF | natural log of non-audit fee paid by the client to its current year auditor. |
Auditor’s attributes | AISpec | auditor’s industry specialization variable which measures auditor market share within the industry-year (based on client number). |
Auditor’s attributes | DOMN | industry dispersion measure which measures market dominance of an auditor as it obtained more clients in the industry. |
Audit engagement’s attributes | AOP | dummy variable equal to 1 if the client received a modified audit opinion or worse, and 0 if the client received an unqualified audit opinion with additional notes or better. |
Audit engagement’s attributes | ACC | audit quality measure, measured by the absolute value of total discretionary accruals estimated using the modified Jones (1991) model. |
Audit engagement’s attributes | POW | client’s bargaining power with its auditor in the industry, calculated by the relative size of the client’s audit fee divided by the sum of the auditor’s total audit fee received from all its clients in the industry. |
Audit engagement’s attributes | DIST | competitor distance, measured by the smallest absolute audit market share (based on client number) difference between audit leader and its closest competitor within an industry. |
Audit engagement’s attributes | GFC | dummy variable equal to 1 if the audit took place in fiscal year 2008 to control for the effect of global financial crisis, and 0 otherwise. |
Audit engagement’s attributes | REG | dummy variable equal to 1 if the audit took place in fiscal year 2007 to control for the amendment of the Financial Instruments and Exchange Act (FIEA) that is effective in fiscal year 2008 (refer to |
(TSE) from fiscal year 2004 to 2011 are employed as the sample of this study. The observation period for the Big 3 period is limited to fiscal year 2011 as additional sample years might aggravate the imbalanced sample between Big 4 period (two fiscal years: 2004-2005) and Big 3 period (six fiscal years: 2006-2011). All of the audit fee and control variables data are obtained from the Nikkei Economic Electronic Database Systems (NEEDS) FinancialQUEST and Japanese securities filings information (yukashoken hokokusho) extracted from the eol database. Most Japanese companies end their fiscal year on March 31. Thus the fiscal year ended March 31, 2004 is considered as fiscal year 2003 or FY2003, consistent with prior literature [
Japanese auditors are allowed to perform joint audit engagement with a single client where each firm formulates policies and procedures with regard to joint audits in its audit manuals, pursuant to Auditing Standards Board Report No. 12 of The Japanese Institute of Certified Public Accountants (JICPA) [
The sample selection process is shown in
Firms listed in First Section of TSE from FY 2004-2011 | 22,824 |
---|---|
- missing audit fee information | (5929) |
- missing regression control variables | (332) |
- firms will multiple auditors (joint audit) | (184) |
- industry with less than ten listed companies within an industry-year | (164) |
- firms from banking, insurance, securities & other financial industries | (905) |
Final sample (firm years) | 15,310 |
consists of 15,310 firm-year observations which represent 2157 unique companies.
The highest peak of audit fee in 2007 can be attributed to more stringent accounting and auditing regulations following the amendment of Financial Instruments and Exchange Law and stricter JICPA self-regulations [
Year | 2004 | 2005 | 2006 | 2007 | 2008 |
---|---|---|---|---|---|
Sample size | 1481 | 1854 | 1873 | 1964 | 2005 |
Number of auditors | 108 | 117 | 120 | 109 | 113 |
