2012年4月17日学术报告通知.doc

Audit Market Concentration, Auditor’s Reputation as Globaland Country-level Market Leader, and Investor-perceived Audit Quality Simon Fung* Hong Kong Polytechnic University Simon.Fung@inet.polyu.edu.hk Ferdinand A Gul Monash University Sunway Campus Ferdinand.gul@buseco.monash.edu.my K. K. Raman University of North Texas raman@unt.edu Kevin Zhu Renmin University Kevin.zhuxindong@gmail.com *Authors’ names are listed alphabetically. Please do not quote or cite without permission. We gratefully acknowledge the helpful comments of Billy Brewster, Gordon Richardson, Soo Yong Kwon, Mujtaba Mian and other workshop participants at the University of Texas at Arlington, Nanyang Technological University and Journal of Contemporary Accounting and Economics 2012 Symposium. 1 Audit Market Concentration, Auditor’s Reputation as Global- and Country-level Market Leader, and Investor-perceived Audit Quality Abstract Motivated by recent calls to break up one or more of the Big audit firms, this study examines whether the largest auditor has the greatest economic incentive to provide a quality audit particularly in countries where it has the largest market share. Using observations for the period 1999-2004 when PwC was the world’s largest auditor and also the market leader (PwC-ML), we show that the favorable effects of Big 4 market dominance (i.e., higher aggregate Big 4 market share) on investorperceived audit quality to be stronger in PwC-ML countries and the adverse effects of within Big 4 concentration (i.e., more unequal market shares among the Big 4) on investor-perceived audit quality to be absent in PwC-ML countries. Similar results are obtained with respect to “factual” audit quality in PwC-ML countries. JEL Classification: L11, L15, M42, M48 Key words: Audit market concentration, Audit quality, Reputation as global- and country-level market leader 2 Audit Market Concentration, Auditor’s Reputation as Global- and Country-level Market Leader, and Investor-perceived Audit Quality 1. INTRODUCTION In this paper, we examine two research questions. First, around the world, does audit market concentration impact investors’ assessment of Big 4 audit quality? 1 Given Big 4 dominance of the audit market, regulators have expressed apprehension about the potential adverse implications of market concentration for investor confidence in the work of auditors (European Commission 2010; GAO 2003, 2008). To quote a former US Secretary of the Treasury: “… we have been left with only four major accounting firms…(which) may not be healthy. The big four dominate the industry in terms of revenues and professional staff. The remaining firms face significant barriers to competing with the big four … The current situation forces us to ask questions about the industry’s sustainability and effectiveness …?” (Paulson 2006). More directly, the GAO (2008) attempted to link auditor concentration to audit quality. “…(S)ome academics, a former regulatory official, and an industry consultant with whom we spoke, expressed concerns that concentration was affecting the quality of the audits. For example, one said that having only four firms in the market resulted in low-quality audits that hurt investors” (GAO 2008, p. 32). The basic concern of regulators is that auditor concentration reduces choice, induces complacency, impairs innovation, contributes to audit leniency, and erodes investor confidence (European Commission 2010; UK House of Lords 2011). For regulators – whose major concern is with maintaining investor trust in the integrity of reported corporate earnings and the orderly functioning of the capital markets – investor perception that audit market concentration impairs audit quality is potentially as serious an issue as direct evidence of factual impairment of audit quality (Francis 2011). Second, are the potential adverse effects of concentration on investor-perceived audit quality increased (or diminished) in countries where the world’s largest Big 4 firm also has the largest market share?2 A suggested remedy for audit market concentration is to break-up one or more of the Big 4 1 The world’s four largest accounting firms (Deloitte, Ernst & Young, KPMG, and PricewaterhouseCoopers) are collectively known as the Big 4. Although our study includes Arthur Andersen (until its dissolution until 2002), for convenience of exposition we refer to the Big 4. Our results are robust to deleting Andersen clients from our study. 2 Although the Big 4 firms consist of global networks of legally independent national member partnerships, they promote their brand names and advertise themselves worldwide as representing “one firm.” However, because audit markets are country-specific due to country-level regulation and licensing (Francis et al. 2012; Francis and Wang 2008), regulators and investors assess the size of a Big 4 firm at both global- and country-levels. We 3 firms (UK House of Lords 2011, p. 14, 62). Presumably, the largest Big 4 firm would be the primary candidate for any break-up mandated by regulators. However, any break-up of the largest auditor is likely to deprive the market of an auditor with the largest investment in reputational capital, i.e., an auditor with a greatest incentive to provide a quality audit. For this reason, whether the largest auditor has a beneficial (or detrimental) effect on audit quality in countries where it has the largest market share is an important research question. In our study, we investigate that question empirically. In our analysis, we focus on the 1999-2004 time period when PricewaterhouseCoopers (PwC) was the world’s largest auditor.3 As the world’s largest auditor, PwC was also the market leader in more countries than any other audit firm. Further, as noted by Sirois and Simunic (2010), the Big 4 firms (including PwC) pursue a strategy heavily reliant on audit technology investments intended to enhance audit quality (factual and/or perceived) in excess of the minimum required to meet professional and legal standards.4 As the world’s largest auditor, PwC potentially represents the firm with the largest costly investments (sunk cost) in reputational capital via strategic spending on personnel and robust training programs, audit technology and systems that are key inputs in the audit process, physical facilities, advertising, etc. to build the size and scale of the overall global organization. Consequently, PwC’s reputation as the largest auditor likely reflects the firm’s “superior” technology and talent. Particularly in auditing where service quality cannot be directly observed and the only characteristic of the audit known to users is the auditor’s identity, investors rely on the auditor’s reputational capital to assess audit quality (Simunic and Stein 1987). Given that the Big 4 firms (including PwC) promote their brand worldwide and advertise themselves as representing “one firm,” the auditor’s global-level reputational capital remains pertinent even in the context of audit markets focus our analysis of investor-perceived audit quality at the country-level since we expect overall investor perceptions to be related more to the auditor’s global- and country-level reputation for market leadership rather than the auditor’s industry-specific investments. 3 As discussed by Francis et al. (2012), there is a possible auditor coding problem in the Global Vantage database beginning 2005. Hence, we restrict our sample period to 1999-2004. Separately, Deloitte attained similar global-level size as PwC (about $27 billion in revenues) in 2010. However, PwC was the largest audit firm during the period of our study (1999-2004). 4 Consistent with this argument, Crespi and Marette (2009) indicate that in most industries high fixed (sunk) costs at the firm-level are positively related to output quality. 4 that are (as noted previously) basically country-specific because of national licensing and regulation of auditors. Separately, as noted by Coffee (2006): “.. if each of the Big Four audits roughly 25 percent of all large public companies, it seems predictable that each will experience one or more clients who will engage in fraudulent financial reporting. But if each suffers a similar level of reputational damage, none is worse off vis-à-vis its competitors. Hence the auditor may conclude that the occasional scandal is tolerable …Ultimately, reputational injury is a relative concept” (p. 159). Put differently, an equal market share implies an equal prospect of reputational injury and an equal incentive for the Big 4 to provide a quality audit. By contrast, as the world’s largest auditor, PwC is likely to incur the largest reputational injury by providing a negligent audit, i.e., suffer a greater loss of revenues/clients to its competitors particularly in countries where it is the market leader. Consequently, PwC is likely to have a greater economic incentive to provide a quality audit especially in countries where it has the largest market share. Potentially, PwC’s incentive to impose stricter quality controls and tighter coordination of national affiliates is likely stronger in countries where its concern for damage to its global reputation as the world’s largest auditor is heightened, i.e., in countries where it also is the market leader. To infer investors’ assessment of audit quality, we utilize two measures, namely (1) the informativeness of clients’ reported earnings, and (2) the cost of capital. The FASB’s conceptual framework defines the ‘‘usefulness’’ of reported earnings to investors in terms of relevance and reliability (SFACs No. 1 [FASB 1978] and No. 2 [FASB 1980]). One measure of investors’ perception of earnings quality is the earnings response coefficient from regressions of returns on earnings (Ghosh and Moon 2005; Schipper and Vincent 2003; Warfield et al. 1995; Teoh and Wong 1993). Another stream of research (e.g. Khurana and Raman 2004; Mansi et al. 2004; Pittman and Fortin 2004), argues and provides evidence that audit quality matters to investors and is associated with the cost of capital, which is a measure of financial reporting credibility. Other things being equal, the higher the investor-perceived quality of the audit, the more credible the reported earnings and the lower the cost of equity and debt financing for companies (DeAngelo 1981). 