The Audit Market Dynamics in a Mandatory Joint Audit Setting: The French Experience

Abstract

Policy makers in France have considered joint audits as a solution to mitigate the audit market concentration and the “systemic” risk associated with Big 4 auditors. We implement a Markovian analysis where audit clients chose between different types of combinations across Big 4 and smaller auditors. Our main findings support the view that the French joint audit system is effective in maintaining market openness and in mitigating the Big 4 domination in the long run. An investigation of the determinants driving changes in joint audit combinations suggests little economic support in favor of two Big 4 combinations, whereas changes in audit clients’ agency costs (e.g., higher ownership concentration) tend to explain the performance of mixed and two non-Big 4 combinations. Overall, this study supports the European Commission’s position on the potential benefits of joint audits in mitigating the market concentration; it also suggests that it might not be necessary to impose mixed joint audits to achieve that objective.

Keywords

joint audit auditor concentration auditor switch Markov chain France

Introduction

The high level of suppliers’ concentration on the market for audit services has gained a growing attention from policy makers, especially since the 2002 demise of Arthur Andersen (The American Assembly, 2005; European Commission, 2010, 2014; Financial Reporting Council, 2010; U.S. Government Accountability Office, 2003). Potential threats associated with extreme concentration levels include both demand and supply side effects. On the demand side is the lack of choice for public interest entities (e.g., Financial Reporting Council, 2010), which are required to use the services of independent auditors. On the supply side are concerns about reduced competition and potential detrimental effects on audit pricing and/or quality.

Among the solutions put forward by policy makers to mitigate this high level of concentration, and the “systemic” importance of Big 4 auditors (European Commission, 2010), stands the use of joint audits, periodic retendering of audit engagements, and mandatory rotation of audit firms. In its 2014 regulation, the European Commission recommends the use of joint audit involving a “systemic” (Big 4) and a nonsystemic audit firm, and requires a mandatory rotation of audit firms after 10 years.¹ This audit regulation has the objective to revitalize the audit market and to favor the emergence of Mid-Tier European-wide audit networks likely to compete with the Big 4.

From an academic standpoint, however, very little is known about the effects of joint audit on the audit market dynamics in general and on its ability to mitigate concentration problems in particular.² Descriptive evidence on the French audit market suggests that the joint audit requirement has resulted in a slightly less concentrated situation as compared with other European markets (Ballas & Fafaliou, 2008; Bédard, Piot, & Schatt, 2014; Broye, 2007), and that it allowed the emergence or survival of a decent fringe of national “Major” networks (Piot, 2007). In fact, most listed French companies choose to pair a Big 4 with a non-Big 4 audit firm (e.g., Marmousez, 2012). France thus seems to illustrate the capacity of joint audit system to alleviate concentration problems. However, the French market did not avoid the concentration trend induced by megamergers, or disappearance, of “Big” audit firms, and has the characteristics of a closed oligopoly (Piot, 2008). These static observations provide information about the immediate consequences of suppliers’ concentration but do not say anything about the long-term effects arising out of a joint audit requirement in terms of future market shares and levels of competition. Specifically, is the greater market openness brought about by the joint audit requirement effective in maintaining, or developing, the market position of non-Big 4 auditors (and specifically Mid-Tier auditors) in the long run? At the audit clients’ level, what are the economic determinants underlying changes in the choice of joint auditors?

In this article, we investigate the dynamics of the French joint audit market. We use a unique database covering the entire population of listed French companies. The French audit setting offers a convenient laboratory in that purpose, for several reasons. First, joint audits have been required by law since 1966, criticized but strongly reconfirmed in the 1980s notably for protectionist reasons (Assemblée Nationale, 1984; Bédard et al., 2014). Second, beyond the minimum requirement of two independent auditors, regulatory texts leave a complete flexibility to the audited company regarding the choice of coauditors. In other words, there is neither requirement nor recommendation for a mixed joint audit configuration (i.e., a Big 4 paired with a non-Big 4 auditor), thus leaving an open space to market mechanisms to operate. Third, the dynamics of auditor–client realignments follow a legally protected 6-year audit engagement, thus putting the auditor choice decision in a long-term “strategic” perspective for auditees.

To project the long-term market structure of joint audits, we use a Markov chain process. We adapt the Markovian analysis to the joint audit setting by considering that audit clients choose between different alternatives in terms of joint audit pairs or combinations. Each joint audit combination corresponds to a defined state of a Markov process. We simulate the long-term market structure for these combinations based on two different market conjunctures: the Big 6 to Big 4 concentration period (1997-2003) and a period of relative stability featuring the Big 4 era (2003-2009). Our main findings support the view that the French joint audit system is effective in maintaining market openness and in mitigating the Big 4 domination in the long run. Specifically, the mixed joint audit arrangement would remain the most common combination, adopted by a majority (54%) of audit clients. We also find that joint audits involving two Big 4 auditors lose in attractiveness in equilibrium, whereas those with two non-Big 4 auditors gain in popularity. The latter combination would outperform the former one with nearly 27% and 19% of audit clients, respectively. Finally, we observe that Mid-Tier auditors play a significant role in the market performance of joint audits with only non-Big 4 auditors. This distinctive role of Mid-Tier auditors is not evident in the case of mixed joint audit combinations. Overall, these findings support the European Commission’s (2010, 2014) position on the potential benefits of joint audits in mitigating the market concentration. They also suggest that it might not be necessary to impose mixed joint audits to reach an equilibrium outcome that preserves non-Big 4 market positions.

To understand the economic determinants of these long-term projections, we use a joint audit switching model. We find little economic support for the demand of two Big 4 combinations in the French setting. Agency costs proxies play in favor of mixed joint audits and also more importantly in favor of two non-Big 4 combinations when ownership gets more concentrated.

This article contributes to the audit market literature in two ways. First, we provide a long-term insight about market structure in a joint audit context, which to our knowledge has not been done before. Second, we add to the auditor choice literature by empirically documenting audit clients’ motivations for switching from one joint audit combination to another.

The remaining of the article is organized as follows. The “Theoretical Framework, Institutional Context, and Data” section depicts the theoretical framework, the relevant traits of the French institutional setting, and the dataset construction. The “Research Design and Results of the Markov Model” section presents the research design and empirical results of the Markovian analysis of the audit market. The “Research Design and Results of the Joint Audit Switching Model” section focuses on the research design and results of the joint audit combination switching model. Finally, the “Discussion and Concluding Remarks” section summarizes and concludes.

Theoretical Framework, Institutional Context, and Data

This section first briefly reviews prior studies about the market effects of audit firms’ mergers in single-audit environments (see “The Market Effects of Audit Firms’ Mergers: Background Literature” section). It then exposes the relevant French audit regulation and market structure (see “The French Institutional Setting and Audit Market Structure” section), and the economics of auditor switching in a joint audit system (see “The Economics of Auditor Switching Applied to a Joint Audit System” section). The choice of periods and data collection aspects is discussed in the “Choice of Periods and Data Collection” section.

The Market Effects of Audit Firms’ Mergers: Background Literature

Structural shocks on the market, such as Big “N” audit mergers or disappearance of a major supplier (e.g., Andersen), can affect the dynamics of market share transfers. Research on the market consequences of audit firms’ mergers try to disentangle between two main theoretical premises: market dominance and efficiency gains. The market dominance hypothesis assumes that the merger will reduce competition and that the newly merged audit firm will increase its market power and profit from abnormal rents. On the contrary, the efficiency gain hypothesis assumes that the merger will result in production efficiencies which, in the presence of sufficient competition, will be passed onto audit clients. Overall, empirical studies investigating the effects of the 1989 Big 8 to Big 6 mergers, or the 1998 PricewaterhouseCoopers (PwC) merger, generally fail to conclude about the existence of dominance effects and hampered competition (Choi & Zéghal, 1999; Iyer & Iyer, 1996; Pong & Burnett, 2006). Some of them even support the efficiency gains hypothesis (Ivancevich & Zardkoohi, 2000).

