Strategic entry deterrence in the audit industry: Evidence from the merger of professional accounting bodies

1 INTRODUCTION

New entrants to a market will increase competition, threatening and disrupting the operations of incumbent firms. Incumbents are thus unlikely to welcome new entrants. Instead, they will resort to various ways to either deter entry before it occurs or drive out entrants ex post. As a strategy to deter ex-ante entry, limit pricing where incumbents lower their preentry prices in response to an increase in entry threat gains support from early theoretical research (e.g., Bain, 1956) as well as later theoretical research incorporating information asymmetry (Milgrom & Roberts, 1982). Compared to theoretical work on limit pricing, empirical studies are less conclusive and relatively sparse with contradicting evidence from survey studies (Singh et al., 1998; Smiley, 1988), and the supporting evidence mostly concentrated in consumer goods industry such as airlines and cable TV industry (Goolsbee & Syverson, 2008; Savage & Wirth, 2005; Seamans, 2013). We provide new evidence from a professional service industry where we rely on a quasi-experiment to identify the entry threat.

Different from the consumer goods such as airlines and TV programs that consumers can directly derive utility from and compare, services such as independent audit services are mainly purchased for signaling quality and providing assurance. Given that the service requires significant expertise, the quality of the service is difficult to evaluate and compare.1 Therefore, the demand of the service relies on the supplier's reputation. It is thus not clear whether the limit pricing documented in consumer goods industries as an entry deterrence strategy will be adopted in a professional service industry.

Professional accountants working in audit firms provide assurance services on firms’ financial statements to investors in capital markets. Thus, the audit industry plays a vital role to firms, capital markets and the overall economy. Auditing requires expertise and thus the audit industry is regulated with a limited number of service providers licensed to operate in the market. Since the 1970s, regulators have been increasingly concerned that the high market concentration could give audit firms monopolistic or oligopolistic pricing powers over their clients. In light of this concern, the potential effects of entry threats on incumbent audit firms’ pricing are in itself a topic that warrants academic attention. In this paper, we investigate whether and when entry threats affect audit pricing using a quasi-experimental setting, where a profession-wide merger at the level of professional accounting bodies induced an exogenous increase in the expected future entries to incumbent audit firms.

The recent staggered approvals by the Canadian provincial legislations to merge Canada's three professional accounting bodies and their respective professional designation offer a unique quasi-experimental setting to identify an exogenous increase in the threat of entry. From 2012 to 2014, the three Canadian professional accounting bodies—the Chartered Accountants (CAs), the Certified Management Accountants (CMAs) and the Certified General Accountants (CGAs)—unified into a new designation, Chartered Professional Accountants (CPAs) (hereafter, the CPA merger). Before the CPA merger, CAs generally enjoyed the reserved privilege of public accounting to themselves. By unifying CAs with the other accounting bodies, the merger relaxes the restrictions to enter the public-accounting industry for two reasons. First, after the CPA merger, the candidacy to apply for public-accounting licenses is no longer restricted to CAs, rather it is extended to all candidates of the newly united CPA designation. Second, the merger harmonizes the certification requirements and thus results in a new requirement level that is lower than those previously for CAs.

Labor economics theory predicts that less restrictive occupational requirements attract more individuals to enter a regulated profession (Lott, 1987). Accordingly, we expect that the approval of the CPA merger leads to an expected increase in the number of future public accountants. An expected increase in the number of public accountants also translates into more qualified individuals who will become eligible to invest and open their own firms. Due to these potential impacts, the approval of the CPA merger may raise the probability of future entry and thus lead to an exogenous increase in the entry threat to the incumbent audit firms.2