5 In a recent study, Francis et al. (2012) use a cross-country research design to examine the relation between audit market concentration and “factual” audit quality.5 Specifically, they analyze the statistical properties of Big 4 clients’ reported earnings with respect to accruals, the likelihood of reporting a loss, and timely loss recognition. In their analysis, they distinguish between two separate dimensions of audit market concentration: (1) Big 4 market dominance (i.e., aggregate Big 4 market share in a country), and (2) concentration within the Big 4 (i.e., uneven market shares among the Big 4 in a country). They argue that Big 4 market dominance in a country is evidence of a strong underlying demand for high-quality audits in that country. Hence, in countries with a lower Big 4 market share, the demand for high-quality audits is lower and the Big 4 must compete on price with other (non-Big 4) auditors resulting in lower audit quality for Big 4 clients. The empirical implication of this argument is that factual audit quality for Big 4 clients will be higher in countries where auditor concentration – as measured by the aggregate Big 4 market share is higher. Francis et al.’s (2012) second metric relates to the degree of concentration within the Big 4 in a country (as measured by the Herfindahl index based on Big 4 market shares in that country). They argue that to the extent that the Big 4 firms have unequal market shares within a country, each firm will have dissimilar scale and will not be able to contest the market on an equal basis with the other three firms. Consequently, factual audit quality for Big 4 clients will be lower in countries where auditor concentration within the Big 4 is higher. However, as noted previously, equal market shares imply an equal prospect for reputational injury and thus an equal incentive for audit quality across all the Big 4 firms (Coffee 2006). By contrast, if market shares are unequal, the auditor with the largest market share is likely to have a greater reputation-based incentive to provide a quality audit. Moreover, in a market with unequal shares, the auditor with the largest market share (particularly PwC, with a global-level reputation as the world’s largest auditor) may pressure the smaller Big 4 firms to maintain service quality so as to 5 Although audit quality cannot be directly observed, it can be inferred from financial statement measures. We use the term “factual” audit quality (consistent with Francis and Ke 2006) to refer to audit quality inferred from such measures. Thus, accruals of smaller magnitude or a higher likelihood of reporting a loss are viewed as indicating higher earnings quality. Since audited earnings are the outcome of negotiations between the client and the auditor, higher earnings quality is viewed as an indicator of higher audit quality. By contrast, the investor-perceived audit quality we examine in our study refers to audit quality inferred from market-based measures such as earnings informativeness or the cost of equity and debt financing. 6 avoid losing additional market share to the market leader. In other words, given PwC’s reputation as the world’s largest auditor, its market leadership at the country-level could be perceived favorably by investors, i.e., the potential adverse effects of unequal Big 4 market shares (i.e., within Big 4 concentration) may be lower in countries where the market leader (PwC) is also the world’s largest auditor. Ultimately, the relation between audit quality for Big 4 clients and auditor concentration in countries where PwC is the market leader remains an empirical question. Our cross-country analysis is in two parts: First, we examine whether the two auditor concentration metrics – Big 4 market dominance and concentration within the Big 4 – affect investorperceived audit quality. As noted previously, the higher the perceived quality of the audit, the greater the credibility of reported earnings, and (other things being equal) the greater the earnings informativeness and the lower the cost of equity and debt financing for the client (DeAngelo 1981; Teoh and Wong 1993). Hence, we infer investor-perceived audit quality from the informativeness of clients’ reported earnings and their cost of equity and debt financing. As noted previously, because audit quality is not directly observable, to regulators and policy makers these investor perceptions are as important as evidence of factual audit quality (Levitt 1998). Moreover, prior research suggests that investor-perceived audit quality can significantly differ from factual audit quality (Francis and Ke 2006).6 Hence, it is important to address the relation between auditor concentration and perceived audit quality, and we do so in our study. Second, we examine whether the relation between the two auditor concentration metrics and investor-perceived audit quality is conditioned by whether PwC has the largest share of the audit market in that country. Given PwC’s reputation as the world’s largest auditor, its global as well country-level market leadership may be expected to have a positive corollary, i.e., the perceived benefits of Big 4 market dominance may be greater and the adverse effects (if any) of within Big 4 concentration may be lower in countries where PwC has the largest market share. 6 As an example, although previous research (e.g., Ashbaugh et al. 2003; Chung and Kallapur 2003) indicated that nonaudit services provided by the incumbent auditor did not impair factual audit quality, these services were perceived by investors as compromising audit quality (Francis and Ke 2006). Largely as a result of these perceptions (and despite the lack of evidence of impaired factual audit quality), the SEC mandated the disclosure of nonaudit fees and the Sarbanes Oxley Act prohibited the incumbent auditor from supplying many types of nonaudit services. These developments underline the overriding importance of investor perception (i.e., investor confidence) for regulators. 7 Our sample consists of 11,823 to 31,811 observations (depending on the analysis) drawn from 29 countries over the 1999-2004 time period.7 Given the cross-country nature of the study, our analysis includes control variables that capture the country’s financial development, legal environment and enforcement of investor protection rights. Our analysis also includes controls for client-specific characteristics. Collectively, our cross-country findings suggest that Big 4 market dominance is associated with higher earnings informativeness and a lower cost of equity and debt financing for clients. Other things being equal, since greater earnings informativeness and a lower cost of equity and debt imply more credible financial reporting (and, by implication, higher investor-perceived audit quality), our findings suggest that investors view a larger market share for the Big 4 as improving audit quality for Big 4 clients in a country. Also, our findings suggest that within Big 4 auditor concentration (as measured by the Herfindahl index based on Big 4 market shares) is associated with lower earnings informativeness and a higher cost of equity and debt financing. Once again, holding other things constant, since lower earnings informativeness and a higher cost of equity and debt financing imply less credible financial reporting (and, by implication, lower investor-perceived audit quality), our finding suggests that investors view unequal market shares among the Big 4 (i.e., market dominance within the Big 4) as lowering audit quality for Big 4 clients in that country. To assess the impact of PwC’s country-level market leadership on these relations, we partition our sample by PwC-ML countries (i.e., countries in which PwC has the highest market share) and other countries. In this analysis, we find the positive effects of Big 4 market dominance, i.e., higher earnings informativeness and a lower cost of equity and debt financing to be stronger in PwC-ML countries than in other (non-PwC ML) countries. Also, we find the negative effects of within Big 4 auditor concentration, i.e., lower earnings informativeness and higher cost of equity and debt financing to be absent in PwC-ML countries and to hold only in non-PwC ML countries. Thus, we 7 As noted previously, although Francis et al. (2012) discuss a possible auditor coding problem in the Global Vantage database beginning 2005, they assume the 2004 auditor for a client is the same in 2005 through 2007 “to retain a longer sample period.” Rather than make such an assumption (which may or may not be valid), we restrict our sample period to 1999-2004. Separately, our sample is limited to 29 countries (versus 42 countries for Francis et al. 2012) due to data availability constraints linked to our investor-perceived audit quality metrics. 8 find that PwC enhances the positive (and mitigates the negative) effects of audit market concentration on perceived audit quality in countries where it is the market leader.8 In additional analysis, we also examine factual audit quality in PwC-ML countries (i.e., countries where PwC is the market leader) and other countries. Consistent with our findings for perceived audit quality, we find (1) the positive effects of Big 4 market dominance to be stronger, and (2) the negative effects of within-Big 4 auditor concentration to be absent for factual audit quality in PwC-ML countries. Thus, our findings are similar for perceived as well as factual audit quality in PwC-ML versus other countries. When we partition our sample by countries with strong investor protection and other countries, we find that strong investor protection has no impact on the positive effects of Big 4 market dominance or the negative effects of within Big 4 concentration for either perceived or actual audit quality. Overall, our results point to PwC’s reputation as the world’s largest auditor having a beneficial effect on audit quality in countries where it is also the market leader. Our paper contributes to the literature in several ways. First, our cross-country study adds to the limited prior research on the cross-country effects of supplier concentration in the audit market. Specifically, we examine the cross-country relations between audit market concentration and audit quality from an investor perspective. Our findings suggest that investor-perceived audit quality is positively related to Big 4 market dominance and negatively related to within Big 4 auditor concentration. These results address potential regulatory concerns over the possible adverse effect of Big 4 auditor concentration for investors’ assessment of audit quality. In particular, our findings suggest that investors are concerned about concentration within the Big 4 rather than overall Big 4 dominance of the audit market. Second, we find the positive effects of Big 4 market dominance to be stronger and the negative effects of within Big 4 concentration to be absent in countries where PwC is the market leader. In other words, we find that PwC has a favorable impact on investor-perceived (as well as factual) Big 4 audit quality in countries where it has the highest market share. Put differently, while 8 By contrast, when we partition our sample by countries in which each of the other three Big 4 firms (i.e., Deloitte, Ernst & Young, or KPMG) is the market leader auditor, we find no difference in the relation between auditor concentration and investor-perceived audit quality across the partitions. 