Studies investigating the market share transfers or changes in concentration metrics around Big “N” megamergers generally observe an increase in the overall level of concentration after the merger. However, they also document that in many instances, the merger allowed a rebalancing of market shares between the main actors, resulting in lower relative concentration levels (usually lower Hirschman–Herfindahl indices).³ These auditor concentration studies have two limitations, though. First, they are based on static or short-term changes in market observations. They do not model the dynamics of market share transfers and thus do not allow projections of future market positions. Second, it is not clear whether these apparently positive rebalancing effects also apply beyond the Big “N” audit networks. In a joint audit setting, it makes sense to appraise how different combinations of audit firms—and notably those involving a Big 4 and a non-Big 4 in view of the European Commission’s concerns—would perform in the long run.

The French Institutional Setting and Audit Market Structure

As pointed out by Francis (2011), institutions shape the supply and demand for audit services, and affect the audit market structure in a way that is not clearly understood yet.

Relevant French audit regulation

The 1966 French Company Law initially required listed entities to have their financial statements certified by at least two statutory auditors. That requirement was criticized in the early 1980s, notably by the stock market authority, for its cost and inefficiency (Marmousez, 2012). However, it was confirmed and even extended in 1984 to all public or private entities required to publish consolidated financial statements. Besides, beyond that legal requirement for joint audits, the French texts—whether from legal, auditing standard, or professional doctrine—do not address the composition of the audit college. Therefore, audited companies have complete flexibility on that ground as far as the two coauditors are effectively independent from one another, that is, belong to independent audit networks.

Another French-specific audit regulation likely to influence the market dynamics relates to the duration of the audit engagement. Auditors are legally appointed for a renewable 6-year term. They cannot be dismissed by their client before the end of that term.⁴ In a market dynamics perspective, imposing a long-term (6 years) audit engagement inevitably affects the timeliness of economic auditor–client realignments. The audit client may not have the faculty to change its auditor promptly when necessary. It results that market observations cannot be designed on the basis of consecutive years, as it is the case for studies carried out on the U.S. or U.K. markets. Once the timing of observations is extended to account for the legal duration of the audit engagement, the context offers relevant theoretical specificities. On the demand side, the 6-year engagement places the appointment of auditors into a strategic perspective for audit clients, thus making the auditor choice decision an important forward-looking one. On the supply side, a guaranteed long-term engagement makes the auditor more comfortable in the starting phase but transfers most of the nonrenewal costs to him or her at the end of the term. Hence, this long-term engagement may motivate audit firms to put more efforts in retaining their existing clients than in trying to attract new ones.⁵

Despite this legal constraint, empirical statistics on auditor tenure suggest that French long-term audit engagement does not result in a lower frequency of auditor switching. The large French companies have a mean (median) auditor tenure between 9 and 10 years (7 and 8 years; Audousset-Coulier, 2008; Piot & Janin, 2007). These figures are similar to U.S. studies using contemporaneous panels from the 1990s or early 2000s (Ghosh & Moon, 2005; Gul, Fung, & Jaggi, 2009; Lim & Tan, 2010). However, they seem to contrast with the stronger inertia observed on the U.K. market, where a Financial Times Stock Exchange (FTSE) 100 auditor remains in place for about 48 years on average and 36 years for FTSE 250 firms (House of Lords, 2010). Overall, the ability of auditors to retain their clients might not depend on a legal time protection of the audit engagement, but rather on the lack of choice associated with an extremely concentrated audit market as the British one (Oxera Consulting, 2006).

French audit market structure: Concentration and competition

In the French context, one motivation of the legislator regarding the joint audit requirement is clearly of a protectionist nature. Indeed, parliamentary debates at the time of extending joint audits in 1984 provide convincing evidence that politicians sought to protect French auditing firms in the face of the rising power of international audit networks (Assemblée Nationale, 1984).

Probably as a result of joint audit, the French audit market appears less concentrated and supposedly more competitive than other mature European markets (Ballas & Fafaliou, 2008). In 2003, the market exhibits a relatively “open” two-tiered structure. Based on clients’ total assets evenly spread among coauditors,⁶ the Big 4 have a global market share of 56.7%, whereas five Mid-Tier audit firms and other local auditors account for 17.8% and 25.5%, respectively (Piot, 2007). Also, the Mid-Tier national audit firms maintained significant positions as the number of “Big” international networks shrunk from six to four. Specifically, Mazars performed well in attracting new audit clients from the merged PwC⁷ network and from the French member of Arthur Andersen (Barbier Frinault). This does not preclude, however, the unavoidable concentration trend associated with the decrease in the number of major players. Indeed, the estimated size of the oligopoly decreased from eight to five main actors (the Big 4 plus Mazars) between 1997 and 2003 (Piot, 2008). This is consistent with the 0.21 Hirschman–Herfindahl index reported by Broye (2007), based on the audit fees effectively received by the coauditors.

To summarize, under the joint audit requirement, a decent fringe of Mid-Tier audit firms have survived, with at least one of them (Mazars) often being in a position to compete with the Big 4. This market openness brought about by joint audit is confirmed by the type of auditor pairs observed in empirical studies. The mixed joint audit combination is indeed the most frequently adopted one and concerns the majority of listed companies (Francis, Richard, & Vanstraelen, 2009; Le Vourc’h & Morand, 2011; Marmousez, 2012). The capacity of a joint audit system to alleviate concentration problems is thus strongly related to the persistence of mixed joint audit combinations.

The Economics of Auditor Switching Applied to a Joint Audit System

Theory and background literature

We refer to the auditor switching literature to investigate the economic determinants of the joint audit market dynamics. Auditor–client realignments may occur due to changes in audit clients’ financial or operating conditions (Johnson & Lys, 1990). Given market competition, audit firms respond to this demand by differentiating themselves in terms of audit technology (Dopuch & Simunic, 1980) or audit quality and reputation (DeAngelo, 1981). Following this, prior studies suggest that auditor–client realignments are associated with changes in audit clients’ agency costs (DeFond, 1992; Francis & Wilson, 1988), or growth and financing needs (Johnson & Lys, 1990). In these studies, the switch in favor of larger or higher profile (Big N) auditors is generally interpreted as a demand for greater levels of audit quality. More in line with audit complexity issues, Francis, Maydew, and Sparks (1999) provide evidence that the use of Big N auditors is more likely for companies with accrual-intensive reporting. They argue that Big N auditors are more able to address the audit complexity of their clients’ operating cycle and can make the reporting of accruals to external parties more credible.

French-specific audit context and literature

As explained in “The French Institutional Setting and Audit Market Structure” section, it is relatively uneasy to investigate the economics of auditor–client realignments in France due to the imposed 6-year audit engagement. Empirically, this requires observing the market over time frames of at least 6 fiscal years to allow for auditor dismissals. To our knowledge, Piot (2003) provides the only study of auditor switches in France. Considering the change in audit clients’ characteristics over the 6 years preceding the switch, he reports a positive association between the appointment of a higher profile auditor and the change in leverage. This is consistent with an agency-related demand for audit quality associated with debt financing. However, the auditor choice decision in his study is made at the level of individual auditor replacements and does not consider the joint audit specificity.

Further empirical studies have addressed the choice of joint audit combinations in France, although they have done so in a nondynamic perspective only. Francis et al. (2009) and Marmousez (2012) both use a cross-sectional sample of French firms in 2003. They reach a similar conclusion that the use of Big 4 auditors (one or two) increases with ownership diffusion, consistent with an agency-related demand for audit quality. The choice of a mixed versus a two non-Big 4 combination is also found to be a matter of audit complexity and international operations.