Year | 2009 | 2010 | 2011 | Average 2004-2011 | Total 2004-2011 |
Sample size | 2010 | 2040 | 2083 | 1914 | 15,310 |
Number of auditors | 112 | 110 | 110 | 112.375 | 199 |
Audit Fee Statistics (million yen) | |||||
Year | 2004 | 2005 | 2006 | 2007 | 2008 |
Mean | 67.03 | 68.94 | 86.30 | 95.96 | 73.36 |
Median | 43.40 | 42.00 | 44.00 | 48.00 | 39.70 |
St. Dev. | 168.94 | 195.56 | 355.30 | 384.06 | 149.85 |
Year | 2009 | 2010 | 2011 | Average 2004-2011 | |
Mean | 75.33 | 74.99 | 72.33 | 77.03 | |
Median | 40.00 | 39.00 | 38.00 | 42.00 | |
St. Dev. | 154.45 | 161.57 | 150.53 | 233.33 | |
Year | Average 2004-2005 | Average 2006-2011 | Relative Change (%) | ||
Mean | 68.09 | 79.52 | 17% | ||
Median | 42.00 | 42.00 | 0% | ||
St. Dev. | 184.19 | 245.23 | 33% |
Independent Variables | Auditor Size (Mean) | Client Size (Mean) | |||||
---|---|---|---|---|---|---|---|
Client’s attributes: | Mean | Std. | Big Na | Mid-Tierb | Other Non-Big N | Large Clientc | Small Clientd |
TA (million yen) | 255,144.57 | 796,194.64 | 297,269.57 | 92,865.40 | 83,557.41 | 300,104.15 | 182,178.65 |
IndPTA | 0.01 | 0.04 | 0.02 | 0.01 | 0.01 | 0.01 | 0.01 |
SUBS | 2.09 | 1.25 | 2.17 | 1.83 | 1.75 | 2.18 | 1.95 |
FORN | 0.15 | 0.22 | 0.15 | 0.13 | 0.11 | 0.15 | 0.13 |
ROI | 0.02 | 0.10 | 0.02 | (0.00) | (0.02) | 0.02 | 0.01 |
LIQ | 0.78 | 0.17 | 0.79 | 0.77 | 0.77 | 0.79 | 0.78 |
LEV | 1.82 | 6.61 | 1.74 | 1.82 | 2.41 | 1.70 | 2.01 |
LOSS | 0.16 | 0.36 | 0.14 | 0.22 | 0.23 | 0.15 | 0.16 |
GAAP | 0.02 | 0.13 | 0.02 | 0.00 | 0.00 | 0.02 | 0.01 |
Auditor’s attributes: | |||||||
TEAM | 1.38 | 1.42 | 1.49 | 1.09 | 0.89 | 1.57 | 1.08 |
TENR | 2.63 | 1.56 | 2.45 | 3.24 | 3.44 | 2.72 | 2.49 |
NAF | 0.24 | 0.78 | 0.29 | 0.03 | 0.01 | 0.37 | 0.03 |
AISpec | 0.20 | 0.11 | 0.24 | 0.03 | 0.02 | 0.25 | 0.12 |
DOMN | 9.76 | 4.10 | 9.82 | 9.82 | 9.26 | 10.21 | 9.02 |
Audit engagement’s attributes: | |||||||
AOP | 0.00 | 0.04 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
POW | 0.20 | 0.32 | 0.06 | 0.51 | 0.91 | 0.06 | 0.42 |
DIST | 0.06 | 0.05 | 0.06 | 0.06 | 0.06 | 0.06 | 0.05 |
GFC | 0.13 | 0.34 | 0.13 | 0.15 | 0.13 | 0.17 | 0.07 |
REG | 0.13 | 0.33 | 0.13 | 0.13 | 0.12 | 0.17 | 0.07 |
ACC | 0.05 | 0.07 | 0.05 | 0.06 | 0.06 | 0.05 | 0.05 |
The table provides the mean and standard deviation of the independent variables included in the regression models categorized by all observation period, auditor size and client size. Notes: aBig N firms include Deloitte Touche Tohmatsu, E&Y Shin Nihon, KPMG AZSA, PwC Aarata, PwC Chuo Aoyama, and PwC Misuzu. bMid-tier firms include unaffiliated and mid-tier local audit firms affiliated with BDO International, Grant Thornton International, Kreston International, NEXIA International, Baker Tilly International, Crowe Horwath, PKF International, Plante & Moran, RSM International, and TIAG (The International Accounting Group). cA client is categorized as large client if the median total assets belong to the upper half (>50th percentile) of the industry- year sample. dA client is categorized as small client if the median total assets belong to the lower half (<50th percentile) of the industry-year sample. Definitions of the independent variables are described in
regression analyses over the observation period. The untabulated average mean (median) non-audit fee (NAF) is 2.52 (0) million yen. This figures shows that it is uncommon for auditors of Japanese listed firms to perform non-audit services. The non-audit services are commonly provided by Big N firms to their large size audit clients. The ratio of non-audit fee to audit fee paid by Japanese listed firms are extremely small (3.17%) when compared to other developed country that has similar audit and legal environment to Japan. In German audit market, the non-audit fee amount to 41.9% of the total fee paid to auditors and is considered to be as important as audit fee [