Using a difference-in-differences design with client-firm, audit-firm, province and year-month fixed effects to exploit the staggered approval of the CPA merger, we examine whether and when (i.e., conditions under which) audit firms, facing an increased probability of future entries, will preemptively lower audit fees to deter potential entries.3 One audit firm can have multiple locations across Canadian provinces, and each firm location can have multiple engagement partners. The partners are required to be designated in their province. For a given auditor and client, only when the auditor's local province (i.e., the province in which her office resides) approves the merger, do we consider the auditor's submarket to be shocked with an increase in entry threats. The same auditor's submarkets in other provinces are used as controls. This alleviates the possibility that changes in fees relate to any specific auditor-firm-wide factors. In addition, even though the approval of the CPA merger increases entry threats to incumbent audit firms, it is unlikely to result in actual entries immediately. Thus, the audit-fee changes upon the merger approval are more likely to reflect audit firms’ preemptive strategies against potential entries than their reactive actions to actual entrants. The staggered-shock design also significantly reduces the likelihood of having a confounding event that explains the treatment effect (Bertrand & Mullainathan, 2003).

Incumbents, threatened by potential future entries, could either lower their prices from a previous monopolistic/oligopolistic level to signal a lower cost (Milgrom & Roberts, 1982) or increase their customer base (Klemperer, 1987) in order to deter entry. In the context of the audit industry, the market is oligopolistic with limited existing service providers and segmented into submarkets with differential entry barriers.4 Further, information is far from complete. The pricing strategy and the cost structure are complicated, as they vary by firms and even differ by engagements for the same partner at a given firm. This makes it difficult to gauge an incumbent firm's profitability. To the extent that some audit firms are threatened by an increase in the probability of new entrants due to the CPA merger, they will be more inclined to agree to fee concessions to preemptively deter entry. Consequently, we expect the CPA merger to lead to lower audit fees.

We further explore the segmented nature of the audit industry to understand the conditions under which the merger leads to significant fee reductions. First, we explore factors that can strengthen the credibility of the threat. We expect that incumbent audit firms in a submarket with high existing entry barriers find the merger to induce a less credible threat. For example, for the market of Big-4 auditors and large auditees where existing barriers are high due to the high requirement of reputation that can only be built over an extended period of time, it is unlikely that the threat induced by the CPA merger would reduce the existing entry barriers to a sufficiently low level to trigger new entrants. As these firms are more protected by the high level of existing entry barriers, there is little need to deter entry. Accordingly, we hypothesize that the fee reduction due to the merger will be mitigated either partially or completely by high entry barriers that make the threat less credible or not credible at all.

Second, we explore cross-sectional variations in the strength of the threat by exploiting institutional differences among Canadian provinces. Ex ante, we expect the shock to be stronger when there are more CMA and CGA members relative to the CA members sitting on the CPA governance boards (i.e., the (CMA + CGA)/CA ratio is higher).5 First, when there are more CMAs and CGAs in a province, we expect a larger increase in the candidate pool for public-accounting licenses.6 Second, the composition indicates the relative bargaining powers of the former CA, CMA and CGA bodies when setting future governing policies. With fewer CAs on the board, new policies are less likely to protect the CAs’ privilege over the CMAs and CGAs, in which case we expect the merger's effect on entry threats to be stronger. We thus predict the fee reduction to be larger in provinces in which fewer CAs relative to CMAs and CGAs are on the CPA governance board.

Our findings support the hypotheses. First, we find that the approval of the CPA merger on average leads to a 7% ($4,191) reduction in audit fees, consistent with at least some incumbent audit firms agreeing to fee concessions in order to deter potential entrants. We further find that the reduction in fees is not significant for the submarket with Big-4 auditors, and is significantly mitigated for submarkets with large clients. In addition, our results show that the effect of the CPA merger is significant only when the increase in entry threats is expected to be greater in provinces in which CMAs and CGAs have a greater presence relative to the CAs on the governing board. Notably, the cross-sectional analyses on the threat intensity provide validation for the use of the CPA-merger setting to identify entry threats. We further provide additional validation of our setting using Big-4 auditors and large clients as placebo tests. In additional analyses, we explore potential mechanisms for audit firms’ strategic entry deterrence and find that the reduction in fees is significant only for nonindustry leaders (and is not significant for industry leaders). This result indicates that auditees’ bargaining power and auditors’ concession are likely mechanisms. Finally, we do not find audit quality changes in response to the approval of the CPA merger.