Choice of Periods and Data Collection

To cope with the 6-year audit engagement and allow each company the opportunity to renew or to replace its coauditors, we design data collection along 6-year intervals, starting in year 1997. This starting point is chosen to provide the latest picture of market positions under the Big 6 era, just before the PwC merger, which took place during year 1998 and was achieved in September 1998. The next data collections are then made at the end of years 2003 and 2009. Hence, the obtained database offers two very different subperiods in terms of market context. The period 1997-2003 can be viewed as a shock or turmoil period, with the Big 6 shrinking to the Big 4. That period fully integrates the market effects of the PwC merger, given that an auditor choice decision made during year 1998 is already placed in a PwC context and will engage the audit client until financial year 2003. The period 1997-2003 may also be partially affected (because of the 6-year inertia) by some immediate consequences of the 2002 Andersen demise, although the French situation remains quite singular on that point. The French audit operations of Arthur Andersen were held by a national member of the network named Barbier Frinault. Following the international network restructuration agreements, Barbier Frinault joined the Ernst & Young network. Although its client portfolio might have been slightly eroded by Andersen’s shredded reputation effects, the use of a distinctive name may have limited these effects. The period 2003-2009 then offers a benchmark of relative market stability, with only one significant concentration event: the acquisition of the second largest Mid-Tier audit firm (Salustro Reydel) by KPMG in 2005.

Common databases have significant caveats in providing reliable auditor names, especially in a joint audit context.⁸ Hence, that information had to be hand-collected from a direct consultation of listed companies’ annual audit report. Table 1 summarizes the available audit client database for the three targeted years, after elimination of firms for which a reliable identification of the auditors was not possible, and firms with a single-audit regime (mostly small entities listed on the over the counter [OTC] market). Table 1 also reports the usable database for both subperiods given the “closed” system requirement of the Markov model (i.e., firms must be present at the beginning and at the end of the period).

Table 1.

Audit Clients’ Database.

Year	1997	2003	2009
Number of French registered listed firms at year end^a	823	917	944
Minus
Auditor name unidentifiable	−8	−30	−44
Single audits	−86	−123	−222
Number of firms with an identified joint audit	729	764	678
Including
Joint audits with three coauditors	16	15	9
Firms on the unregulated (OTC) segment	89	216	213
Balanced panel available for period 1997-2003^b,c	491 firms 986 engagements
Balanced panel available for period 2003-2009^b,c		529 firms1,062 engagements

Note. OTC = over the counter.

The population is targeted according to the year-end “Cote officielle” (i.e., official listing) published by the Stock market authority.

To keep the number of audit engagements constant, cases with three coauditors are kept as such if the audit client appoints three auditors both at the beginning and at the end of the period (assumption of a “stable” audit demand). For firms changing from two to three coauditors (or vice versa), the third added (or removed) coauditor is ignored (assumption of an “unusual” demand for three coauditors). This procedure results in neutralizing 13 (11) engagements for the period 1997-2003 (2003-2009). In each period, only four audit clients post a constant three coauditors college.

We performed independent-sample Student’s comparison tests to ensure that both panels are comparable in terms of the size of audit clients. Specifically, for year 2003, total assets and sales are not statistically different between the two panels (p values of .81 and .48, respectively). Similarly, total assets and sales are not statistically different when comparing year 1997 of the first panel (1997-2003) with year 2003 of the second panel (2003-2009) (p values of .35 and .51, respectively).

Table 2 reports the market positions of the Big 6/4 and Mid-Tier audit firms at the end of each period, based on the total population available. It pictures the two-tiered structure of the French audit market: (a) Big 6/4 auditors, steadily improving their market position but retaining “only” 37% of the audit engagements offered in 2009; (b) Mid-Tier audit firms, generally holding individually more than 20 engagements (i.e., more than 1.5% of the total number of engagements); and (c) other smaller auditors labeled “Local,” who individually represent less than 0.5% of the engagements.⁹

Table 2.

Market Positions of the Main Audit Firms by Number of Audit Engagements.

	1997		2003		2009
	Frequency	%	Frequency	%	Frequency	%
Ernst & Young	96	6.2	190	11.4	173	10.9
Deloitte & Touche	52	3.3	124	7.4	154	9.7
Pricewaterhouse/PwC	67	4.3	110	6.6	97	6.1
Coopers & Lybrand	48	3.1	Merged into PwC in 1998
KPMG	98	6.3	128	7.7	160	10.1
Barbier Frinault (Andersen)	82	5.3	Joined Ernst & Young in 2003
Total Big 6/4	443	28.4	552	33.1	584	36.8
Mazars	96	6.2	90	5.4	127	8.0
Salustro Reydel	34	2.2	40	2.4	Acquired by KPMG
Amyot Exco/Grant Thornton	26	1.7	57	3.4	54	3.4
Calan Ramolino	42	2.7	Acquired by Deloitte in 1998
Constantin & Associés	21	1.3	28	1.7	31	2.0
Fidulor	23	1.5	Joined Grant Thornton in 2001
Total Mid-Tiers	242	15.5	215	12.9	212	13.4
Other “Local” (<0.5%)	875	56.1	899	54.0	791	49.8
Total audit engagements	1,560	100	1,666	100	1,587	100

Note. PwC = PricewaterhouseCoopers. Shaded areas indicate significant mergers between audit firms during the period of analysis.

Table 3 reports the observed transition matrix for the six possible joint audit combinations. Panels A and B address the concentration (1997-2003) and stability (2003-2009) periods, respectively.

Table 3.

Observed Transitions for Joint Audit Combination Types.

Panel A: Concentration Period (1997-2003).
Lost to gained from	B4-B4	B4-MT	B4-Loc	MT-MT	MT-Loc	Loc-Loc	Opening position	Clients lost
B4-B4	28	8	4	0	1	0	41	13
B4-MT	8	38	3	3	1	1	54	16
B4-Loc	16	10	115	0	7	8	156	41
MT-MT	0	0	0	3	4	1	8	5
MT-Loc	1	9	9	6	45	6	76	31
Loc-Loc	1	6	49	0	14	86	156	70
Closing position	54	71	180	12	72	102	491
Clients gained	26	33	65	9	27	16		176
Panel B: Stability Period (2003-2009).
\Lost to gained from	B4-B4	B4-MT	B4-Loc	MT-MT	MT-Loc	Loc-Loc	Opening position	Clients lost
B4-B4	52	7	7	0	1	0	67	15
B4-MT	2	43	3	1	5	0	54	11
B4-Loc	22	11	151	1	10	13	208	57
MT-MT	0	0	0	2	1	0	3	1
MT-Loc	0	7	9	2	37	6	61	24
Loc-Loc	4	2	25	0	19	86	136	50
Closing position	80	70	195	6	73	105	529
Clients gained	28	27	44	4	36	19		158

Note. Based on the number of audit clients per joint audit combination. The empirical basis includes a balanced panel of 491 (529) audit clients for the 1997-2003 concentration (2003-2009 stability) periods (see Table 2 for details). Shaded areas designate the number of unchanged positions. B4 = Big 4; MT = Mid-Tier; Loc = other “Local” auditors.