9 unequal market shares among the Big 4 have a negative effect on investor perceptions (and factual audit quality) in other countries, we find no such effect in countries where PwC has the largest market share. Thus, our findings suggest that PwC’s reputation as the world’s largest auditor has a positive corollary effect on audit quality – both perceived and factual – in countries where it is also the largest auditor. Third, prior research in the economics literature (e.g., Dick 2007; Ellickson 2007) suggests that the dominant firms in banking and supermarkets incur large firm-level fixed (sunk) costs in technology (R&D) and advertising. To the extent that service quality in these industries is related to the fixed investments (sunk costs) rather than variable costs, local markets dominated by these large firms are associated with higher service quality. Thus, Crespi and Marette (2009) empirically demonstrate that overall market concentration and high fixed (sunk) costs are positively related to quality in most industries. Other things being equal, the largest firm in any industry is also likely to have the highest fixed costs as it builds scale and breadth.9 However, to our knowledge, the prior industrial organization literature has not examined whether country-level markets dominated by the world’s single largest firm in the industry are also associated with higher quality. In our study, we examine this question for the audit industry at the country-level.10 Our findings suggest that PwC has a favorable effect on the relation between auditor concentration and investor-perceived (and factual) audit quality in countries where it is the market leader. Thus, we find no evidence to suggest that a break-up of PwC (the world’s largest Big 4 firm) – as a remedy for auditor concentration -- would improve either perceived or factual audit quality. Rather, the evidence suggests that a break-up of the largest Big 4 firm could adversely affect audit quality in countries where it is the market leader. The rest of the paper is organized as follows: in section 2 we provide background information relating to audit market concentration and investor-perceived audit quality and develop our empirical predictions. Section 3 presents our research method and data, while results and findings from additional analyses are reported in section 4. Section 5 provides concluding remarks. 9 Thus, Deloitte invested more than $1 billion in scale and breadth prior to catching up with PwC in global-level size in 2010 (inaudit.com/audit/...audit/deloitte-is-the-world’s-top-audit-firm-1287/). 10 As noted by Francis et al. (2012), due to local licensing requirements and restrictions on the cross-border flow of labor, each country constitutes a unique practice and localized audit market. 10 2. HYPOTHESES DEVELOPMENT As noted earlier, the thrust of the argument regarding concentration is that it potentially limits choice, induces complacency in the incumbent auditor, lowers auditor skepticism, reduces the rigor of audit procedures, result in a more lenient audit, and impairs investor confidence in the integrity of reported earnings (European Commission 2010; GAO 2003, 2008; UK House of Lords 2011; US Treasury 2008). Still, the prior literature on industrial organization (I/O) theory is inconclusive on the effects of concentration on output quality (Demsetz 1973; Robinson and Chiang (1996); Schmalensee 1978; Shaked and Sutton 1987; Spence 1975; Sutton 1991). Thus, while Cohen and Mazzeo (2004), Domberger and Sherr (1989), Dranove and White (1994), and Mazzeo (2003) suggest that concentration is negatively correlated with service quality in several industries such as hospital services, legal services, airlines, media, and banking, Crespi and Marette (2009) and George (2007) suggest otherwise, i.e., that greater concentration is associated with higher service quality. Thus, based on current research evidence from markets other than auditing, there is no a priori reason to believe that concentration impairs audit quality. 2.1 Big 4 Market Share and Investor-Perceived Audit Quality Basically, the auditor’s role is to add credibility to the client’s reported earnings by restraining the exercise of managerial discretion in financial reporting and thereby limiting the risk of financial statement misstatements. However, audit quality is not directly observable since the output is normally a standardized report, i.e., an unqualified (clean) audit opinion. Moreover, only the auditor can determine the amount of effort needed to plan and conduct the audit (Causholli et al. 2010). Further, the consumer of the audited financial statements is the investor rather than the client who selects the auditor and pays for the audit. Thus, the supplier (auditor) potentially knows more about the quality of the audit than either the purchaser (the client) or the consumer (the investor). Consequently, the investor relies on the auditor’s reputational capital to assess audit quality. Along the same lines, DeAngelo (1981) suggests that audit quality is related to the size of the auditor. Specifically, she argues that incumbent auditors earn client-specific quasi rents (due to the technological advantages of incumbency) and that these quasi-rents incentivize the auditor to maintain audit quality, i.e., serve as collateral against opportunistic behavior on the part of the auditor. Further, 11 although quasi-rents are not observable, in the aggregate they are expected to vary with the size of the auditor as proxied by aggregate revenues and/or aggregate client sales. Consequently, although auditor size can never be a perfect surrogate for audit quality, a large auditor may be viewed by investors as having a greater incentive to tell the truth and thus rationally perceived as providing a higher quality audit. Similarly, other research (e.g., Simunic and Stein 1987) suggests that investors impute audit quality based on the auditor’s brand name/reputation. Specifically, the Big 4 incur large fixed (sunk) costs (in human capital including professional staffing/training, technical resources, advertising, etc.) to build the necessary scale and breadth to provide a quality-differentiated audit. Moreover, fixed (sunk) costs in advertising can in its own right be a component of perceived quality (Becker and Murphy 1993). Consistent with these arguments, prior research (e.g., Becker et al. 1998; Francis et al. 1999; Teoh and Wong 1993) suggests that Big 4 auditors provide higher audit quality – both factual and perceived -- relative to other (non-Big 4) auditors. In a cross-country context, Francis et al. (2012) suggest that the market share for the Big 4 in a country is an observable proxy for the demand for high quality audits in that country. Consequently, the higher the Big 4 market share, the greater the demand for high quality audits, and the lower the need for the Big 4 to compete with other non-Big 4 auditors based on price. Hence, they predict (and find) that the factual quality of Big 4 audits is higher in countries with a larger Big 4 market share. As noted previously, this argument is consistent with economic theory which suggests that to the extent that service quality is related to fixed (sunk) costs, local markets dominated by the large firms are associated with higher service quality (Sutton 1991; Dick 2007; Ellickson 2007). In our study, to the extent that investor perceptions are consistent with the notion that the large audit firms (with their large fixed/sunk costs in reputational capital, i.e., professional staffing/training, R&D, technical resources, advertising, etc.) provide higher quality audits, the relation between the Big 4 market share in a country and perceived audit quality may be expected to be positive. Alternatively, to the extent that investors share the concerns expressed by audit regulators that market dominance by the Big 4 restricts choice and induces auditor complacency (as discussed 12 previously), the relation between Big 4 market share in a country and perceived audit quality could be negative. In our cross-country study, our first hypothesis (stated in the null form) is as follows: H1: There is no relation between the Big 4 market share in a country and investor-perceived audit quality. 2.2 Unequal Big 4 Market Shares and Investor-Perceived Audit Quality Francis et al.’s (2012) second argument is that concentration within the Big 4 (as measured by the Herfindahl index) impairs choice and reduces the incentive for Big 4 auditors to provide highquality audits. They report that higher concentration within the Big 4 is associated with lower factual audit quality. With respect to our study, to the extent that investors share the concerns expressed by audit regulators that audit market concentration limits competition, restricts choice and induces auditor complacency (as discussed previously), the relation between concentration (as measured by the Herfindahl index) in a country and perceived audit quality could be negative. However, it is not clear whether higher concentration necessarily implies less competition. Specifically, recent research (Dedman and Lennox 2009) suggests that the link between the Herfindahl index and perceived competition is not straightforward. Moreover, Baumol et al. (1982) and Stiglitz (1987) suggest that even a market with two suppliers can be intensely competitive as long as the threat of entry by new rivals exists, i.e., competition can be intense even in highly concentrated markets with unequal market shares. Thus, our second hypothesis (stated in the null form) is as follows: H2: There is no relation between within Big 4 audit market concentration in a country and investor perceived audit quality. 2.3 PwC’s Country-Level Market Leadership and Investor-Perceived Audit Quality As noted previously, Dick (2007) and Ellickson (2007) suggest that service quality in certain industries (such as banking and supermarkets) is related to fixed investments (sunk costs) rather than variable costs. Further, Crespi and Marette (2009) empirically demonstrate that overall market concentration and high fixed (sunk) costs at the firm-level are positively related to quality in local markets in most industries. However, the prior economics literature (to our knowledge) has not examined whether market leadership at the country-level by the largest firm in the industry is associated with higher service quality. To the extent that the largest firm in the industry also has the 13 highest investment (fixed/sunk costs) in attributes pertinent to output quality, service quality could be higher in countries where it is also the market leader. With respect to auditing, PwC (as the world’s largest auditor) may be expected to have the highest investment in reputational capital, i.e., fixed/sunk costs in professional staffing and robust training programs, R&D, technical resources including proprietary audit methodologies, and advertising. Specifically, large expenditures (sunk costs) in advertising can be critical to service quality as a means of segmenting the market by quality signals, as a barrier to entry to generate the economic profits necessary to incur the large sunk costs in professional staffing and technical resources, and as a necessary component of perceived quality in its own right (Becker and Murphy 1993; Crespi and Marette 2009; Schmalensee 1978; Sutton 1991). For these reasons, given PwC’s reputation as the world’s largest auditor, its market leadership at the country-level could be perceived favorably by investors. Our last two hypotheses, also stated in the null form, are as follows: H3: There is no difference between PwC-ML countries (i.e., countries where PwC is the market leader) and other countries in the relation between Big 4 market share and investor-perceived audit quality. H4: There is no difference between PwC-ML countries (i.e., countries where PwC is the market leader) and other countries in the relation between within Big 4 audit market concentration and investor-perceived audit quality. 