This paper offers several contributions. First, this study contributes to the economics and management literature on entry deterrence (e.g., Bresnahan & Reiss, 1991; Goolsbee & Syverson, 2008; McCann & Vroom, 2010; Parise, 2018; Prince & Simon, 2015; Sigfried & Evans, 1994; Simon, 2005; Whinston & Collins, 1992). Prior research shows that the incumbent firms in the airline industry reduce prices, lower service quality and extend debt maturity to longer periods in response to the rise of a credible entry threat (Goolsbee & Syverson, 2008; Parise, 2018; Prince & Simon., 2015). Seamans (2012) further finds that incumbent firms in cable television use limit pricing in a way consistent with the degree of information asymmetry as predicted in Milgrom and Roberts (1982). The existing evidence is mostly concentrated in consumer industries. Different from prior research, we examine entry deterrence in an important professional service industry where information asymmetry is high but the demand of the service is highly dependent on a supplier's reputation.

Second, our evidence adds to the debate about the competitiveness of the audit industry that has long since been a concern for regulators. Although we have accumulated some knowledge about the effect of audit-firm exits, due to mergers, from both anecdotal evidence (e.g., Arthur Andersen's bankruptcy) and archival research (e.g., Ding & Jia, 2012), little is understood about the threat of potential entrants. The CPA mergers across different provinces offer a unique quasi-experiment to study how an exogenous increase in entry threats causally affects audit pricing. In addition, the audit industry in Canada is segmented into multiple audit submarkets with differential entry barriers, and hence, provides a useful setting to study when preemptive entry deterrence arises as an equilibrium choice.

Finally, the empirical evidence in this paper is of potential interest to regulators and rule-setting agencies in the context of the recent US regulatory reform to harmonize cross-state occupational licensing rules. Although our setting is about the unification of licensing requirements across job functions and the policy debate in the United States is about the unification of licensing requirements across states, the underlying similarities between the two lie in an increase in labor mobility and a relaxation of occupation restrictions. Both these factors are identified among the most important deterrents to foster competition (Council of Economic Advisors, 2016).7 We believe that our study provides timely and relevant evidence on effects of the unification of licensing requirements on professional service fees.8

2 EMPIRICAL SETTING

2.2 The merger of the Canadian professional accounting bodies and their respective designations

The recent provincially legislated unification of multiple Canadian accounting designations has significantly changed the public-accounting licensing landscape in Canada. Historically, the accounting profession in Canada consisted of three professional bodies, each with their own professional designations and certification requirements—the CAs, CMAs and CGAs. The certification requirements for CAs were much more closely aligned with the requirements to obtain public-accounting licenses than those for CMAs and CGAs.10

In contrast, the CGA and CMA designations had less restrictive education requirements and lower training requirements, and were generally regarded as insufficient to practice public accounting, and thus were more likely to be chosen by candidates who would accept a smaller set of career options in order to incur lower costs. In summary, the premerger certification requirements led to a situation where CMA and CGA members were less likely to obtain public licenses, and practically reserved the privilege of practicing public accounting for the CA designation.11