Table 3 displays losses in rows and gains in columns, with both types of movement equaling each other as a result of the “closed” system requirement. Overall, 176 (35.8%) of the audit clients changed their joint audit configuration during the 1997-2003 concentration period. A slightly lower proportion (29.9%) did so during the 2003-2009 stability period. The following example provides a more in-depth reading of Table 3. For the concentration period (1997-2003), the two Big 4 combination was used by 41 audit clients at the beginning (1997); 28 firms had maintained that combination by the end of the period (2003); 13 firms switched to another combination during the period (including eight in favor of the Big 4-Mid-Tier arrangement). In the same time, 26 other firms adopted the two Big 4 combination (including 16 previously using the Big 4-Local arrangement). Overall, this results in 54 audit clients using two Big 4 auditors at the end of the period (2003), with a net gain for that combination of 13 (26 minus 13) audit clients.

A descriptive look at the main movements shows the following trends. During the concentration period, the Big 4 strengthened their positions for all the three joint audit combinations that concern them. The Mid-Tier audit firms also progressed when paired with a Big 4, but struggled to maintain other positions, that is, when they operate with a non-Big 4. As a result, the smaller “Local” auditors clearly lost their retention capacities, with nearly half of the audit clients abandoning the two Local combination, most of them appointing a Big 4 auditor. During the stability period (2003-2009), the transition pattern was slightly different. Although the Big 4 continued to increase their presence when coupled with another Big 4 or with a Mid-Tier, the Big 4-Local combination declined at the benefit of combinations made of non-Big 4 auditors. Hence, the Mid-Tier firms progressed globally, whether they operate with a Big 4 or with a Local coauditor. Overall, the stability period seems marked by a lower Big 4 domination and a slightly higher market performance of Mid-Tier auditors.

Research Design and Results of the Markov Model

The Markov Chain Modeling of Audit Markets

The use of a Markov chain allows to model the market dynamics. It complements prior static studies. Markov chains are widely used in finance and economics, mostly for modeling asset prices or regime-switching dynamics (Hamilton & Raj, 2002). A Markov chain modeling offers a convenient tool to appraise the dynamics of an oligopoly market, as is the case for audit services, owing to the small number of main suppliers. At each transition step, the individual market performance of each audit firm i depends on two key parameters: (a) the ability to retain its existing audit clients (retention probability, R_i), and (b) the ability to attract audit clients from other competitors (attractiveness parameters, A_i).

Three important features of the Markovian model should be underlined. First, the model works as a “closed” system of demanders and suppliers operating over a time period including one or several steps. In other words, there are no incoming or outgoing actors on both demand and supply sides while the system operates. Demanders consist in a balanced panel of audit clients present during the whole period; suppliers refer to a finite number of “states” corresponding to the main audit firms, categories of audit firms, or combinations of audit firms in a joint audit setting. Basically, all states are accessible (i.e., an audit client can switch from state i to any state j at some step n) and communicate with each other (i.e., a client can decide to move back to state i in a future step). Second, the probability to move from state i to state j only depends on state i and not on previous states (the system has no memory).¹⁰ Third, the Markov process is stationary, that is, transition probabilities are time-invariant.

Empirical studies using Markov models on audit markets are scarce and only relate to single-audit contexts. In the United States, Comunale and Sexton (2003) use the S&P 500 firms over the period 1995-1999 to estimate the effect of Arthur Andersen’s collapse on the long-term market share of the other Big 4 audit firms. Their simulations conclude that the disappearance of Arthur Andersen would result in relatively even long-term market share increases for the remaining Big 4 audit firms. Duxburry, Moizer, and Wan Mohamed (2007) investigate the market effects of the 1998 PwC megamerger in the United Kingdom. They show that the PwC merger changed the market dynamics in favor of the non-Big 5 auditors. More importantly, their findings suggest that the PwC merger produced negative synergies, which materialize in a sharp deterioration of both the client retention and attractiveness capacities of PwC, as compared with the same performance indicators for Pricewaterhouse alone in the premerger period.¹¹ The French audit context is supposedly more open due to the joint audit environment, and more long-term oriented due to the 6-year legal engagement. In this setting, we investigate the market performance factors estimated using a Markov chain model for two different market contexts: the Big 6 to Big 4 concentration period (1997-2003), and a Big 4 market stability period (2003-2009). Market projections that remain valid under both contexts are then less likely to be context-specific (see “Choice of Periods and Data Collection” section).

The Markov Chain Modeling Applied to the French Joint Audit Context

We adapt the Markovian analysis to the joint audit environment by considering that auditor choice by audit clients consists in selecting a given pair (or combination) of independent auditors. A pairwise definition of the model’s states is therefore necessary to ensure that the diffusion process meets the Markov chain assumption that an audit client has, in theory, access to all the possible states, and that the evolution of the process only depends on its present state. If possible states were defined at the auditor level, this assumption would be limiting in a joint audit setting because an audit client cannot move to auditor i if the other coauditor is already auditor i. Hence, a first approach is to posit that each state of the model corresponds to an auditor pair (i, j), where i and j are two different auditors (i≠j) chosen among a set of K independent ones. In that individual auditor-based approach, the system has a finite number of C(K, 2) = [K! / 2! × (K− 2)!] states. For example, in a market that would be restricted to the Big 4 auditors plus a homogeneous category of “other” auditors, the auditor-pairing approach would result in [5! / 2! × 3!] = 10 possible auditor pairs.

Another approach is to take a “macro” view of the market dynamics by not considering individual auditors per se, but categories of auditors, that is, the Big 4 and other possible classes of smaller actors. This approach assumes that audit clients do not target their auditor choice process to a specific auditor, but rather to a certain type of auditors, for example, a Big 4 for their higher profile. This view is consistent with the agency-related auditor choice literature in the United States (DeFond, 1992; Francis & Wilson, 1988) and in France (Broye, 2001; Francis et al., 2009; Marmousez, 2012; Piot, 2003). It assumes that auditors within a same class provide homogeneous audit services, and that the choice of the auditor itself is of secondary order for audit clients.

This “macro” approach is also consistent with the preoccupations of policy makers regarding the audit market concentration. Specifically, the European Commission considers as a priority the development of European-wide Mid-Tier audit networks able to mitigate the systemic importance of the Big 4. It is thus strongly relevant to investigate how the non-Big 4 auditors, whether Mid-Tier or smaller ones, behave and perform in a joint audit environment. A Markovian approach of the types of joint audit combinations will provide some insight on that point. Given that an audit client can appoint two different auditors of the same type (e.g., two Big 4), the number of states differs from the auditor-based approach and is formally equal to C(K, 2) +K, with K being this time the number of auditor classes. For instance, given a two-tiered structured audit market, thus offering three classes of auditors (denoted by Big 4, Mid-Tier, and Local), there will be six possible combinations.

We rely on that “macro” perspective of the joint audit system to appraise the joint audit combination dynamics at two different levels: (a) a dual market segmentation opposing Big 4 and non-Big 4 auditors, and (b) a two-tiered market segmentation where the non-Big 4 are spread among Mid-Tier and other Local audit firms. The first level corresponds to the most common distinction made in auditing research, and will provide some insight into the long-run viability of mixed (a Big 4 with a non-Big 4) joint audit combinations. The second level is in line with the two-tiered structure of the French market and prior related research (e.g., Francis et al., 2009). This more detailed segmentation will allow inferences about the long-term market performance of Mid-Tier audit actors, consistent with the European Commission’s view of the potential benefits of joint audits.

Level 1 results in three possible joint audit combinations, that is, two Big 4 and one Big 4 paired with a non-Big 4 and two non-Big 4. Level 2 would provide a six-state system: two Big 4, one Big 4 and one Mid-Tier, one Big 4 and one Local, two Mid-Tier, one Mid-Tier and one Local, and two Local auditors. However, the two Mid-Tier combination is marginally represented in our sample (only eight observations in 1997, and three in 2003; see Table 3). Hence, following Francis et al. (2009), we merge the two Mid-Tier joint audits with the one Mid-Tier and one Local ones, thus resulting in a five-state system.