3. METHODOLOGY AND DATA 3.1 Perceived Audit Quality, Big 4 Market Share and Within Big 4 Market Concentration We examine investor-perceived audit quality in terms of the informativeness of reported earnings and investors’ required rate of return (cost of capital) for audit clients. Specifically, we examine whether the Big 4 market share and the level of concentration within the Big 4 (unequal market shares among the Big 4) in a country is associated with perceived audit quality for Big 4 clients. We also examine whether these associations are different in PwC-ML countries (i.e., countries where PwC is the market leader) versus other countries. We employ two measures of audit market concentration, namely Big 4 market share (Big4Share) and market concentration within the Big 4 (Big4Concen). In particular, Big4Share is measured as the percentage of listed companies audited by the Big 4 in a country for each test year. Also, market concentration within the Big 4 (Big4Concen) is measured as the Herfindahl index based 14 on Big 4 market shares. If all Big 4 firms have equal market shares, the Herfindahl index will have a value of 0.25 (for Big 4) or 0.20 (for Big 5), and if one firm has the entire Big 4 market share the index will have a value of 1.00.11 Note that Francis et al. (2012) compute these measures on a country-industry-year basis in examining the association between audit market concentration and factual audit quality measures. However, our focus is on investors’ perception of an individual auditor’s global- and country-level reputation for quality, which is not necessarily relevant to the firm’s industry-related investments. Moreover, the number of sample observations available for computing the concentration metrics on a country-industry-year basis is potentially too small and the tests may not be sufficiently powerful.12 As such, we compute the concentration measures on a country-year basis rather than a country-industry-year basis. In additional tests, we assess the results based on the country-industry-year measures as in Francis et al. (2012). As discussed in Francis et al. (2012), both observable and unobservable country-level characteristics are likely to be correlated with both the overall market share of the Big 4 and the level of concentration within the Big 4. Hence, in our analysis, we control for a set of country-level characteristics believed to influence the effects of audit market structure on audit quality: the country’s financial development, national income per capita, rule of law and proxies for a country’s legal environment and legal enforcement of investor rights. Financial market development and investor protection have been shown in prior research to be associated with audit as well as financial reporting quality (e.g., Ball et al. 2000; Hung, 2000; Leuz et al. 2003; Francis et al. 2003; Choi and Wong 2007; Kwon et al. 2007; Francis and Wang 2008). Thus, controlling for these country-level characteristics is important in determining whether audit market concentration affects perceived audit quality. These country-level control variables are discussed further in the next section. 11 As an additional test, we also compute Big4Share and Big4Concen after excluding Arthur Andersen, i.e., using only the extant Big 4 firms for our entire sample period. Results are qualitatively similar. Results are also robust when we use clients’ total assets as the basis to measure Big4Share and Big4Concen. 12 The average number of observations is 158 per country-year when the concentration metrics are measured on a country-year basis but drops dramatically to 12 per country-industry-year when the metrics are measured on a country-industry-year basis using two-digit SIC codes as in Francis et al. (2012). In alternative country-industryyear analysis using GIND industry classifications in Worldscope, we obtain similar inferences as those reported in the paper using country-year analysis. Francis et al. (2012) also report similar inferences for their factual audit quality analysis regardless of whether concentration is measured on a country-year or a country-industryyear basis. 15 Data on the audit market concentration test variables and client-level control variables are obtained from the Compustat Global Vantage Database for the years 1999 through 2004. We do not include years after 2004 due to the existence of an auditor coding problem in the Global Vantage database beginning in 2005, as noted by Francis et al. (2012). While the sample firms from Global Vantage tend to be the larger listed clients, it is not necessarily a problem in the context of our study since the primary concern with respect to audit market concentration is the limited choice faced by larger listed clients (GAO 2003; Oxera 2006; U.S. Treasury 2008). 3.2 Empirical Models We test our hypotheses using the following empirical models: 3.2.1 Earnings Informativeness We examine the auditor reputation effects through a test of earnings informativeness. In particular, if the auditor’s reputation leads investors to perceive higher audit quality, the reported earnings is likely to be viewed as more informative. In other words, the higher the perceived audit quality, the stronger the association between reported earnings and stock returns. Consistent with prior studies (e.g. Teoh and Wong 1993; Warfield et al. 1995; Firth et al. 2007), we estimate the following regression model: RETURN = β0 + β1EARN + β2TestVar + β3EARN*TestVar + α k ClientControls + λkCountryControls + b k EARN*ClientControls +ρk EARN*CountryControls + FixedEffects + error (1) In model (1), the variables of interest are the interaction terms between EARN and our test variables (TestVar) Big4Share and Big4Concen. The effects of the test variables on earnings informativeness is tested by the coefficient (β3) on EARN*Big4Share and on EARN*Big4Concen. The model includes a number of client-specific controls, namely client size (SIZE), leverage (LEV), profitability (ROA), growth (GROWTH), systematic risk (BETA) and total accruals (TOT_ACC). In addition to client-specific controls, we also control for a number of country-level characteristics that might affect cross-country variations in returns. These include the gross national income per capita (LNGNI), financial market development (FIN_DEVEL), a country’s level of investor protection (INV_PROT) and rule of law (RULE_OF_LAW). Following Francis et al. (2012), 16 the investor protection control variable is a single parsimonious measure constructed based on a factor analysis using five investor protection variables (from La Porta et al. 1998, 2006) and the size of a country’s securities regulatory staff (from Jackson and Roe 2009). The variable RULE_OF_LAW measures “the perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence” (Kaufmann et al. 2008). All these variables are defined in the APPENDIX. Finally, industry and year fixed effects are included in the model, and the estimations are clustered at the client level in the regressions. 3.2.2 Cost of Capital Investor perception of audit quality is also expected to be demonstrated in the client’s cost of capital, and we test our hypotheses by examining the association between Big4Share or Big4Concen and the client’s cost of equity and debt capital. In particular, if the client is perceived to have received a higher quality audit, equity and debt investors are likely to require a lower rate of return. Based on a number of prior studies (e.g. Francis et al. 2005a; 2005b), we estimate the following regression model: COC = β0 + β1TestVar + α k ClientControls +λk CountryControls +FixedEffects + error (2) In model (2), COC denotes the measures of cost of equity and debt capital (IndEP and CostDebt, respectively) and the variables of interest (TestVar) are Big4Share and Big4Concen. Following Francis et al. (2005b), we measure the cost of equity capital as the industry-adjusted E/P ratio (IndEP), using the price earnings multiple as a short-hand valuation metric (e.g. Liu et al. 2002; Dechow and Dichev 2002). We use the earnings-price ratio to address concerns related to small values of earnings in the denominator, and we industry-adjust the ratio based on Alford’s (1992) finding that industry membership works well for selecting firms that are comparable in terms of risk and growth. The cost of debt, on the other hand, is measured as the ratio of interest expense in year t to average interest-bearing debt outstanding during year t (CostDebt). Similar to the earnings informativeness test, both client-specific and country-level control variables are included in the model. In terms of client-specific controls, we include size (SIZE), leverage (LEV), profitability (ROA) and total accruals 17 (TOT_ACC) for both the IndEP and CostDebt models (Kaplan and Urwitz 1979; Palepu et al. 2000; Francis et al. 2005b; Bae and Goyal 2009). In addition we control for growth (GROWTH) and systematic risk (BETA), which attempt to capture the variations in IndEP that are potentially unrelated to audit market concentration. Specific to the CostDebt model, we also include the client firm’s tangible assets (TANGIBLE) as an additional client-specific control since these assets reduce the cost of financial distress by making it easier for lenders to monitor borrowers (Bae and Goyal 2009). Country-level controls are included in a similar fashion as in the earnings informativeness model, and include the gross national income per capita (LNGNI), financial market development (FIN_DEVEL), a country’s rule of law (RULE_OF_LAW) and a country’s level of investor protection (INV_PROT). For the CostDebt model, we also control for the country’s lending interest rate (LRATE), following prior literature (e.g. Francis et al. 2005a). Industry and year fixed effects are also included in the models, with standard errors being clustered at the client level in the regressions. Again all the variables are defined in the APPENDIX. 3.3 Countries Where PwC is the Market Leader versus Other Countries In order to examine the effects of PwC’s country-level market leadership on the association between Big4Share (Big4Concen) and investor-perceived audit quality (Hypotheses 3 and 4), we partition our sample by countries where PwC is the market leader (i.e., has the largest market share) and other countries. We then estimate our equations (1) and (2) for each of the two set of countries separately. We then conduct an F-test on the difference in the coefficients for the variables of interest between the two sub-samples (i.e., observations in countries where PwC is the market leader and other countries). While an alternative specification of this analysis is to include an interaction variable between our variable of interest (Big4Share or Big4Concen) and an indicator variable (PwCML) for countries where PwC is the market leader, our inferences are drawn based on the sub-sample analyses for two reasons: (1) the number of observations in each of the two sub-samples is sufficiently large for separate estimation, and (2) tests based on the full sample with an interaction variable may 18 be less precise under certain circumstances when coefficients on the control variables differ between the two groups (e.g. Hardy 1993).13 3.4 Sample Selection and Descriptive Statistics Table 1 reports the sample selection process. We begin with all client-year observations available in Compustat Global Vantage with positive total assets during 1999–2004, which yields 75,792 firm-year observations with 15,061 unique client firms from 29 countries. Following Francis et al. (2012), we exclude 19,337 firm-year observations with a non-Big 4 auditor, since we are studying the effects of market concentration on Big 4 audits.14 We also exclude 4,560 observations with missing country-level control variables, and 3,496 financial and utility firms. With this primary sample (N = 48,399), we exclude observations with missing client-level data for the respective tests, resulting in the earnings-informativeness sample (N = 16,617), the cost of equity sample (N = 11,823) and the cost of debt sample (N = 31,811). Note (from Panel A of Table 3) that a large portion of client-year observations are from the US (N = 17,392 in the primary sample). As reported later in section 4, the results are robust to excluding from our analysis the US as well as other countries (one at a time) with more than 1,000 observations each (Australia, Canada, France, Germany, Italy, Malaysia, Singapore, Sweden, United Kingdom). [Insert Table 1 Here] Table 2 reports the number of countries in which an individual Big 4 firm is the market leader (i.e., has the largest market share), both using all countries available in Compustat Global Vantage (Panel A) and using the 29 sample countries included in our study (Panel B). Results show that for all of our test years, PwC is the market leader in more countries than any other auditor. [Insert Table 2 Here] Panel A of Table 3 presents descriptive statistics for country-level variables. As for our test variables, Big4Share (the Big 4 market share in a country for a particular year) has a mean value of 51 percent and ranges from a low of 18 percent in Australia to a high of 75 percent in Chile. Big4Concen (the within Big 4 concentration measure, as measured by the Herfindahl index) has a mean value of 0.35, and there is considerable variation across the 29 countries, with the lowest and 13 See Gul et al. (2009) for more discussion. 