The CPA merger was initiated by the governing bodies of the CAs, CMAs and CGAs with a purpose to harmonize different certification requirements and to promote the accounting profession. In October 2011, Quebec was the first province to release an update on the CPA merger, followed by British Columbia, Alberta, Newfound and Labrador, Saskatchewan, Manitoba, Nova Scotia, Yukon, Prince Edward Island and lastly Ontario. By the end of 2014, all provincial governments had officially approved the CPA merger and amended provincial legislations in their jurisdictions (see Table 1 for approval dates).12 There are two points worth mentioning. First, a province (e.g., Ontario) has full independence over its merging decision, and thus, the province's merger decision is independent of other provinces. Thus, ex ante, there is significant uncertainty regarding whether and when the merger would be passed, as well as which accounting bodies would join the merger. In fact, several provinces had been back and forth on the merger proposal. For example, the merger in Quebec in 2012 was the fifth attempt since 1973 that Quebec's accounting profession tried to merge forces.13 As another example, the CGA and CMA in Ontario announced that they were withdrawing from merger discussions in that province on May 15, 2012.14 Thus, it was difficult to anticipate the voting results or approval dates. Second, the fact that the merger was ultimately approved does not mean that it was the existing CAs who favored the change. In fact, a designation's (i.e., CA's, CMA's and CGA's) decision to join the merger is made through members’ votes, and thus, independent of other designations. Alberta CAs initially voted no on the merger. Therefore, the result is more likely to represent a compromise between different parties rather than a unanimous decision. The exception is Quebec, in which case the merger decision was voted by the board of directors (rather than members) of each of the three designations.

TABLE 1. Approval dates of the CPA merger by Canadian provinces

Province	Province abbreviation	Approval date
Quebec	QC	May 16, 2012
Saskatchewan	SK	June 6, 2013
Newfoundland and Labrador	NL	June 21, 2013
Ontario	ON	June 27, 2013
Alberta	AB	June 28, 2013
British Columbia	BC	September 25, 2013
Manitoba	MB	January 30, 2014
New Brunswick	NB	April 23, 2014
Nova Scotia	NS	May 24, 2014

• Note: This table presents the approval dates of the CPA merger in Canadian provinces for firms in our final sample. A detailed timeline of the key developments in Canadian provinces is presented separately in Table A1.

After the merger, the unified CPA designation also has a public-accounting stream and a nonpublic-accounting stream, similar to the structure of the previous CA designation. The certification requirements for the unified CPA designation are similar to those of the previous CA designation in that the public-accounting stream is closely aligned with public-accounting licensures, and the nonpublic-accounting stream has a similar education requirement with more flexible training options. The merger mandated all registered and prospective CA, CMA and CGA candidates who had not yet received their designations, to be moved to the CPA program, thus making the CMA and CGA candidates equally eligible as the CA candidates to enroll in the CPA public-accounting stream.15

The staggered passage of the CPA unification act leads to the expectation of a gradual growth in qualified public accountants for two reasons. First, the CPA merger extends the candidacy to enter the public-accounting stream from a restricted group of CAs to all candidates of the unified CPA designation. It leads to higher enrollment in the CPA public-accounting stream, and thus increased the number of eligible candidates who can automatically qualify for public licenses when receiving professional designations. Second, a key mandate of the merger is to harmonize the differences in certification requirements among CA, CGA and CMA designations. The unification process results in a certification requirement level that is below that of the previous CA designation. For instance, the minimum university average grade requirement reduces to 65% under the new CPA program from the previous CA requirement of 70% in Ontario.16 Further, certain in-class training, which was the main form of the CA education, is replaced by online education.

According to labor economics theory, less restrictive licensing requirements will draw more individuals to enter the profession. To validate this assumption, we collect statistics from CPA annual reports released by provincial CPA associations. Figure 1 shows that the number of CPA members on average increased by 9% within five years after the CPA merger. According to staff members at CPA Ontario, the number of students who choose the public-accounting stream in the new CPA program exceeds the number of CA students before the merger date and the examination pass rate is higher. These individuals will eventually qualify as public accountants, which reduce salaries to a new equilibrium where the expected return of becoming a public accountant equals the cost of obtaining the license (Lott, 1987). Therefore, we expect that the approval of the CPA merger leads to an expected increase in the number of qualified public accountants.