In a first phase, we model the dynamics of joint audit combinations using a Markov process. To this end, we construct the observed transition matrix between the different states of the system, for both subperiods separately. Observed transitions are reported in Table 3 for the six possible joint audit combinations. Based on this matrix, we construct the matrices of observed transition probabilities p_ij for the two levels of analysis (i.e., the three-state and five-state systems). If the system meets the properties of a stationary ergodic process,¹² it will converge to an equilibrium position as the number of steps (transition opportunities) increases. Hence, there exists a vector of steady-state probabilities π_j, summing to one for j = 1 to K, which can be interpreted as the long-term representation of each of the K states of the system (Comunale & Sexton, 2003).

In a second phase, we express the transition probabilities in terms of retention and attractiveness parameters. The estimated transition probabilities ${\hat{p}}_{ij}$ are then expressed as follows: (a) the retention probability R_ij of a given state if i = j, and (b) the probability to switch (i.e., nonretention) from an existing state i to a different state j. This is expressed as the probability to leave state $i (1 - R_{i})$ , times the probability to join state j conditional on leaving state i (relative attractiveness of the targeted state j, with i≠j). Formally, for a system of K finite states,

\begin{matrix} {\hat{p}}_{ij} = R_{i} if i = j, \\ {\hat{p}}_{ij} = (1 - R_{i}) \times \frac{A_{j}}{\sum_{k \neq i} A_{k}} if i \neq j . \end{matrix}

We estimate the retention and attractiveness parameters R_i and A_i by minimizing the sum of the squared differences between the observed transition probabilities ${\hat{p}}_{ij}$ and the estimated transition probabilities $p_{ij}$ . We constrain this minimization to obtain the same steady-state probabilities using the estimated transition probabilities $({\hat{π}}_{ij})$ and observed transition probabilities $({\hat{π}}_{ij})$ . We also constrain R_i, as probabilities, to be comprised between 0 and 1. Finally, we constrain the attractiveness parameters to sum to 1 so that the denominator of $p_{ij}$ (for i≠j) represents the sum of all attractiveness parameters excluding that of auditing firm i. The ratio A_j / (1 −A_i) thus represents the probability that an audit client abandoning state i will move to state j, then resulting in ${\hat{p}}_{ij}$ being the probability to move from state i to state j. Formally, we specify, as in Duxburry et al. (2007), the following minimization problem under constraints:

\min_{R_{i}, A_{i}} {\sum_{i = 1}^{K} \sum_{j = 1}^{K} {(p_{i j} - {\hat{p}}_{i j})}^{2} | π_{j} = {\hat{π}}_{j}, j = 1, \dots, K; 0 \leq R_{i} \leq 1, i = 1, \dots, K; \sum_{j = 1}^{K} A_{j} = 1; A_{j} \geq 0} .

This system is solved using nonlinear optimization (Excel Solver add-in) to find out estimated values of R_i and A_i. These values of R_i and A_i thus produce an estimated transition matrix as close as possible to the observed one, with identical market shares.

Empirical Results

Table 4 reports estimation results for the Level 1 of the analysis, which considers joint audit combinations based on a dual segmentation of the audit market (Big 4 and non-Big 4 auditors).

Table 4.

Estimation Results in a Big 4 Versus Non-Big 4 Market Segmentation.

Type of joint audit combination	Period 1997-2003			Period 2003-2009
	π_i	R_i	A_i	π_i	R_i	A_i
Two Big 4	0.2302	0.6867	0.0560	0.1903	0.7754	0.0469
Mixed (Big 4 + non-Big 4)	0.5675	0.7799	0.8981	0.5413	0.8150	0.8812
Two non-Big 4	0.2023	0.7043	0.0459	0.2684	0.7621	0.0719
Sum of squares	0.00027			0.00002

Note. π_i designates the long-term steady-state probability of the joint audit combination; R_i is the estimated probability that an audit client will continue the same joint audit combination; and A_i is the relative attractiveness of joint audit combinations.

Table 4 calls for several comments. First, for both periods, the mixed joint audit combination remains the most popular one in the long run, representing 56.7% and 54.1% of audit clients for the concentration and stability periods, respectively. Whereas extant research on the structure of the French audit market only proposes spot observations (e.g., Francis et al., 2009; Marmousez, 2012), we provide an insight on the long-term structure of this market. Results obtained on the U.K. market suggest that the Big 4 capture 82% of audit clients in the long term (Duxburry et al., 2007, Table IV). Our results are consistent with these, as 80% (23.0% + 56.7%) of the French listed companies would appoint at least one Big 4 auditor when considering the concentration period (1997-2003). However, thanks to the joint audit, an almost similar proportion of firms (56.7% + 20.2%) would also appoint at least one non-Big 4 auditor. On that point, the joint audit system contributes to the nonsupremacy of Big 4 auditors. This observation is also valid based on a stability context (2003-2009), although the projected market structure moves toward a lower proportion of both two Big 4 and mixed joint audit combinations, which benefits to tandems of non-Big 4 auditors. The long-term market power of the Big 4 seems stronger based on the concentration period as compared with the stability period, which is rather in line with the market dominance theory. Over the stability period, the dynamics of joint audit combinations give more weight to non-Big 4 auditors, and thus confirm the market openness allowed by a required joint auditing system.

Looking at the retention and attractiveness factors provides an understanding of the change in market structure (steady-state probabilities) observed between the concentration and stability periods. Specifically, the significant increase in joint audit pairs made of non-Big 4 auditors (+6.5 points) owes to noticeable improvements in both the retention and attractiveness factors. In addition, the noticeable drop in joint audit pairs made of two Big 4 auditors (−4 points) can be attributed to a lower attractiveness of this combination.

In sum, findings based on a Big 4 versus non-Big 4 audit market segmentation suggest that the joint audit system is likely to mitigate the domination of the Big 4 audit firms. However, contrary to the view of regulators (e.g., European Commission), this effect is not necessarily induced by mixed joint audit arrangements, specifically in a context where the number of “Big” suppliers is stable. We therefore investigate more thoroughly the role of Mid-Tier auditors among the non-Big 4 to get a more precise understanding of the market dynamics.

Table 5 reports estimation results for the Level 2 of the analysis, considering joint audit combinations based on a two-tiered market segmentation (Big 4, Mid-Tier, and Local auditors). We restrict the number of possible states to five (instead of six) given the particularly low occurrence of joint audits involving two Mid-Tier auditors (see “The Markov Chain Modeling Applied to the French Joint Audit Context” section).

Table 5.

Estimation Results in a Two-Tier Market Segmentation.

Type of joint audit combination	Period 1997-2003			Period 2003-2009
	π_i	R_i	A_i	π_i	R_i	A_i
Two Big 4	0.2265	0.7100	0.2098	0.1875	0.7842	0.1363
One Big 4 + One Mid-Tier	0.2725	0.7348	0.2399	0.2687	0.8082	0.1836
One Big 4 + One Local	0.2853	0.6920	0.3321	0.2772	0.7216	0.3405
One Mid-Tier + One Local	0.1449	0.6959	0.1275	0.1749	0.6571	0.2252
Two Local	0.0708	0.5383	0.0907	0.0917	0.6201	0.1145
Sum of squares		0.07620			0.03098

The long-term projection regarding the two Big 4 joint audits is consistent with the one depicted under the three-state model, that is, a four-point drop in steady-state representativeness from the concentration to the stability period, associated with a loss in attractiveness.