14 We use all observations to calculate the Big 4 market share in a country. 19 highest values of 0.24 and 0.63 for the US and Philippines, respectively. The inter-quartile ranges for Big4Share and Big4Concen are 0.36 to 0.64 and 0.28 to 0.39 respectively.15 Other country-level variables also vary widely. Financial market development (FIN_DEVEL) has a mean of 0.97 and ranges from 0.19 in Austria to 3.45 in Hong Kong. RULE_OF_LAW has a mean of 1.18 and ranges from -0.49 in Argentina to 1.96 in Switzerland. Finally, INV_PROT is calculated using the single significant factor that loads from a principal components analysis of COMMON, ANTIDIR, DISC, LIAB, PUBENF and SEC_STAFF. By construction, it has a mean value of zero and ranges from a low of -2.23 in Germany to 0.91 in Singapore. The correlations among country-level variables are reported in Panel B of Table 3. Big4Concen is highly correlated with INV_PROT (-0.605), suggesting that controlling for investor protection is important in our estimation. In addition, the correlation coefficient between PwC-ML and INV_PROT is 0.569, suggesting that PwC market leadership is capturing more than just the effects of investor protection. Also, as one would expect, the individual investor protection characteristics (COMMON, ANTIDIR, DISC, LIAB, PUBENF and SEC_STAFF) are highly correlated, justifying the principal components analysis employed to obtain the measure of INV_PROT. Finally, relatively high correlations are found among the following variables: FIN_DEVEL and INV_PROT (0.516), RULE_OF_LAW and LRATE (-0.467) and RULE_OF_LAW and LNGNI (0.838). In untabulated sensitivity analyses, we exclude each of these variables (one at a time) from the model and re-estimate our analyses. We obtain qualitatively similar results, suggesting that collinearity is not likely to be a serious issue in our specification. [Insert Table 3 Here] Panel A (B) Table 4 reports descriptive statistics (correlations) of the client-level control variables used in the study. The mean (median) value of RETURN is 0.133 (0.018), and the mean (median) value of EARN is 0.013 (0.031). As for the cost of capital measures, while the mean of IndEP (cost of equity measure) is 0.047, that of CostDebt (cost of debt measure) is higher at 0.111. The two metrics are not directly comparable because while the cost of debt is measured by the client’s 15 In an untabulated sensitivity test, we delete the top and bottom five percent of client-year observations with extreme values of Big4Share and Big4Concen. Results are qualitatively similar, suggesting that our results are not driven by these extreme values. 20 interest expense scaled by the average interest-bearing debt, the cost of equity is measured by the industry-adjusted earnings-price ratio, both of which follow Francis et al. (2005b). Moreover, the measures are computed based on separate sets of observations. As additional tests, we repeat our analyses using alternative cost of equity measures, and results are qualitatively similar (discussed later in the paper). No sign of serious collinearity problems is observed from the correlation results reported in Panel B.16 [Insert Table 4 Here] 4. RESULTS 4.1 Hypothesis 1: Big 4 Market Share and Perceived Audit Quality Table 5 presents the results of the tests of Hypothesis 1. All models are significant at p < 0.01, and the control variables are generally significant in a direction consistent with prior research. Panel A reports the results on the effects of Big 4 market share on earnings informativeness measured by the earnings-returns association. We find that the coefficient for EARN*Big4Share is significantly positive at 0.530 (p = 0.040), suggesting that the relation between returns and earnings is stronger (i.e., investors find reported earnings to be more informative) in countries where the Big 4 have a larger market share. Put differently, perceived audit quality is higher in these countries, i.e., reported earnings are viewed by investors as more credible and exhibit a stronger earnings-returns association. The results for the control variables in Panel A are consistent with those in the prior literature. Panels B and C report results for the association between Big 4 market share and our two cost of capital measures. After controlling for client-level and country-level factors as well as industry and year fixed effects, we find that the coefficients for Big4Share are -0.102 (p < 0.01) and -0.025 (p = 0.066) in the IndEP and CostDebt models, respectively. These findings suggest that both equity and debt investors view audit quality for Big 4 clients to be higher in countries where the Big 4 have a larger market share, and require a lower risk premium (and a lower required rate of return) in 16 The correlation between EARN and ROA is 0.566, which is the highest correlation observed among client- level independent variables. In a sensitivity test, we exclude ROA from our estimation and results are qualitatively similar to our main findings (not tabulated for brevity). 21 providing finance to these clients. Once again, the results for the control variables in Panels B and C are generally significant and consistent with prior research. [Insert Table 5 Here] 4.2 Hypothesis 2: Within Big 4 Market Concentration and Perceived Audit Quality We test Hypothesis 2 by examining the effects of within Big 4 market concentration on earnings informativeness and the cost of capital. Table 6 presents the results of our empirical estimations. In Panel A, while the coefficient for EARN is positive and significant (2.255, p = 0.035), the interaction term EARN*Big4Concen is negative and significant (-1.076, p = 0.074), suggesting that higher within Big 4 market concentration is associated with a weaker earnings-returns association i.e., less informative earnings. These results echo Francis et al. (2012) and regulatory concerns that lower contestability of market share (as measured by a higher Herfindahl index) adversely affects audit quality, leading investors to incorporate less of the reported earnings information into stock returns. Panels B and C of Table 6 report the results of the association between the within Big 4 concentration metric and cost of capital measures. Results show that the higher the within Big 4 market concentration (Big4Concen), the higher the IndEP (coef = 0.051, p = 0.070) and CostDebt (coef = 0.067, p = 0.074). Put differently, both equity and debt investors appear to require a higher rate of return (i.e., a higher cost of capital) for client firms in markets with higher within Big 4 market concentration, suggesting that audit quality is perceived to be lower in these markets. [Insert Table 6 Here] 4.3 Hypothesis 3: Effects of Big4Share in PwC-ML Countries vs. Other Countries To test Hypothesis 3, we estimate our equations (1) and (2) separately for countries where PwC is the market leader and other countries. Panel A of Table 7 reports the results of the estimation for the earnings-returns association, with a sample of 10,764 firm-year observations in PwC-ML countries (representing 65% of the full sample) and a sample of 5,853 firm-year observations in other countries (representing 35% of the full sample). The coefficient for EARN*Big4Share is significant (with a positive sign) only in the PwC-ML=1 sub-sample, suggesting that investors perceive reported earnings in countries with a higher Big 4 market shares to be more credible only in countries where 22 PwC is the market leader. Also, the magnitude of the coefficient for EARN*Big4Share is higher in PwC-ML countries (0.693) than in other countries (0.420), and the difference is statistically significant (F-value = 5.35, p < 0.01). Thus the results for Hypothesis 3 suggest that earnings informativeness is increasing in Big 4 market share only in countries where PwC is the market leader. Panels B and C of Table 7 report the results of the association between Big 4 market share and cost of capital measures for PwC-ML countries and other countries. Panel B documents a negative association between Big4Share and IndEP for both PwC-ML countries (N = 7,641, representing 65% of the full sample) and other countries (N = 4,182, representing 35% of the full sample), consistent with the results reported previously in Panel B of Table 5. In addition, the results of the F-test (F-value = 4.00, p < 0.01) suggest that the coefficient for Big4Share in PwC-ML=1 countries (-0.108) is significantly smaller (more negative) than that in other countries (-0.092), suggesting that although a higher Big 4 market share is associated with a lower cost of equity in all countries, it is lower still in countries where PwC is the market leader. Panel C of Table 7 suggests similar inferences when we examine creditors’ required rate of return. The results indicate that the negative association between Big4Share and CostDebt previously documented in Panel C of Table 5, is driven by the sub-sample of client firms in countries where PwC is the market leader (N = 21,413, representing 73% of the full sample). The coefficient for Big4Share is significantly negative for PwC-ML countries (-0.043, p = 0.017), but not significantly different from zero for other countries (-0.014, p = 0.436). Note that the difference in the magnitude of the coefficients is statistically significant (F-value = 4.79, p < 0.01). As in the case of equity investors, the results suggest that the higher the Big 4 market share, the lower the creditors’ required rate of return (cost of debt capital) for clients in countries where PwC is the market leader. [Insert Table 7 Here] 4.4 Hypothesis 4: Effects of Big4Concen in PwC-ML Countries vs. Other Countries Because the world’s largest auditor (PwC) has the most reputation capital at risk and potentially the necessary incentive to provide a quality audit particularly in countries where it is also the market leader, the joint effect of PwC’s global- and country-level reputation as market leader may be expected to mitigate the negative consequences of audit market concentration on perceived audit 23 quality. Once again, we test our hypothesis