Only 250 firm-years (1.63% of total sample which consists of 39 unique companies) employ non-Japanese accounting standards (SEC registrants or IFRS). The mean (median) audit fee paid by clients who adopt non-Japanese accounting standards (SEC registrants or IFRS) is 1040.63 million yen (530 million yen). These figures are significantly higher than audit fee paid by clients who follow Japanese GAAP (J-GAAP) with a mean (median) fee of 61.03 million yen (41.8 million yen). The higher audit fee paid by adopters of non-Japanese accounting standards is consistent with prior study [
To ensure that the multicollinearity problem does not introduce bias the regression results, the variance inflation factor (VIF) for the audit fee regression models is calculated. The VIF value of ten is generally considered as rules of thumb to indicate excessive or serious multi-collinearity [
The FeePrem interaction variable examine whether the higher audit fee paid to Big N auditors is differentially affected by client market segment (H1). The estimate of interaction variable between FeePrem and ClientSeg (α4 coefficient in Equation (1)) is positive and statistically significant, suggesting that Big N auditors receive higher fee as client size becomes larger. Hypothesis 1 argues that pricing between Big N and non-Big N auditors is regarded to be competitive if Big N firms earn consistent fee premium through audit service differentiation that does not vary with the size of clients [
Variable | Equation (1): Small and Large Client Segments (2004-2011) | |||
---|---|---|---|---|
Coef. | t-stat | p-value | ||
FeePrem | 0.235 | 12.435 | 0.000 | *** |
Big3Per | 0.164 | 12.780 | 0.000 | *** |
ClientSeg | (0.089) | (5.006) | 0.000 | *** |
FeePrem × ClientSeg | 0.076 | 4.531 | 0.000 | *** |
TA | 0.251 | 33.862 | 0.000 | *** |
IndPTA | 1.428 | 7.291 | 0.000 | *** |
SUBS | 0.093 | 15.089 | 0.000 | *** |
FORN | 0.051 | 2.416 | 0.016 | ** |
ROI | (0.351) | (7.832) | 0.000 | *** |
LIQ | 0.124 | 4.378 | 0.000 | *** |
LEV | 0.001 | 1.842 | 0.065 | * |
LOSS | 0.093 | 9.194 | 0.000 | *** |
GAAP | 1.504 | 24.641 | 0.000 | *** |
TEAM | 0.014 | 4.381 | 0.000 | *** |
TENR | (0.015) | (5.055) | 0.000 | *** |
NAF | 0.137 | 13.515 | 0.000 | *** |
AISpec | 0.140 | 2.435 | 0.015 | ** |
DOMN | 0.016 | 6.814 | 0.000 | *** |
AOP | (0.215) | (1.998) | 0.046 | ** |
POW | 0.201 | 9.526 | 0.000 | *** |
DIST | (0.561) | (4.031) | 0.000 | *** |
GFC | (0.187) | (1.349) | 0.177 | |
REG | 0.100 | 0.724 | 0.469 | |
ACC | 0.020 | 0.270 | 0.787 | |
n | 15,310 | |||
Industry dummy variables | Included | |||
Year dummy variables | Included | |||
Adj. R-Squared | 64.02% |
*, ** and *** represent statistical significance at the 10%, 5%, and 1% level, respectively. The dependent variable is AF (natural log of total audit fee). Definitions of the independent variables are described in
higher fee premium in the large clients segment compared to the fee premium earned in the small clients segment.
The results of the regression estimate suggests a non-competitive audit pricing market between Big N and non-Big N auditors in which Big N auditors are paid 7.91% higher fee from their large clients pay compared to fee that small clients’ paid to their Big N or non-Big N auditors. The results from
To complement the results of
Variable | Equation (1): Large Client Segment | Equation (1): Small Client Segment | ||||||