With regard to mixed joint audits (i.e., a Big 4 paired with a Mid-Tier or with a Local auditor), both combinations post a slight decrease in representativeness between the two periods. However, their long-term position has different underlying rationales: The performance of Big 4-Mid-Tier combinations relies primarily on the ability to retain audit clients, whereas the performance of Big 4-Local combinations is mostly attributable to their ability to attract audit clients. This opposition is even more striking based on the stability period simulations. This is consistent with the bilateral monopoly nature of the audit industry (DeAngelo, 1981): Big 4-Mid-Tier combinations impose higher switching costs to audit clients—probably due to the stronger brand name of the audit college and/or to the larger size of audit clients opting for these combinations—as compared with Big 4-Local combinations.

Finally, both joint audit combinations involving only non-Big 4 auditors (a Mid-Tier paired with a Local, or two Local auditors) tend to improve their long-term position in the stability period, as compared with the concentration period. The performance patterns do differ though: The Mid-Tier-Local performance owes only to gains in relative attractiveness, whereas joint audits involving two Local auditors improve both their retention and attractiveness factors. However, joint audits involving only Local auditors remain by far the least demanded option, with only 9% of audit clients in the long run versus nearly twice as more when a Mid-Tier auditor gets in the college. In brief, tandems made of two Local auditors perform better in a period of stability, although they cannot compete with Mid-Tier-Local combinations. These findings thus support the argument that Mid-Tier audit firms play an important part among non-Big 4 auditors in the French joint audit environment. This part probably owes to differentiated audit production capacities that smaller Local auditors cannot offer.

Research Design and Results of the Joint Audit Switching Model

Specification of the Joint Audit Switching Model

We investigate the change in joint audit combinations by testing a dynamic model over the 2003-2009 stability period. This change analysis is consistent with the previous Markov chain modeling, in which companies can decide to keep the same joint audit combination or to change for a different one. The model integrates three categories of predictors: agency costs of financing, audit complexity and risk, and supply side effects. It is formally specified as follows, with the Δ_ prefix designating the change (end-of-period minus beginning-of-period) in predictor values:

\begin{matrix} Change in J A = β 0 + β 1 \cdot Δ_CHS + β 2 \cdot Δ_TotDebt + β 3 \cdot Issue + β 4 \cdot Δ_LnAsset \\ + β 5 \cdot Δ_IntanGW + β 6 \cdot Δ_LongAcc + β 7 \cdot Δ_ShortAcc \\ + β 8 \cdot Δ_ForSales + β 9 \cdot Δ_ROA + β 10 \cdot Andersen \\ + β 11 \cdot Salustro + ε . \end{matrix}

The change in joint audit (Change in JA) is a categorical variable coded 0 if the firm kept the same joint audit combination during the period and takes a nominal value to identify each type of change otherwise. Three variables capture the agency-related demand for audit quality/reputation:

CHS: The percentage of Closely Held Shares (from Worldscope), that is, the shareholding of insiders and identified blockholders;

TotDebt: The ratio of total debt to total assets;

Issue: A dummy variable coded 1 if common stock has increased by more than 10% over the period, and 0 otherwise.

The next six variables capture audit clients’ complexity and risk, and therefore relate to a demand for audit technology:

LnAsset: The natural log of year-end total assets;

IntanGW: The ratio of goodwill and other intangibles to total assets;

LongAcc: The ratio of long-term accruals (depreciation, amortization, and depletion expenses) to sales;

ShortAcc: The ratio of short-term accruals (obtained by difference between total accruals and long-term accruals) to sales;

ForSales: The percentage of foreign sales divided by total sales;

ROA: The return on assets (net income divided by total assets).

Finally, we also integrate in the model two supply side effects associated with audit firm mergers that could have triggered auditor switching decisions in the period of analysis:

Andersen: A dummy variable coded 1 if the audit client was audited by the ex-Andersen network French member (Barbier Frinault), and 0 otherwise¹³;

Salustro: A dummy variable coded 1 if the audit client was audited by Salustro Reydel before 2005, and 0 otherwise. Salustro Reydel was a significant Mid-Tier French audit firm but was acquired by KPMG in 2005.¹⁴

All the accounting data are extracted from Worldscope for fiscal years 2003 and 2009. The final dataset consists of 400 audit clients.¹⁵

Empirical Results

Table 6 reports the results from a multinomial logit analysis of the changes in joint audit combinations. The base outcome is the group of firms which did not change their joint audit combination over the period.

Table 6.

Multinomial Logit Regression Results for the Joint Audit Change Model.

Explanatory variables	Two Big 4 → Mixed(n = 10)		Mixed → Two Big 4(n = 15)		Mixed → Two non-Big 4(n = 25)		Two non-Big 4 → Mixed(n = 34)
Explanatory variables	Coefficient	z statistic	Coefficient	z statistic	Coefficient	z statistic	Coefficient	z statistic
Intercept	−5.00***	−6.33	−3.58***	−7.69	−2.58***	−8.26	−2.32***	−8.23
Δ_CHS	0.38	0.22	1.90	1.34	2.44**	2.17	0.81	0.79
Δ_TotDebt	1.36	0.60	−2.81*	−1.80	0.93	0.78	−0.47	−0.50
Issue	1.49*	1.79	0.39	0.67	−0.28	−0.57	0.37	0.92
Δ_LnAsset	−0.28	−0.69	−0.34	−0.94	−0.42	−1.45	0.12	0.66
Δ_IntanGW	−2.17	−1.30	1.81	1.25	1.16	1.05	−1.41	−1.45
Δ_LongAcc	1.34	0.41	−3.24	−1.37	−3.98**	−2.45	−0.79	−0.54
Δ_ShortAcc	−0.20	−0.21	1.12**	2.28	0.91**	2.17	0.65*	1.80
Δ_ForSales	4.13***	2.63	1.49	1.12	1.74	1.67	0.28	0.34
Δ_ROA	3.06	1.26	−1.74	−0.71	1.53	0.96	0.40	0.28
Andersen	1.54*	1.91	0.81	1.16	−1.07	−1.01	−15.51	−0.01
Salustro	1.26	1.51	−16.23	0.00	−0.69	−0.64	−16.21	−0.01

Note. The panel is composed of 400 audit clients observed in 2003 and 2009. The base outcomes are clients who did not change their joint audit combination during that period (N = 316). Model specification: Change in JA = β0 +β1·Δ_CHS+β2·Δ_TotDebt+β3·Issue+β4·Δ_LnAsset+β5·Δ_IntanGW+β6·Δ_LongAcc+β7·Δ_ShortAcc+β8·Δ_ForSales+β9·Δ_ROA+β10·Andersen+β11·Salustro+ϵ. Overall model quality indicators: Chi-square = 78.05 (p = .0012); pseudo-R² = .1244. CHS = Closely Held Shares; TotDebt = The ratio of total debt to total assets; LnAsset = the natural log of year-end total assets; IntanGW = the ratio of goodwill and other intangibles to total assets; LongAcc = the ratio of long-term accruals (depreciation, amortization, and depletion expenses) to sales; ShortAcc = the ratio of short-term accruals (obtained by difference between total accruals and long-term accruals) to sales; ForSales = the percentage of foreign sales divided by total sales; ROA = return on assets; JA = joint audit.

*, **, and *** are 10%, 5%, and 1% significance levels, respectively (two-tailed).

In such a dynamic perspective, regression coefficients indicate the determinants of the performance of the three joint audit combinations. Overall, the performance of mixed and two non-Big 4 combinations—as underlined by the Markov chain model—can be explained by changes in financing and audit complexity characteristics of audit clients, as well as by supply side market concentration effects.