by separately estimating equations (1) and (2) for countries where PwC is the market leader and other countries. Panel A of Table 8 shows that the negative association between within Big 4 market concentration and earnings informativeness (reported previously in Panel A of Table 6) is driven by client observations in countries where PwC is not the market leader (coef for EARN*Big4Concen = -2.216, p = 0.013). Specifically, the coefficient for EARN*Big4Concen is not significant (0.030, p = 0.974) for client firms in PwC-ML=1 countries, and the difference in the size of the coefficients is statistically significant (F-value = 6.92, p < 0.01). The results show that while within Big 4 market concentration has no effect on earnings informativeness in countries where PwC is the market leader, it does lower earnings informativeness in other countries. Panels B and C of Table 8 report results for the association tests between within Big 4 market concentration and our two cost of capital measures for PwC-ML=1 countries and other countries separately. The results show that the positive association between Big4Concen and IndEP and CostDebt (as documented in Panels B and C of Table 6), i.e., the adverse effects of auditor concentration, is driven by the sub-sample of client observations from countries where PwC is not the market leader. Note that the coefficients for Big4Concen are not significantly different from zero for PwC-ML=1 countries in both the IndEP (-0.012, p = 0.745) and CostDebt regressions (0.041, p = 0.478). Further, these findings (or rather the lack of findings) hold despite the large sample sizes involved (N = 7,641 for the IndEP test, and N = 21,413 for the CostDebt test) for the PwC-ML=1 countries. [Insert Table 8 Here] Collectively, our findings suggest that auditor concentration has a significant impact on investor-perceived audit quality as proxied by earnings informativeness and the investors’ required rate of return on equity and debt. Specifically, we examine two alternative measures of auditor concentration, overall Big 4 market share (Big4Share) and unequal market shares within the Big 4 (Big4Concen). With respect to the first measure Big4Share, Big 4 audit clients in countries with a larger Big 4 market share are perceived to have higher audit quality, i.e., other things being equal, the higher the Big 4 market share, the greater the earnings informativeness and the lower the investors’ 24 required rate of return for equity and debt. Further, these capital market effects are stronger in countries where PwC is the market leader. With respect to the second measure Big4Concen, Big 4 audit clients in countries with more concentration within the Big 4 are perceived to have lower audit quality, i.e., other things being equal, the more unequal the Big 4 market shares, the lower the earnings informativeness and the higher the investors’ required rate of return for equity and debt. However, these negative capital market effects are observed only in countries where PwC is not the market leader. Put differently, these negative effects are not observed in countries where PwC is the market leader. All these findings are robust to a number of client-level, industry-level and countrylevel controls, as well as the adjustment for potential client-level clustering effects. 4.5 Effects of Audit Market Concentration on Factual Audit Quality in PwC-ML=1 Countries vs. Other Countries Francis et al. (2012) examine the association between the two measures of audit market concentration (Big4Share and Big4Concen) and factual audit quality as proxied by total accruals (TOT_ACC), abnormal accruals (AB_ACC) and the probability of reporting a profit (Prob.[PROFIT]). While the focus of our study is perceived audit quality, we also test whether PwC market leadership at the country-level impacts the relation between Big4Share and Big4Concen and factual audit quality. Specifically, we re-do the analyses in Tables 7 and 8 using Francis et al.’s (2012) factual audit quality metrics as our dependent variables.17 The results of these analyses are reported in Table 9. After excluding observations with missing values in computing the factual audit quality measures, the sample include 16,546 client-year observations, with 11,359 observations for the PwC-ML=1 countries (i.e., countries where PwC is the market leader) and 8,187 observations for other countries. For brevity, in Table 9 we report only the coefficients of interest. 17 Francis et al.’s (2012) factual audit quality metrics are defined as follows: (1) TOT_ACC is net income before extraordinary items less cash flows from operations, scaled by lagged total assets; (2) AB_ACC is the residual from a regression predicting non-discretionary accruals as calculated in Dechow et al (1995) and Kothari et al (2005). The model specification used is: NDAt = α + β1 (1/ASSETSit-1) + β2 (∆SALESit - ∆ARit) + β3 (PPEit) + β4 (ROAit) + Year Fixed Effects + Industry Fixed Effects + Country Fixed Effects + ε where NDA is non-discretionary accruals (net income less cash flows, scaled by lagged assets), ASSETS is a firm’s total assets, SALES is sales, AR is accounts receivable, PPE is gross property, plant and equipment and ROA is return on assets. AB_ACC is the residual from this regression calculated over separate industry-year groupings where industries are based on 2-digit SIC codes; and (3) PROFIT is 1 when a firm’s net income for the year is above zero, and 0 otherwise. 25 Panel A of Table 9 presents the results of the association between Big4Share (Big 4 market share) and factual audit quality for PwC-ML=1 countries and other countries separately. Consistent with Francis et al.’s (2012) findings, we find that Big4Share is negatively associated with all three measures of factual audit quality (TOT_ACC, AB_ACC and Prob.[PROFIT]), but the association is stronger (or exists) only in countries where PwC is the market leader. Recall that the higher the factual audit quality metrics, the lower the audit quality. Hence, the results suggest that factual audit quality is higher in countries when Big 4 market share is higher, and that such association is stronger (or exists) only in countries where PwC is the market leader. Similarly, Panel B of Table 9 reports that Big4Concen is positively associated with all three factual audit quality metrics but only in countries where PwC is not the market leader. Since higher values of the factual audit quality metrics imply lower factual audit quality, our findings suggest that the adverse impact of within Big 4 market concentration on factual audit quality is observed only in countries where PwC is not the market leader. Collectively, our findings suggest that the positive effects of Big4Share for factual audit quality (previously documented by Francis et al. 2012) are stronger (or exist) only in countries where PwC is the market leader. Similarly, our findings suggest that the negative effects of Big4Concen for factual audit quality (also previously documented by Francis et al. 2012) are present only in countries where PwC is not the market leader. Notably, these findings are consistent with the results of our main tests (reported previously) for perceived audit quality. [Insert Table 9 Here] 4.6 PwC Country-Level Market Leadership and Investor Protection As noted previously, the correlation between our PwC country-level market leader variable (PwC-ML) and the investor protection variable (INV_PROT) is 0.569 (see Panel B of Table 3). This finding potentially suggests that the differential results we find for countries where PwC is the market leader and other countries may be due to differences in the level of investor protection. In other words, in countries where PwC is the market leader, (1) the stronger favorable effects of Big 4 market dominance (Big4Share), and (2) the absence of adverse effects of within Big 4 concentration (Big4Concen), could be due either to PwC’s greater global- and country-level reputation capital at risk (as we have argued) or simply due to stronger investor protection in these countries. 26 To examine this possibility, we repeat the analyses in Tables 7 and 8 based on partitioning the sample into high and low investor protection countries, i.e., based on whether INV_PROT is above or below the sample median. Our results (untabulated) show that the findings observed in Tables 7 and 8 do not hold when we partition our sample observations by countries with high vs. low investor protection. These findings suggest that the results reported in Tables 7 and 8, i.e., that the favorable effects of Big 4 market dominance (Big4Share) are stronger and that the adverse effects of within Big 4 concentration (Big4Concen) are absent in countries where PwC is the market leader (relative to other countries) is not driven by differences in the level of investor protection. To further investigate the unique effects of PwC’s global- and country-level market leadership, we re-estimate Tables 7 and 8 using country-level partitions based on market leadership by each of the other remaining Big N firms, i.e., rather than partition by PwC-ML, we partition by EYML, DTT-ML and KPMG-ML to identify countries where the market leader is Ernst & Young, Deloitte and KPMG, respectively. However, for these alternative partitions, our results (untabulated) suggest no differences between countries where EY (or DTT or KPMG) is the market leader and other countries. Collectively, our findings suggest unique reputation effects associated with PwC’s joint global- and country-level market leadership in enhancing the positive and mitigating the negative effects of audit market concentration on audit quality. As a caveat, we note that although we show that the unique effects associated with PwC’s global- and country-level leadership cannot be explained by country-level differences in investor protection, it is always possible that PwC’s country-level market leadership is capturing some unobserved country-level characteristic. This remains an open question and we leave it for future research. 4.7 Additional Analyses 4.7.1 Alternative Measures of Big4Share and Big4Concen We use clients’ total assets (rather than total sales) as the basis to measure Big4Share and Big4Concen and re-run all the analyses. The results are qualitatively similar to our previously reported findings. In addition, we follow Francis et al. (2012) in measuring Big4Share and 27 Big4Concen at the country-industry-year (rather than country-year) level.18 Our expectation is that investor-perceived audit quality is related more to the auditor’s overall global- and country-level market leadership rather than industry-specific investments. Consistent with this expectation, for the alternative analyses, we find that while the signs of the variables of interest are similar to those reported in Tables 5 through 8, the results are weaker (results not tabulated for brevity). For example, For the earnings informativeness test, the coefficient on EARN*Big4Share for the overall sample in the earnings informativeness test remains positive, but the magnitude is smaller and not statistically significant (coef = 0.195, p = 0.252). However, when we partition the sample into PwC-ML=1 and other countries, we find the coefficients for EARN*Big4Share to be positive and significant for the PwC-ML=1 countries, negative for the PwC-ML=0 countries, and significantly different between the two sub-samples (F-value = 4.79, p < 0.01). For the cost of equity test, the coefficient on Big4Share remains negative and significant in the overall sample, but smaller (coef = -0.089, p < 0.01). Further, consistent with the information previously reported in Table 7, the coefficient on Big4Share remains more strongly negative in the PwC-ML=1 sub-sample (coef = -0.091, p < 0.01) than in the PwCML=0 sub-sample (coef = -0.077, p < 0.01), and the difference in coefficients between the two subsamples is significant (F-value = 3.30, p = 0.03). Finally, for the cost of debt test, the results using the alternative Big4Share measure show that the coefficient (although smaller) remains negative and significant in the overall sample (coef = -0.014, p = 0.055). Once again, consistent with the information previously reported in Table 7, the coefficient on Big4Share remains more strongly negative in the PwC-ML=1 sub-sample (coef = -0.039, p < 0.01) than in the PwC-ML=0 sub-sample (coef = -0.003, p = 0.792), and the difference in coefficients is significant (F-value = 5.91, p < 0.01). The results are similar for the Big4Concen measure. 