With more qualified public accountants, more individuals will also be eligible to invest and open their own firms. Although the newly designated CPAs may not be able to immediately start firms to serve large clients, existing CPAs, especially those working at the management and senior management levels, would find it more feasible to start their own practices. Because the CPA merger will increase the supply of newly designated individuals, it thus attracts both new and more experienced CPA holders to start new businesses. The expectation of an increase in the supply of public accountants and audit firms finds support among practitioners and in statistics. A report issued by CPA Canada in 2012 identified the expected establishment of more audit firms in Canada as a major consequence of the merger. According to the statistics from CPA British Columbia's annual reports, the number of public licensed firms increases after the merger.17

In summary, we argue that the approval of the CPA merger increases the expected probability that new firms will enter the audit industry, and thus raises the level of entry threats to incumbent audit firms, especially non-Big-4 firms and those who audit smaller clients.

The unique setting ensures that the change is primarily driven by entry threats in expectation rather than an actual increase in the number of public accountants or audit firms. Any real effect of the CPA merger on audit firms (e.g., audit quality, supply of public accountants, etc.) would likely take effect in the long run, as the total time required for an individual to fully complete CPA licensing process is 4.5 years (i.e., 2 years of coursework plus 30 months of practice experience). Yet, young professionals’ decisions to start new firms are part of their career planning, and thus, fostered early in their career. Thus, it is optimal for audit firms to react quickly to strengthen existing client relations through fee concession and to signal lower profits to potential entries. Therefore, we consider the CPA merger to be a relatively clean setting to examine the causal impacts of entry threats.

3 HYPOTHESES DEVELOPMENT

When threatened by potential entries, incumbent firms can preemptively lower their prices and thus depress short-term profitability in an attempt to deter new entrants and secure higher profits in the future. With complete information, preemptive actions are not subgame perfect and are thus irrational (Selten, 1978). However, when information asymmetry exists, incumbent firms can signal lower profitability to potential entrants by lowering prices preemptively, so preemptive entry deterrence arises as an equilibrium strategy (Milgrom & Roberts, 1982; Seamans, 2013).

Next, we explore how the predicted fee reduction varies with the extent to which the CPA merger exogenously increases entry threats. Ex ante, we expect the shock to be stronger in provinces where it is more likely that new entrants will materialize. We expect the new entrants and thus the entry threats to be stronger when more CMA and CGA members serve on the governance boards relative to the CA members for two reasons. First, CPA Canada states that the board composition represents the composition of accountants (including CAs, CMAs and CGAs) in the province, so we expect more new entrants in provinces with more CMA and CGA members relative to the CA members. For example, to a province with 90% of the accountants being CAs, the entry-threat shock is weaker than in a province with 10% of the accountants of being CAs.

4 RESEARCH DESIGN

4.2 Identification strategy and empirical models

We use a difference-in-differences design with a strict fixed-effects structure that exploits the approvals of the CPA merger as a source of exogenous variations in potential entry threats to identify the effects of entry threats on audit fees.19 Our research design is similar to Bertrand and Mullainathan (2003) in that we use staggered shocks that affect different firms at different points in time as our identification strategy. Figure 2 illustrates the staggered difference-in-differences design by presenting auditor-client examples in different provinces. In particular, because several provinces in our sample approved the CPA merger in the same calendar year, we further use the timing of the audit engagements to maximize the extent of the staggered nature of the shocks in our setting. We rely on the CPA unification timeline to identify the first event for which the uncertainty of the merger is substantially resolved in each province.20 Generally speaking, the legislation process can be broadly divided into three phrases: the initiation (for public feedback), voting approval and passage of legislation. For most provinces, the first dates that the uncertainty resolves would be the approval date from members’ votes.21

There are several advantages of this research design. First, the difference-in-differences specification employed with exogenous shocks enables us to make causal inferences (e.g., Hope et al., 2017). Second, the staggered shocks allow us to have a larger and more comparable control sample. Specifically, we can use all firms with auditors in the provinces that are not approving the CPA merger at a given time as controls to those with auditors in the provinces that are approving the CPA merger at the same time. Third, using staggered shocks, we test whether the trends of fees in all provinces that have approved the CPA merger would be the same as the trends in absence of the staggered approvals, whereas a single-shock design would test whether the premerger period fees are on average lower than postmerger period fees pooled together. Thus, staggered shocks significantly reduce the likelihood of having a confounding factor that explains the treatment effect, because such a confounding factor is unlikely to be correlated with the staggered approvals of the CPA merger.22