---|---|---|---|---|---|---|---|---|
Coef. | t-stat | p-value | Coef. | t-stat | p-value | |||
FeePrem | 0.465 | 14.378 | 0.000 | *** | 0.156 | 7.833 | 0.000 | *** |
Big3Per | 0.437 | 22.037 | 0.000 | *** | 0.395 | 18.030 | 0.000 | *** |
TA | 0.293 | 23.551 | 0.000 | *** | 0.167 | 20.732 | 0.000 | *** |
IndPTA | 0.658 | 3.466 | 0.001 | *** | 5.093 | 1.877 | 0.061 | * |
SUBS | 0.111 | 10.925 | 0.000 | *** | 0.075 | 13.428 | 0.000 | *** |
FORN | 0.011 | 0.325 | 0.745 | 0.078 | 3.402 | 0.001 | *** | |
ROI | (0.502) | (3.789) | 0.000 | *** | (0.249) | (5.504) | 0.000 | *** |
LIQ | 0.141 | 2.802 | 0.005 | *** | 0.037 | 1.160 | 0.246 | |
LEV | 0.005 | 1.663 | 0.096 | * | 0.001 | 1.072 | 0.284 | |
LOSS | 0.083 | 4.679 | 0.000 | *** | 0.063 | 5.724 | 0.000 | *** |
GAAP | 1.454 | 19.736 | 0.000 | *** | - | - | - | |
TEAM | 0.006 | 1.249 | 0.212 | 0.040 | 10.986 | 0.000 | *** | |
TENR | (0.015) | (3.113) | 0.002 | *** | (0.018) | (5.711) | 0.000 | *** |
NAF | 0.116 | 9.965 | 0.000 | *** | 0.031 | 2.392 | 0.017 | ** |
AISpec | 0.226 | 2.722 | 0.007 | *** | 0.073 | 1.068 | 0.286 | |
DOMN | 0.026 | 10.153 | 0.000 | *** | 0.011 | 5.449 | 0.000 | *** |
AOP | (0.021) | (0.121) | 0.904 | (0.186) | (1.426) | 0.154 | ||
POW | 0.463 | 11.342 | 0.000 | *** | 0.032 | 1.303 | 0.193 | |
DIST | (0.803) | (5.167) | 0.000 | *** | (0.328) | (2.641) | 0.008 | *** |
GFC | (0.240) | (0.849) | 0.396 | (0.982) | (42.982) | 0.000 | *** | |
REG | (0.540) | (1.604) | 0.109 | (0.049) | (0.659) | 0.510 | ||
ACC | 0.122 | 1.009 | 0.313 | 0.040 | 10.986 | 0.000 | *** | |
n | 7597 | 7713 | ||||||
Industry dummy variables | Included | Included | ||||||
Year dummy variables | Included | Included | ||||||
Adj. R-Squared | 60.19% | 13.68% |
*, ** and *** represent statistical significance at the 10%, 5%, and 1% level, respectively. The dependent variable is AF (natural log of total audit fee). Definitions of the independent variables are described in
of the audit market concentration regression model (Equation 1) using the client segment size subsample of large client segment (sample size = 7597 firm-years) and small client segment (sample size = 7713 firm-years) following [
The difference-in-difference (DiD) interaction variable between FeePrem and Big3Per (α4 coefficient in Equation (2)) is positive and significant at 5%; providing empirical support of a non-competitive audit pricing market between Big N and non-Big N after the audit market transition. Hypothesis 2 argues that audit market becomes less competitively priced if audit fee is positively associated with Big N audit fee premium following the transition from Big 4 to Big 3 period. Using the economic significance measurement approach of [
To complement the results of
To control for auditor selectivity bias inherent in prior audit fee studies, the two-stage Heckman (1979) procedure is employed [
Variable | Equation (2): Transition from Big 4 to Big 3 Period | |||
---|---|---|---|---|
Coef. | t-stat | p-value | ||
FeePrem | 0.240 | 10.467 | 0.000 | *** |
Big3Per | 0.120 | 5.348 | 0.000 | *** |
ClientSeg | (0.028) | (2.733) | 0.006 | *** |
FeePrem × Big3Per | 0.035 | 1.871 | 0.061 | ** |
TA | 0.251 | 33.775 | 0.000 | *** |
IndPTA | 1.449 | 7.399 | 0.000 | *** |
SUBS | 0.093 | 15.052 | 0.000 | *** |
FORN | 0.053 | 2.479 | 0.013 | ** |
ROI | (0.362) | (7.964) | 0.000 | *** |
LIQ | 0.121 | 4.252 | 0.000 | *** |
LEV | 0.001 | 1.865 | 0.062 | * |
LOSS | 0.095 | 9.314 | 0.000 | *** |
GAAP | 1.510 | 4.710 | 0.000 | *** |
TEAM | 0.014 | 4.478 | 0.000 | *** |
TENR | (0.015) | (4.995) | 0.000 | *** |
NAF | 0.138 | 13.710 | 0.000 | *** |
AISpec | 0.129 | 2.221 | 0.026 | ** |