4.7.2 Alternative Cost of Equity Capital Measures We use an alternative cost of equity capital measure based on the Easton (2004) model in lieu of IndEP and re-run all the tests. Our findings are robust to the alternative cost of equity capital metric. 18 Note that we use GIND industry classifications available in WorldScope because the number of observations for each country-industry-year combination using two-digit SIC (as in Francis et al. 2012) is too small to compute meaningful Big4Share and Big4Concen measures for our sample. 28 4.7.3 Inclusion of Both Big4Share and BigConcen Following Francis et al. (2012), we include both Big4Share and Big4Concen in the same model and re-run all the tests. Although untabulated, our findings remain unchanged. 4.7.4 Country Exclusions We exclude each country in our sample one at a time and re-perform our analyses to check whether our results are driven by the observations from any given country. The untabulated results are qualitatively similar to our main results. 4.7.5 Exclusion of Andersen Clients Our sample includes Andersen clients although (for convenience of exposition) we use the term Big 4 throughout the paper. We exclude Andersen clients from our sample and re-run all the analyses. The results are qualitatively similar to our main findings. 4.7.6 Exclusion of Year 2002 We exclude year 2002 observations from our sample because Arthur Andersen’s collapse that year resulted in an unusually large number of auditor-client realignments during the fiscal year. Our results are robust to the exclusion of year 2002 observations from our sample. 4.8 Other Sensitivity Tests In an untabulated sensitivity test, we also delete from our sample the top and bottom five percent of client-year observations with extreme values of Big4Share and Big4Concen. The findings remain unchanged, indicating that our results are not driven by observations with extreme values. 5. CONCLUDING REMARKS Audit market concentration and its potential impact on audit quality is currently an important public policy issue (European Commission 2010; GAO 2003, 2008; Oxera 2006; UK House of Lords 2011). In particular, a suggested remedy for audit market concentration is to break-up the large Big 4 firms. Presumably, the largest Big 4 firm would be the primary target for any such break-up enforced by the anti-trust authorities. Our analysis examines the 1999-2004 time period when the world’s largest auditor was PwC. 29 Since audit quality is not directly observable, investors rely on the auditor’s reputation to assess the quality of the audit. For regulators, evidence about investors’ perception of audit quality is potentially as important as evidence about factual audit quality (Levitt 1998). To our knowledge, there is little or no prior research on the relation between auditor concentration and investor-perceived audit quality. Our cross-country study provides timely evidence on this important public policy issue. We examine whether two auditor concentration metrics – Big 4 market dominance and concentration within the Big 4 – affect investor-perceived audit quality as proxied by the informativeness of reported earnings as well as the investors’ required rate of return for equity and debt financing. Our findings suggest that investor-perceived audit quality is higher in countries with greater Big 4 market dominance and lower in countries with higher within Big 4 auditor concentration. We also examine whether the relation between the two auditor concentration metrics and investor-perceived audit quality is impacted by whether the world’s largest auditor (PwC) is also the country-level market leader. Equal market shares imply an equal prospect for reputational injury and thus an equal incentive for audit quality across all the Big 4 firms (Coffee 2006). By contrast, as the world’s largest auditor, PwC faces a higher prospect of reputational injury (i.e., loss of revenues/clients and market share) particularly in countries where it is also the market leader. Our findings suggest that PwC’s reputation as the world’s largest auditor has a favorable joint effect on investor-perceived Big 4 audit quality in countries where it is also the largest auditor, i.e., we find the positive effects of Big 4 market dominance to be stronger and the negative effects of within Big 4 concentration to be absent in countries where PwC is the market leader. In additional analysis, we find similar results also with respect to factual audit quality. Collectively, our results address potential regulatory concerns over the possible adverse effect of Big 4 auditor concentration for investors’ assessment of audit quality. In particular, we find no evidence to suggest that a break-up of the world’s largest Big 4 firm (PwC) to lower audit market concentration would improve perceived (or factual) audit quality. Rather, the evidence suggests that a break-up of the world’s largest auditor could adversely affect audit quality in countries where it is also the market leader. Our findings are important in documenting the joint effect of a Big 4 firm’s global- 30 and country-level reputation as market leader in enhancing the positive -- and mitigating the negative – consequences of audit market concentration on perceived (as well as factual) audit quality. Altogether, our study contributes to the limited prior research on auditor concentration by providing more complete evidence on the relation between concentration and both factual and perceived audit quality. REFERENCES Alford, A. 1992. The effect of the set of comparable firms on the accuracy of the price-earnings valuation method. Journal of Accounting Research 30: 94–108. Ashbaugh, H., R. LaFond, and B. Mayhew. 2003. Do nonaudit services compromise auditor independence? Further evidence. The Accounting Review 78(3): 611-639. Bae, K., and V. K. Goyal. 2009. Creditor rights, enforcement, and bank loans. The Journal of Finance 64: 823-60. Ball, R., S.P Kothari, and A. Robin. 2000. The effect of international institutional factors on properties of accounting earnings. Journal of Accounting and Economics 29: 1-51. Baumol, W., J. Panzer, and R. Willig. 1982. Contestable Markets and the Theory of Industrial Structure. San Diego, CA: Harcourt Brace. Becker, G., and K. Murphy. 1993. A simple theory of advertising as a good or bad. Quarterly Journal of Economics 108(4): 941-964. Becker, C., M. DeFond, J. Jiambalvo, and K. Subramanyam. 1998. The effects of audit quality on earnings management. Contemporary Accounting Research 15(1): 1-24. Causholli, M., W. Knechel, H. Lin, and D. Sappington. 2010. Competitive procurement of a credence good: The case of auditing. Working paper: University of Florida. Choi, J., and T. Wong., 2007. Auditors’ governance functions and legal environments: an international investigation. Contemporary Accounting Research 24: 1-36. Chung, H., and S. Kallapur. 2003. Client importance, nonaudit services, and abnormal accruals. The Accounting Review 78(4): 931-956. Coffee, J. 2006. Gatekeepers: The Professions and Corporate Governance. New York, NY: Oxford University Press. Cohen, A., and M. Mazzeo. 2004. Competition, product differentiation and quality provision: An empirical equilibrium analysis for bank branching decisions. Financial and Economics Discussion Series. Washington, DC: Federal Reserve Board. Crespi, J., and S. Marette. 2009. Quality, sunk costs and competition. Review of Marketing Science 7: 1-34. DeAngelo, L., 1981. Auditor size and audit quality. Journal of Accounting and Economics 3: 183199. 31 Dechow, P., Dichev, I., 2002. The quality of accruals and earnings: the role of accrual estimation errors. The Accounting Review 77 (Supplement), 35–59. Dechow, P., R. Sloan, and A. Sweeney. 1995. Detecting Earnings management. The Accounting Review 70 (2): 193-225. Dedman, E., and C. Lennox. 2009. Perceived competition, profitability and the withholding of information about sales and the cost of sales. Journal of Accounting and Economics 48: 210-230. Demsetz, H. 1973. Industry structure, market rivalry and public policy. Journal of Law and Economics 16: 1-9. Dick, A. 2007. Market size, service quality and competition in banking. Journal of Money, Credit, and Banking 39: 49-81. Domberger, S., and A. Sherr. 1989. The impact of competition on pricing and quality of legal services. International Review of Law and Economics 9: 41-56. Dranove, D., and W. White. 1994. Recent theory and evidence on competition in hospital markets. Journal of Economics and Management Strategy 3(1): 169-209. Easton, P. 2004. PE ratio, PEG ratios, and estimating the implied expected rate of return on equity capital. The Accounting Review 79 (1): 73-95. Ellickson, P. 2007. Does Sutton apply to supermarkets? The RAND Journal of Economics 38: 43-59. European Commission (EC). 2010. Green paper. Audit policy: lessons from the crisis. http://ec.europa.eu/internal_market/consultations/docs/2010/audit/green_paper_audit_en.pdf Financial Accounting Standards Board (FASB). 1978. Objectives of Financial Reporting by Business Enterprise. Statement of Financial Accounting Concepts No. 1. Norwalk, CT: FASB. Financial Accounting Standards Board (FASB). 1980. Qualitative Characteristics of Accounting Information. Statement of Financial Accounting Concepts No. 2. Stamford, CT: FASB. Firth, M., P. M. Y. Fung and O. M. Rui, 2007, Ownership, two-tier board structure, and the informativeness of earnings – Evidence from China, Journal of Accounting and Public Policy, 26: 463-496. Francis, J., I. Khurana, and R. Pereira. 2003. The role of accounting and auditing in corporate governance and the development of financial markets around the world. Asia-Pacific Journal of Accounting and Economics 10: 1-30. Francis, J., I, Khurana, and R, Pereira. 2005a. Disclosure incentives and effects on cost of capital around the world. The Accounting Review 80: 1125-62. Francis, J., R. La Fond, P. Olsson, and K. Schipper. 2005b. The market pricing of accruals quality. Journal of Accounting and Economics 39: 295-327. Francis, J. 2011. A framework for understanding and researching audit quality. Auditing: A Journal of Practice & Theory 30: 125-152. Francis, J., and B. Ke. 2006. Disclosure of fees paid to auditors and the market valuation of earnings surprises. Review of Accounting Studies 11: 495-523. 32 Francis, J., P. Michas, and S. Seavy. 