We additionally employ an extensive fixed-effects structure to address concerns over omitted-variable biases. Perhaps most important, we use client-firm fixed effects to address potential concerns over firm-specific and time-invariant omitted variables. As such, we only compare the within-firm variations in fees. That is, we test whether fees for a given client have decreased after the CPA merger relative to its past fees. We further include audit firm, province (of the audit firm) and year-month fixed effects in our model to address potential omitted-variable biases over auditor-specific and time-invariant omitted variables, province-specific and time-invariant omitted variables, and time-specific and firm-invariant omitted variables. As such, we essentially compare whether the audit firms located in a particular province, which is approving the CPA merger, experience a different trend in fees compared to those in other provinces that are not approving the merger. We estimate the effect of the CPA merger on fees using the following empirical model:

$$\begin{equation}\ln AuditFee{s_{i,j,k,t}} = {\alpha _i}\ + {\delta _j} + {\pi _k} + {\lambda _t} + {\beta _1}Merge{r_{k,t}} + \gamma {{\rm X} _{i,j,k,t}} + {\varepsilon _{i,j,k,t}},\end{equation}$$(1)

where $$i$$ indexes auditees, $$j$$ indexes audit firms, $$k$$ indexes provinces of audit firms, $$t$$ indexes time, $$\ln AuditFees$$ denotes the natural logarithm of audit fees charged by auditor $$j$$ located in province $$k$$ to auditee $$i$$ in year $$t$$, $${{\alpha}}$$ denotes the firm fixed effects, $${{\delta}}$$ denotes the audit-firm fixed effects,$$\ \pi $$ denotes the province fixed effects and $${{\lambda}}$$ denotes the time fixed effects.

$$ Merger $$ is an indicator variable equal to one if an auditee $$i$$ files its financial statements audited by audit firm $$j$$ after the date that the CPA merger is approved by province $$k$$, and 0 otherwise, and $$ {\rm X} $$ is a vector of control variables. We define $$Merger$$ using the financial statements filing date, as it closely coincides with the auditor's sign-off and invoicing dates.23^,24 Figure 2 provides auditor-client examples to illustrate how we define $$Merger$$.

Following prior literature, we additionally include several time-varying control variables that have been shown to explain variations in audit fees (e.g., Aobdia, 2019; Bruynseels & Cardinaels, 2013; Gul & Goodwin, 2010; Hope et al., 2012). Our control variables include market-to-book ratio, leverage ratio, return on assets, issuance of new debt, loss in the current year, total assets, asset growth, current ratio, December year-end indicator and the audit-reporting lag.25 In addition, to ensure that our results are not driven by a change in audit quality, we further control for the propensity to just meet or beat a zero earnings threshold (i.e., small profits). We also control for audit risk and client complexity (proxied by the number of business segments, percentage of foreign sales to total sales and restatements), as well as auditor independence (proxied by going-concern opinions). $${\beta _1}$$estimates the effect of the CPA merger on fees, and as posited by H1, we expect $${\beta _1}$$to be negative.26

H2 predicts that the effect of the CPA merger on fees is mitigated by the existence of high entry barriers. We estimate the mitigating effect using the following empirical model:

$$\begin{equation}\ln AuditFee{s_{i,j,k,t}} = {\alpha _i}\ + {\delta _j} + {\pi _k} + {\lambda _t} + {\beta _1}Merge{r_{k,t}} + {\beta _2}Merge{r_{k,t}} \times Entry\,Barrie{r_{i,j,k,t}} + {\beta _3}\,Entry\,Barrie{r_{i,j,k,t}} + \gamma {{\rm X} _{i,j,k,t}} + {\varepsilon _{i,j,k,t}},\end{equation}$$(2)