DOMN | 0.016 | 6.827 | 0.000 | *** |
AOP | (0.216) | (2.011) | 0.044 | ** |
POW | 0.200 | 9.367 | 0.000 | *** |
DIST | (0.561) | (4.011) | 0.000 | *** |
GFC | (0.194) | (1.297) | 0.195 | |
REG | 0.090 | 0.603 | 0.547 | |
ACC | 0.026 | 0.351 | 0.726 | |
n | 15,310 | |||
Industry dummy variables | Included | |||
Year dummy variables | Included | |||
Adj. R-Squared | 63.97% |
*, ** and *** represent statistical significance at the 10%, 5%, and 1% level, respectively. The dependent variable is AF (natural log of total audit fee). Definitions of the independent variables are described in
Variable | Equation (1): Big 4 Period (2004-2005) | Equation (1): Big 3 Period (2006-2011) | ||||||
---|---|---|---|---|---|---|---|---|
Coef. | t-stat | p-value | Coef. | t-stat | p-value | |||
FeePrem | 0.312 | 5.482 | 0.000 | *** | 0.230 | 10.741 | 0.000 | *** |
ClientSeg | (0.049) | (1.151) | 0.250 | (0.101) | (4.719) | 0.000 | *** | |
FeePrem × ClientSeg | 0.013 | 0.316 | 0.752 | 0.095 | 4.801 | 0.000 | *** | |
TA | 0.226 | 14.050 | 0.000 | *** | 0.257 | 29.058 | 0.000 | *** |
IndPTA | 0.997 | 2.270 | 0.023 | ** | 1.677 | 6.888 | 0.000 | *** |
SUBS | 0.051 | 4.203 | 0.000 | *** | 0.106 | 13.996 | 0.000 | *** |
FORN | 0.143 | 2.827 | 0.005 | *** | 0.026 | 1.033 | 0.302 | |
ROI | (0.244) | (3.194) | 0.001 | *** | (0.399) | (7.689) | 0.000 | *** |
LIQ | (0.083) | (1.235) | 0.217 | 0.141 | 4.559 | 0.000 | *** | |
LEV | 0.002 | 0.552 | 0.581 | 0.001 | 1.782 | 0.075 | * | |
LOSS | 0.047 | 1.568 | 0.117 | 0.102 | 8.684 | 0.000 | *** | |
GAAP | 1.018 | 7.712 | 0.000 | *** | 1.541 | 21.049 | 0.000 | *** |
TEAM | 0.024 | 3.319 | 0.001 | *** | 0.012 | 3.182 | 0.001 | *** |
TENR | 0.014 | 0.278 | 0.781 | (0.014) | (5.044) | 0.000 | *** | |
NAF | 0.264 | 6.831 | 0.000 | *** | 0.115 | 10.152 | 0.000 | *** |
AISpec | 0.352 | 2.054 | 0.040 | ** | 0.061 | 0.973 | 0.330 | |
DOMN | 0.017 | 2.636 | 0.008 | *** | 0.017 | 7.092 | 0.000 | *** |
AOP | (0.170) | (1.068) | 0.286 | (0.114) | (0.947) | 0.344 | ||
POW | 0.344 | 5.489 | 0.000 | *** | 0.172 | 6.876 | 0.000 | *** |
DIST | (0.004) | (0.007) | 0.994 | (0.569) | (3.281) | 0.001 | *** | |
GFC | (0.235) | (2.387) | 0.017 | ** | (0.260) | (13.570) | 0.000 | *** |
REG | 0.312 | 5.482 | 0.000 | *** | (0.134) | (5.043) | 0.000 | *** |
ACC | (0.049) | (1.151) | 0.250 | 0.124 | 1.650 | 0.099 | * | |
n | 3,335 | 11,975 | ||||||
Industry dummy variables | Included | Included | ||||||
Year dummy variables | Included | Included | ||||||
Adj. R-Squared | 27% | 62.52% |
*, ** and *** represent statistical significance at the 10%, 5%, and 1% level, respectively. The dependent variable is AF (natural log of total audit fee). Definitions of the independent variables are described in
procedure assumes that auditor size variable (BigN) is endogenous in the audit fee model where companies are not randomly assigned to audit firms and clients has the choice whether to hire large or small audit firms [
B i g N i , t = α 0 + α 1 T A i , t + α 2 L E V i , t + α 3 G A A P i , t + α 4 L I Q i , t + α 5 F O R N i , t + α 6 R O I i , t + α 7 L O S S i , t + ε i , t (3)
where:
B i g N = dummy variable equals to 1 if the client is audited by one of the Japanese Big N auditors (Deloitte Touche Tohmatsu, E&Y Shin Nihon, KPMG AZSA, PwC Aarata, PwC Chuo Aoyama, and PwC Misuzu), and 0 otherwise.
T A = natural log of clients’ total assets.
L E V = ratio of the client’s total liabilities to total equity.
G A A P = dummy variable equal to 1 if the client is a SEC registrant or an IFRS adopter, and 0 otherwise.
L I Q = ratio of the client’ current assets (less inventories) to current liabilities.