More specifically, the attractiveness of mixed versus two Big 4 joint audits is related to equity issuance (Issue, p < .10), greater international operations (Δ_ForSales, p < .01), and the fact that the company was audited by Arthur Andersen before 2003 (Andersen, p < .10). The first two effects appear to be at odds with the conventional agency and audit complexity literature (although the equity issuance effect is weakly significant). The latter may account for detrimental reputation effects on the Big 4 “label” caused by the Andersen failure. In addition, greater debt financing seems to reduce the propensity to adopt a two Big 4 joint audit, thus contributing to the retention of a mixed combination. In fact, only an increase in short-term accruals seems to play against the use of mixed joint audits, in favor of both two Big 4 and two non-Big 4 joint audits (Δ_ShortAcc, p < .05 in both cases). This particular two-way effect could be explained by differences in audit technology between the coauditors in the case of mixed joint audits (Sirois & Simunic, 2011), which incurs potential free-riding by the non-Big 4, and a lower precision of audit evidence (Deng, Lu, Simunic, & Ye, 2014). The appointment of a second Big 4 when short-term accruals increase is also consistent with a demand for higher credibility of reported earnings (Francis et al., 1999).

Regarding the performance of joint audits involving two non-Big 4, the underlying economics is somewhat different. Basically, the attractiveness of two non-Big 4 joint audits is associated with increases in ownership concentration and in short-term accruals (Δ_CHS and Δ_ShortAcc, both at p < .05). French companies have greater propensity to dismiss their only Big 4 auditor when they get exposed to fewer agency costs of equity. This is consistent with prior cross-sectional empirical findings documenting that French family-owned companies avoid the use of Big 4 auditors (Francis et al., 2009; Marmousez, 2012).

To summarize, we find little economic support for the demand of two Big 4 joint audit combinations in the French setting. Proxies related to agency costs of financing play in favor of mixed joint audits and also more importantly in favor of two non-Big 4 combinations when ownership gets more concentrated. However, these results should be taken cautiously. The 6-year window imposed by the French research context may introduce some noise in the estimation of regression coefficients. More specifically, the change in joint audit can happen any year during the observation period, whereas the change in economic conditions is measured between both ends of that period (i.e., financial years 2009 minus 2003). This misalignment may explain some of the unexpected results we get, regarding audit complexity factors in particular.

Discussion and Concluding Remarks

The European Commission (2014) encourages joint audits as a possible solution to mitigate the audit market concentration, specifically the “systemic” position of Big 4 auditors, and to favor the emergence of Mid-Tier audit firms offering an alternative to the Big 4. This belief led the Commission to encourage mixed joint audit combinations, that is, joint audits pairing a Big 4 with a non-Big 4 auditor.

We investigate this proposal by modeling the dynamics of the French audit market, where virtually all listed companies are subject to a joint audit requirement, and where audit clients are free to choose any combination of coauditors that are independent from one another. We adapt the Markov chain model (Comunale & Sexton, 2003; Duxburry et al., 2007) to joint audit combinations to infer the relative attractiveness of each type of joint audit combination both in the short term and in the long run. We consider two periods representing different market contexts: a significant market concentration period (1997-2003) during which the Big 6 merged into the Big 4, followed by a stability period (2003-2009) characterized by a Big 4 audit market.

Our main findings stand at three levels. First, in a stability versus concentration context, the two Big 4 combination comes out less attractive in the long run, and the two non-Big 4 one as more attractive. Specifically, based on market share transfers observed during the 2003-2009 stability period, 19% (27%) of audit clients should, in equilibrium, choose two Big 4 (two non-Big 4) auditors, as compared with 23% (20%) for the earlier 1997-2003 concentration period. This change in steady-state probabilities could be consistent with the presence of market dominance effects at the time of the Big “N” megamergers and differs with findings reported around the 1998 PwC merger (Duxburry et al., 2007). Second, the mixed combination remains the most attractive, with 54% of audit clients in the long run. This steady-state figure is consistent with current empirical observations (e.g., Francis et al., 2009; Marmousez, 2012) and with the European Commission’s recommendation. More specifically, in this configuration, the fact that the non-Big 4 auditor belongs to the Mid-Tier or Local auditors does not make a real difference: Both combinations perform equivalently in the long run. Third, conversely, our findings suggest that when two non-Big 4 auditors are involved as joint auditors, the fact that one of them belongs to the Mid-Tier results in a more comfortable long-term performance, with 17% of audit clients versus only 9% for joint audits involving two Local auditors. Hence, overall, the French joint auditing environment stands as a viable solution to open the market to non-Big 4 auditors, and Mid-Tier auditors take an important part in sustaining the long-term market performance of non-Big 4 auditors.

We also investigate the firm-level economic determinants driving changes in joint audit combinations. To this end, we estimate a dynamic joint audit switching model over the stability period integrating three categories of variables: agency costs of financing, audit complexity and risk, and supply side effects. Our findings complete and corroborate the inferences from the Markovian analysis. Changes in audit clients’ agency costs are associated with the performance of mixed and two non-Big 4 joint audit combinations; the demand for two non-Big 4 combinations notably increases with ownership concentration. Little economic support is found for the choice of two Big 4 combinations in the French setting.

These findings globally support the positive effects of a mandatory joint auditing system on the audit market dynamics, in line with the position expressed by the European Commission after the 2010 Green Paper. As such, our study contributes to the debate among regulators and professional bodies regarding the benefits and costs of joint auditing.

The Markovian approach of market dynamics is subject to limitations, the principal one being the assumption of a “closed” system that ignores potential new entrants, and specifically the arrival of new audit clients (e.g., newly listed companies). However, the attractiveness parameter reflects the ability of a given combination to gain new audit clients from existing alternative combinations. The high relative attractiveness factor of mixed joint audits (0.89 in the three-state model) is likely to support a dominant long-term position of that combination if new entrants were to be considered. In addition, economic factors such as a damaged reputation of the “Big N” after the 2002 Andersen demise or the higher cost of a two Big 4 college in a period of crisis (André, Broye, Pong, & Schatt, 2016) could also have contributed to observed changes in market shares over this period.

Footnotes

Acknowledgements

The authors are grateful to the Editor-in-Chief, Bharat Sarath; the anonymous reviewers; seminar participants at the 2014 European Accounting Association Annual Congress (Tallinn); and at the 2014 French Accounting Association Annual Conference (Lille) for their suggestions and comments. Charles Piot also thanks the CNRS (French National Center for Scientific Research) for administrative support (CNRS delegation).

Authors’ Note

A first version of this article was completed during the authors’ sabbatical year at Beedie School of Business, Simon Fraser University (Vancouver).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors acknowledge financial support from Grenoble Ecole de Management and from Université Grenoble Alpes (Université Pierre Mendès-France Grant PRS No. 67).

Notes

References

The American Assembly. (2005, May 23). The future of the accounting profession: Auditor concentration. New York, NY: Columbia University.

André

Broye

Pong

Schatt

(2016). Are joint audits associated with higher audit fees?European Accounting Review, 25, 245-274. doi:10.1080/09638180.2014.998016

Assemblée Nationale. (1984). Journal Officiel de la République Française – Débats parlementaires, 1ère séance du 03 février 1984 [Official Journal of the French Republic Parliamentary debates, 1st meeting on 3 February 1984], 485-506

Audousset-Coulier

(2008). La publication des honoraires d’audit par les sociétés françaises: deux études de déterminants. Les déterminants du caractère volontaire de la publication des honoraires d’audit et les déterminants des honoraires d’audits publiés [The disclosure of audit fees by French companies: Two studies of determinants. The determinants of voluntary disclosure of audit fees and the determinants of published audit fees] (Doctoral dissertation). HEC Paris, France.

Ballas

Fafaliou

(2008). Market shares and concentration in the EU auditing industry: The effects of Andersen’s demise. International Advances in Economic Research, 14, 485-497.

Bédard

Piot

Schatt

(2014). An evaluation of the French experience with joint auditing. Retrieved from http://ssrn.com/abstract=2165595

Broye

(2001). Choix des cabinets d’audit et évaluation des titres introduits au second marché [Auditor choice and the valuation of new issues]. Comptabilité– Contrôle – Audit, 7(2), 105-124.