where $$Entry\ Barrier$$ refers to one of two empirical proxies of the existing entry barriers: Big-4 auditors and auditee size. First, the submarket of Big-4 audit firms has a high entry barrier, as Big-4 audit firms have accumulated large reputation and human capital, established a global network and invested in fixed assets and technologies. Second, the submarket for large auditees also likely has a high barrier to enter because large auditees require their auditors to have high industry expertise and reputation as well as deeper pockets for potential litigation (Simunic, 1980), all of which impose a high barrier to enter this submarket. H2 posits that the effect of CPA merger is insignificant or less pronounced when the entry barriers are high; accordingly we expect $${\beta _1}$$ to be negative and $${\beta _2}$$ to be positive.

Finally, H3 examines whether the effect of the CPA merger on audit fees is more pronounced when the shock intensity of the CPA merger on entry threats is stronger. We estimate the effects of strong and weak shocks using the following models:

$$\begin{equation}\ln AuditFee{s_{i,j,k,t}} = {\alpha _i}\ + {\delta _j} + {\pi _k} + {\lambda _t} + {\beta _1}Merge{r_{k,t}} + {\beta _2}Merge{r_{k,t}} \times Strong\,Shoc{k_k} + \gamma {{\rm X} _{i,j,k,t}} + {\varepsilon _{i,j,k,t}},\end{equation}$$(3)

where $$Strong\ Shock$$ is an indicator variable equal to 1, if the number of CMA and CGA members on the governance board relative to the CA members (i.e., the (CMA + CGA)/ CA ratio) is greater than the median, and 0 otherwise. β₁ estimates the effect due to a weak shock, and the sum of β₁ and β₂ estimates the effect of a strong shock. We expect both $${\beta _1}$$ and $${\beta _2}$$ to be negative. We further validate our setting with placebo tests, using audit submarkets for Big-4 auditors and large clients, in which case the entry threats induced by the CPA merger are ex ante less credible to incumbents.

5 EMPIRICAL RESULTS

Table 4 presents the empirical results on the effect of the CPA merger on audit fees, corresponding to H1. Column (1) presents the base model; the coefficient on $$Merger$$ is negative (−0.0633) and significant at the 1% level (using two-tailed tests). In Column (2), we find that Merger continues to be significantly negative after including audit-firm, province and time fixed effects (−0.0825), and after further controlling for time-varying auditee characteristics (−0.0679) in Column (3). Finally, in Column (4), we further control for audit quality and continue to find that Merger is significantly negative (−0.0692), suggesting that the CPA merger results in an average reduction in fees of 6.92% after controlling for audit quality and other time-varying firm characteristics.27 In addition, the coefficients on Market-to-Book, Return on Assets (ROA), Total Assets, and Current Ratio are also significant. It is also worth noting that the adjusted R²s from the regressions are high due to the inclusion of the very extensive fixed effects. The high explanatory power indicates that omitted variables are unlikely to be a major concern.

Table 5 provides empirical results on whether the effect of the CPA merger is mitigated when the existing entry barriers are high (H2). We provide separate results for the cross-sectional analyses of Big-4 auditors and auditee size. Columns (1) and (2) show that the reduction in fees is driven by non-Big-4 auditors and muted for Big-4 auditors; in particular, the coefficients on $$Merger$$ are negative (−0.1334 and −0.1121, respectively) and significant at the 1% level, and the coefficients on $$Merger \times Entry\,Barrier$$ are positive (0.1763 and 0.0908, respectively) and significant at the 1% level, suggesting that fees decrease by 11.21% for non-Big-4 firms and do not change significantly for Big-4 auditors ($${\beta _1} + {\beta _2}$$ insignificant).