F O R N = ratio of the client’s overseas sales to net sales.
R O I = ratio of the client’s net income to total assets.
L O S S = dummy variable equal to 1 if the client incurred a net loss in the previous fiscal year, and 0 otherwise.
The untabulated results from the probit model show that the likelihood ratios for Big N and non-Big N auditors are significant (p-value of chi-squared test is less than 0.01), which suggest that the probit model of Equation 3 can effectively differentiate between Big N and non-Big N auditors [
Year-by-year analysis of the cross-sectional regression models (Equation (1)) is conducted to rule out the possibility that repeated observations in the cross-sectional regressions have inflated the significance of the coefficients [
The main regression models (Equation (1)) are re-estimated on a balanced set of sample which include four fiscal years period (Big 4 period sample from 2004 to 2005 and Big 3 period sample from 2006 to 2007). The 4-years balanced sample is comprised of 7172 firm-years. The findings on the audit pricing competitiveness among individual firms at the industry level are also mixed. However, these findings are conditional to the low value of the adjusted R-squared for Equation (1) using the balanced sample that is significantly lower (24%) compared to the full sample in
The main empirical tests are re-estimated again on another set of sample which exclude fiscal year 2006 from the sample (observation number = 13,437 firm-years). The empirical findings of the 2006 fiscal year excluded sample are consistent with the results of the balanced 4-years period sample. Untabulated regression estimation results for Equation (1) for the 2006 fiscal year excluded sample show a significantly lower degree of adjusted R-squared (12%) compared to the full sample in
The 2010 European Commission report predicts that the collapse of one of the Big 4 large audit firms could potentially impair the stability of the financial system [
The descriptive statistics results of audit fee show that the average audit fee during the Big 3 period (2006-2011) is 16.8% higher than that of the Big 4 period (2004-2005). In addition, the ratio of non-audit fee to audit fee paid by Japanese listed firms is extremely small (3.17%). Japanese audit market is highly concentrated, with the three largest Big N auditors controlling more than 70% of the audit market share. Results of the industry-level audit market concentration analysis show that the industry level audit market concentration in the Big 3 period (2006-2011) is consistently higher compared to the Big 4 period (2004- 2005). Although market share of Japanese Big N firms (based on average client numbers per fiscal year) in the Big 3 period (2006-2011) declined by 4.45% compared to the prior Big 4 period (2004-2005); non-Big N firms’ market share (based on average client numbers per fiscal year) increased by 21.42% in the same period (based on author’s calculation).
The combination of the declining Big N firms market share, the rising non- Big N auditors’ market share, weaker audit service differentiation of large auditors, and low audit quality expectation for large auditors contribute to greater likelihood for Japanese Big N and non-Big N auditors to compete for clients. Those factors motivate this study to further examine audit pricing competitiveness between Big N and non-Big N auditors in Japan. This research investigates audit pricing competitiveness between Big N and non-Big N auditors using the Big N audit fee premium differential between large and small clients. The empirical results show that a non-competitive audit pricing market exists between large and smaller size auditors where Big N auditors earn increasingly higher audit fee as their audit clients become bigger. In addition, the difference-in-difference analysis results show that the transition from Big 4 to Big 3 period contributes to a less competitive audit pricing between Big N and non-Big N auditors. Thus, the concerns regarding the lack of a competition between Big N and non-Big N auditors in the Japanese audit market following the transition from Big 4 to Big 3 audit market is warranted.
This study provides evidence of a non-competitive audit pricing between Big N and non-Big N auditors that can be explained for the following reasons. First, prior empirical study has shown that Japanese listed firms are concerned with the good reputation of their auditors [
The adverse effect of Big 4 to Big 3 transition on audit pricing competition provides urgency for audit market reform as another reduction in the number of large auditors could seriously impair the stability of financial markets. Although higher market concentration may reduce audit costs by economies of scale, audit services may become overpriced in the long run [
This study has a number of limitations. The observation period for the Big 3 period is limited to fiscal year 2011 as additional sample years might aggravate the imbalanced sample between Big 4 period (two fiscal years: 2004-2005) and Big 3 period (six fiscal years: 2006-2011). Audit firms do not disclose necessary information to calculate marginal costs and marginal revenues of audit services [
Frendy (2018) Big N and Non-Big N Audit Pricing Competitiveness in Japan during Transition from Big 4 to Big 3 Period. Open Journal of Accounting, 7, 42-72. https://doi.org/10.4236/ojacct.2018.71004