Broye

(2007). Concentration du marché de l’audit: un état des lieux [Audit market concentration in France: Current situation]. Revue Française de Comptabilité, 399, 34-37.

Broye

Schatt

(2012). Comment ont évolué les honoraires d’audit en France pendant la crise financière? [How did audit fees evolved in France during the financial crisis?]. Revue Française de Comptabilité, 456, 24-27.

10.

Choi

M. S.

Zéghal

(1999). The effect of accounting firm mergers on international markets for accounting services. Journal of International Accounting, Auditing & Taxation, 8, 1-22.

11.

Comunale

C. L.

Sexton

T. R.

(2003). Current accounting investigations: Effect on Big 5 market shares. Managerial Auditing Journal, 18, 569-576.

12.

DeAngelo

L. E.

(1981). Auditor independence, “low balling,” and disclosure regulation. Journal of Accounting & Economics, 3, 113-127.

13.

DeFond

M. L.

(1992). The association between changes in client firm agency costs and auditor switching. Auditing: A Journal of Practice & Theory, 11(1), 16-31.

14.

Deng

Simunic

D. A.

(2014). Do joint audits improve or impair audit quality?Journal of Accounting Research, 52, 1029-1060. doi:10.1111/1475-679X.12060

15.

Dopuch

Simunic

(1980). The nature of competition in the auditing profession. In Buckley

J. W.

Easton

J. F.

(Eds.), Regulation and the accounting profession (pp. 77-94). Belmont, CA: Lifetime Learning.

16.

Duxburry

Moizer

Wan Mohamed

(2007). Effects of the PricewaterhouseCoopers merger on the UK audit services market. Managerial Auditing Journal, 22, 121-138.

17.

European Commission. (2010, October 13). Green Paper “Audit policy: Lessons from the crisis.”Brussels, Belgium: Author.

18.

European Commission. (2014, May 27). Regulation (EU) No 537/2014 of the European Parliament and of the Council of 16 April 2014 on specific requirements regarding statutory audit of public-interest entities and repealing Commission Decision 2005/909/EC. Official Journal of the European Union, 2014(L158), 77-112.

19.

Financial Reporting Council. (2010, June). Choice in the UK audit market—Fifth progress report. London, England: Author.

20.

Francis

J. R.

(2011). A framework for understanding and researching audit quality. Auditing: A Journal of Practice & Theory, 30, 125-152.

21.

Francis

J. R.

Maydew

E. L.

Sparks

H. C.

(1999). The role of Big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory, 18(2), 17-34.

22.

Francis

J. R.

Richard

Vanstraelen

(2009). Assessing France’s joint audit requirement: Are two heads better than one? Auditing: A Journal of Practice & Theory, 28(2), 35-63.

23.

Francis

J. R.

Stokes

Anderson

D. J.

(1999). City markets as a unit of analysis in audit research and the re-examination of Big 6 market shares. Abacus, 35, 185-206.

24.

Francis

J. R.

Wilson

E. R.

(1988). Auditor changes: A joint test of theories relating to agency costs and auditor differentiation. The Accounting Review, 63, 663-682.

25.

French National Assembly (1984). Official Journal of the French Republic - Parlementary debates, 1st session of February 3rd, 1984

26.

Ghosh

Moon

(2005). Auditor tenure and perceptions of audit quality. The Accounting Review, 80, 585-612.

27.

Gul

F. A.

Fung

S. Y. K.

Jaggi

(2009). Earnings quality: Some evidence on the role of auditor tenure and auditors’ industry expertise. Journal of Accounting & Economics, 47, 265-287.

28.

Hamilton

J. D.

Raj

(Eds.). (2002). Advances in Markov-switching models. Heidelberg, Germany: Physica-Verlag.

29.

House of Lords. (2010). Call for evidence—Auditors: Market concentration and their role. Available from http://www.parliament.uk

30.

Ivancevich

S. H.

Zardkoohi

(2000). An exploratory analysis of the 1989 accounting firm megamergers. Accounting Horizons, 14, 389-401.

31.

Iyer

V. M.

Iyer

G. S.

(1996). Effect of Big 8 mergers on audit fees: Evidence from the United Kingdom. Auditing: A Journal of Practice & Theory, 15, 123-132.

32.

Johnson

W. B.

Lys

(1990). The market for audit services—Evidence from voluntary auditor changes. Journal of Accounting & Economics, 12, 281-308.

33.

Le Vourc’h

Morand

(2011, November 9). Final report: Study on the effects of the implementation of the acquis on statutory audits of annual and consolidated accounts including the consequences on the audit market. Paris, France: ESCP Europe.

34.

Lim

C.-Y.

Tan

H.-T.

(2010). Does auditor tenure improve audit quality? Moderating effects of industry specialization and fee dependence. Contemporary Accounting Research, 27, 923-957.

35.

Marmousez

(2012). Étude d’une exception française: Les déterminants du choix de la composition du collège de commissaires aux comptes [Examining a French exception: The determinants of the choice of joint-auditors]. Comptabilité Contrôle Audit, 18, 121-152.

36.

Minyard

D. H.

Tabor

R. H.

(1991). The effect of Big Eight mergers on auditor concentration. Accounting Horizons, 5, 79-90.

37.

Oxera Consulting. (2006, April). Competition and choice in the UK audit market (Report prepared for Department of Trade and Industry and Financial Reporting Council). Retrieved from http://www.oxera.com/Latest-Thinking/Publications/Reports/2006/Competition-and-choice-in-the-UK-audit-market.aspx

38.

Piot

(2003). Coûts d’agence et changements de commissaire aux comptes: Une approche empirique [Agency costs and auditor changes: Some evidence in France]. Comptabilité– Contrôle – Audit, 9(2), 5-30.

39.

Piot

(2007). Auditor concentration in a joint auditing environment: The French market 1997-2003. Managerial Auditing Journal, 22, 161-176.

40.

Piot

(2008). Concentration et compétitivité du marché de l’audit en France: Une étude longitudinale 1997-2003 [Concentration and competitiveness of the French audit market: A longitudinal study 1997-2003]. Finance Contrôle Stratégie, 11(4), 31-63.

41.

Piot

Janin

(2007). External auditors, audit committees and earnings management in France. European Accounting Review, 16, 429-454.

42.

Pong

C. K. M.

Burnett

(2006). The implications of merger for market share, audit pricing and non-audit fee income: The case of PricewaterhouseCoopers. Managerial Auditing Journal, 21(1), 7-22.

43.

Ratzinger-Sakel

N. V. S.

Audousset-Coulier

Kettunen

Lesage

(2013). Joint audit: Issues and challenges for researchers and policy makers. Accounting in Europe, 10, 175-199.

44.

Sirois

L.-P.

Simunic

D. A.

(2011). Auditor size and audit quality revisited: The importance of audit technology. Retrieved from http://ssrn.com/abstract=1694613

45.

Thavapalan

Moroney

Simnett

(2002). The effect of the PricewaterhouseCoopers merger on auditor concentration in Australia: A note. Accounting & Finance, 42, 153-167.

46.

Tonge

S. D.

Wootton

C. W.

(1991). Auditor concentration and competition among the large public accounting firms: Post-merger status and future implications. Journal of Accounting and Public Policy, 10, 157-172.

47.

U.S. Government Accountability Office. (2003). Public accounting firms—Mandated study on consolidation and competition. Washington, DC: Author.

48.

Wolk

C. M.

Michelson

S. E.

Wootton

C. W.

(2001). Auditor concentration and market shares in the US: 1988-1999 a descriptive note. British Accounting Review, 33, 157-174.