Similarly, in Columns (3) and (4), the coefficients on $$Merger$$ are negative (−0.1146 and −0.0905, respectively) and significant at the 1% level, and the coefficients on $$Merger \times Entry\,Barrier$$ are positive (0.1450 and 0.0436, respectively, although the latter is only marginally significant), indicating that the fee reduction, on average, is 9.05% for small clients and 4.69% for large clients.28 Collectively, these findings lend support to H2 and imply that the audit-fee reduction due to the CPA merger is mitigated in audit submarkets where the existing entry barriers are high due to higher reputation.

Next, Table 6 shows the empirical results on the ex-ante shock intensity (H3) for each audit submarket. In Columns (1) and (3), the coefficients on $$Merger \times Strong\,Shock$$ are both negative (−0.0889 and −0.0644, respectively), but significant only in Column (1). The result indicates that a stronger shock leads to lower audit fees for audit firms in the submarket of non-Big-4 auditors, where existing barriers are low and thus the entry threats to incumbent audit firms are ex-ante credible. Columns (1) and (3) show that β₁ is statistically insignificant, and the sum of β₁ and β₂ is statistically significant, using both non-Big-4 auditor and smaller auditee specifications. These findings suggest that the CPA merger leads to lower audit fees only in the submarkets of non-Big-4 auditors and smaller clients when the ex-ante merger shock intensity is stronger.

We further provide placebo tests, using audit submarkets where existing entry threats are high (i.e., submarkets for Big-4 auditors and large auditees) and hence the entry threats induced by the shock are ex ante less credible to incumbent firms. In Columns (2) and (4), the coefficients on $$Merger \times Strong\,Shock$$ are statistically insignificant, revealing that (1) CPA merger identifies credible entry threats in submarkets where existing entry barriers are low, and (2) audit firms that face credible entry threats reduce service prices. In terms of the economic significance, incumbent audit firms in submarkets for non-Big-4 auditors and small auditees, on average, reduce audit fees by 8.93% and 12.89%, respectively, when they face ex-ante stronger shocks on entry threats, whereas the fee reductions due to a stronger shock are statistically insignificant and lower in magnitudes for Big-4 auditors and large auditees (4.87% and 5.87% in Columns (2) and (4), respectively).

6 ADDITIONAL ANALYSES AND ROBUSTNESS TESTS

7 CONCLUSION

We study the “microeffect of a macromerger” of self-regulatory professional accounting bodies in Canada, which offers a unique setting to study the causal effect of an increase in the threat of entry on audit fees. We document that audit fees decrease in response to an exogenous increase in entry threats induced by the CPA merger, and that the fee reduction is more pronounced when the entry threat is expected to be greater and less pronounced when the entry threat is expected to be lower due to higher reputation. Our evidence suggests that the degree of the fee reduction is related to the relative bargaining power between the auditor and the auditee. In sum, the negative effect that the approval of the CPA merger has on audit fees suggests that audit firms are either more willing to give clients’ fee breaks proactively or are more likely to agree to clients’ fee concessions due to the increased entry threat. In either case, the fee reduction reflects audit firms’ increased incentive to deter future entry.

The paper also contributes to the literature on entry deterrence in the broader management and economics literature by providing empirical evidence on a professional service industry. The findings indicate that occupational licensure creates an entry barrier that protects the small players in the audit industry, and that relaxing licensing requirements induces these firms to strategically lower their prices to deter potential entrants. Occupational licensure has little impact on the big players in the audit industry due to their reputation for expertise. Equally importantly, the magnitude of fee reduction due to the CPA merger is affected by existing entry barriers, which leads to different consequences for Big-4 and non-Big-4 auditors, and small and large clients.

ACKNOWLEGEMENTS

We appreciate valuable comments from Daniel Aobdia, Muhammad Azim, Mahfuz Chy, Michael Marin, Michael Shen, Shibin Tang, Dushyant Vyas, Jingjing Wang, Ping Zhang, and seminar participants at the Financial Accounting and Reporting Section (FARS) Midyear Meeting, the Canadian Academic Accounting Association (CAAA) Annual Meeting, and the Rotman School of Management.