2012. Does audit market concentration harm the quality of audited earnings? Evidence from audit markets in 42 countries. Contemporary Accounting Research, forthcoming. Francis, J., E. Maydew, and H. Sparks. 1999. The role of Big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice and Theory 18(2): 17-34. Francis, J., and D. Wang. 2008. The joint effect of investor protection and Big 4 audits on earnings quality around the world. Contemporary Accounting Research 25(1): 157-91. George, L. 2007. What's fit to print: The effect of ownership concentration on product variety in daily newspaper markets. Information Economics and Policy 19 (October): 285-303. Ghosh, A. and D. Moon. 2005. Auditor tenure and perceptions of audit quality. The Accounting Review 80(2): 585-612. Government Accountability Office (GAO). 2003. Public Accounting Firms: Mandated Study on Consolidation and Competition. Washington DC: GAO. Government Accountability Office (GAO). 2008. Audits of Public Companies: Continued Concentration in Audit Market for Large Public Companies Does Not Call for Immediate Action. Washington DC: GAO. Gul, F.A., S.Y-K. Fung and B. Jaggi. 2009. Earnings Quality: Some Evidence on the Role of Auditor Tenure and Auditors’ Industry Expertise. Journal of Accounting and Economics 47(3): 265-287. Hardy, J. A. 1993. Regression with dummy variables. Sage University Paper series on Quantitative Application in the Social Sciences, 07-093, Sage, Newbury Park, CA Hung, M., 2000. Accounting standards and value relevance of financial statements: an international analysis. Journal of Accounting and Economics 30: 401-420 Jackson, H. and M. J. Roe. 2009. Public and Private Enforcement of Securities Laws: Resource-Based Evidence. Journal of Financial Economics 93(2): 207-238. Kaplan, R., Urwitz, G., 1979. Statistical models of bond ratings: a methodological inquiry. Journal of Business (April), 231–261. Kaufmann, D., A. Kraay, and M. Mastruzzi. 2008. Governance matters VII: aggregate and individual governance indicators, 1996-2007. World Bank Policy Research working paper No. 4653 Khurana, I., and K. Raman. 2004. Litigation risk and the financial reporting credibility of Big 4 versus non-Big 4 audits: Evidence from Anglo-American countries. The Accounting Review 79(2): 473-495. Kothari, S., A. Leone, and C. Wasley. 2005. Performance matched discretionary accrual measures. Journal of Accounting and Economics 39(1): 163-197. Kwon, S., C. Lim, and P. Tan. 2007. Legal systems and earnings quality: The role of auditor industry specialization. Auditing: A Journal of Practice and Theory 26: 25-56. La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R. Vishny. 1998. Law and finance. The Journal of Political Economy 106: 1113-55. 33 La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R. Vishny. 2006. What works in securities laws? The Journal of Finance 61: 1–32. Leuz, C., D. Nanda, and P. Wysocki. 2003. Earnings management and investor protection: An international comparison. Journal of Financial Economics 69: 505-527. Levitt, A. 1998. The Numbers Game. Unpublished remarks of the Chairman of the US Securities and Exchange Commission delivered at the New York University Center for Law and Business (September): www.sec.gov. Liu, J., Nissim, D., Thomas, J., 2002. Equity valuation using multiples. Journal of Accounting Research40, 135–172. Mansi, S.A., W.F. Maxwell, and D.P. Miller. 2004. Does Auditor Quality and Tenure Matter to Investors? Evidence from the Bond Market. Journal of Accounting Research 42(4): 755-793. Mazzeo, M. 2003. Competition and service quality in the US airline industry. Review of Industrial Organization 22: 275-296. Oxera. 2006. Competition and Choice in the UK Audit Market. Oxera Consulting Limited: Oxford, UK. Palepu, K., Healy, P., Bernard, V., 2000. Business Analysis and Valuation, second ed. Southwestern Publishing. Paulson, H. 2006. Remarks by Treasury Secretary Henry M. Paulson on the Competitiveness of US Capital Markets – Economic Club of New York, NY. www.treas.gov. Pittman, J. A. and S. Fortin. 2004. Auditor choice and the cost of debt capital for newly public firms. Journal of Accounting and Economics 37: 113-136. Robinson, W., and J. Chiang. 1996. Are Sutton’s predictions robust? Empirical insights into advertising, R&D, and concentration. Journal of Industrial Economics (December): 389-408. Schipper, K. and L. Vincent. 2003. Earnings quality. Accounting Horizons (Supplement) 17: 97–110. Schmalensee, R. 1978. A model of advertising and product quality. Journal of Political Economy 86(3): 485-503. Shaked, A., and J. Sutton. 1987. Product differentiation and industrial structure. Journal of Industrial Economics 36: 131-152. Shepherd, W., and J. Shepherd. 2004. The Economics of Industrial Organization. Fifth edition. Long Grove, IL. Waveland Press. Simunic, D., and M. Stein. 1987. Product Differentiation in Auditing: Auditor Choice in the Market for Unseasoned New Issues. Vancouver, Canada: The Canadian Certified General Accountants’ Research Foundation. Sirois, L., and D. Simunic. 2010. Auditor size and audit quality revisited: The importance of audit technology. Ssrn.com. Spence, A. 1975. Monopoly, quality and regulation. Bell Journal of Economics 6:407-414. 34 Stiglitz, J. 1987. The causes and consequences of the dependence of quality on prices. Journal of Economic Literature 25, 1–48. Sutton, J. 1991. Sunk Costs and Market Structure. Cambridge, MA: MIT Press. Teoh, S., and T. Wong. 1993. Perceived auditor quality and the earnings response coefficient. The Accounting Review 68: 346-366. UK House of Lords. 2011. Auditors: Market Concentration and their Role. Select Committee on Economic Affairs. US Treasury. 2008. Advisory Commission on the Auditing Profession: Final Report. (0ctober 6): www.treas.gov. Warfield, T. D., J. J. Wild, and K. L. Wild. 1995. Managerial ownership, accounting choices, and informativeness of earnings. Journal of Accounting and Economics 20 (1): 61-91. 35 APPENDIX Variable Variable Definitions Definition Dependent Variables: RETURN = 12-month stock returns ending three months after fiscal yearend. IndEP = cost of equity capital, computed as industry-adjusted E/P ratio (following Francis et al. 2005b). CostDebt = cost of debt capital, computed as the ratio of interest expense in year t to average interest-bearing debt outstanding during year t (following Francis et al. 2005b). Test Variables: Big4Share = the percentage of listed clients in a country with a Big 4 auditor in year t. The higher the metric, the higher the aggregate market share of the Big 4 in that country. Big4Concen = Herfindahl index based on Big 4 market shares, calculated as å [s/ S]2 where s is the dollar value of total client sales for each of the Big 4 auditors in a country, and S is the total dollar value of client sales for all Big 4 auditors in that country. The square is summed over all Big 4 auditors that perform audits in a given country-year. The higher the metric, the higher the concentration within the Big 4. PwC-ML = 1 if PwC is the market leader, i.e., has the single largest market share based on client sales (or assets) in a given country-year, and 0 otherwise. Client-Level Control Variables: EARN = earnings per share scaled by stock price at prior fiscal year-end. SIZE = the natural log of sales in year t. LEV = total liabilities scaled by total assets in year t. ROA = the return on assets measured as net income divided by total assets in year t. GROWTH = the natural log of one plus the percentage change in book value of equity over the past 5 years. BETA = the 5-year rolling beta from client-specific CAPM estimations using the past 5 years of data. TANGIBLE = the ratio of fixed assets to total assets in year t. 36 TOT_ACC = total accruals, i.e., net income before extraordinary items less cash flows from operations, scaled by lagged total assets. Country-Level Control Variables: FIN_DEVEL = the country’s total market capitalization in year t scaled by GDP, from World Bank. LRATE = the country-level lending interest rate, from World Bank. LNGNI = the natural log of one country’s gross national income per capita. COMMON = 1 if a country’s legal origin is based on English common law, and 0 otherwise, from La Porta et al. (1998). ANTIDIR = the measure of anti-director rights taken from La Porta et al. (1998). DISC = the disclosure rights index from La Porta et al. (2006). LIAB = the liability standard index from La Porta et al. (2006). PUBENF = the public enforcement of securities laws index from La Porta et al. (2006). SEC_STAFF = the size of a country’s securities regulator staff scaled by total population, taken from Jackson and Roe (2009). INV_PROT = a factor analysis (principal components analysis) with varimax rotation of the variables COMMON, ANTIDIR, DISC, LIAB, PUBENF and SEC_STAFF. All variables load on one factor. RULE_OF_LAW = the rule of law from Kaufmann et al. (2008). Data are from Compustat Global Vantage unless noted otherwise. 37 TABLE 1 Sample Selection Firm-year observations Number of distinct client firms Firm-year observations in Compustat Global Vantage with positive total assets from 1999 to 200419 Less: Firm-years audited by a non-Big 4 auditor Firm-years with missing country-level data Financial and Utility firms Primary sample 75,792 15,061 (19,337) (4,560) (3,496) 48,399 (3,267) (1,022) (615) 10,157 For Earnings-informativeness test Primary sample less: Firm-years with missing client-level data Earnings-informativeness sample (31,782) 16,617 (3,939) 6,218 For Cost of Equity test Primary sample less: Firm-years with missing client-level data Cost of Equity sample (36,576) 11,823 (4,429) 5,728 For Cost of Debt test Primary sample less: Firm-years with missing client-level data Cost of Debt sample (16,588) 31,811 (2,029) 8,128 19 Following Francis and Wang (2008), we exclude observations from Japan, South Korea, India and Pakistan due to the potential miscoding of the auditor identification variable. TABLE 2 38 Number of countries in which an individual Big 4 firm is market leader (i.e., has the largest market share based on client sales or assets) Panel A Based on all Global Vantage countries 1999 2000 2001 2002 2003 2004 Arthur Andersen 12 9 7 Deloitte 5 10 7 15 12 12 Ernst & Young 6 6 8 11 12 11 KPMG 14 12 11 10 11 15 PwC 23 23 28 25 25 26 Total 60 60 61 61 60 64 Panel B Based on countries in our sample 1999 2000 2001 2002 2003 2004 Arthur Andersen 5 4 3 Deloitte 1 3 2 5 5 7 Ernst & Young 5 4 5 8 7 5 KPMG 5 6 5 7 3 5 PwC 13 12 14 9 14 12 Total 29 29 29 29 29 29 Note: PwC was the market leader in Canada, Finland, Hong Kong, Mexico, Malaysia, and US throughout 19992004. In addition, it was the market leader in Australia, Brazil, Belgium, Singapore, Sweden, Switzerland, and UK in 1999, Australia, Brazil, Israel, Singapore, Switzerland, and Netherlands in 2000, Australia , Brazil, Chile, Germany, New Zealand, Singapore, Switzerland, and UK in 2001, Brazil, Sweden, and Switzerland in 2002, Australia, Brazil, Germany, Switzerland, Sweden, Netherlands, Philippines, and UK in 2003, and Brazil, Denmark, Israel, Switzerland, Sweden, and UK in 2004. 39 TABLE 3 Panel A COUNTRY Argentina Australia Austria Belgium Brazil Canada Chile Denmark Finland France Germany Greece Hong Kong Ireland Israel Italy Malaysia Mexico Netherlands New Zealand Norway Philippines Singapore Spain Sweden Switzerland Thailand U.K. U.S. Means N 96 1665 274 459 634 2820 378 724 634 1666 1875 200 742 281 123 1227 2830 349 988 318 813 104 2245 739 1476 1014 882 5451 17392 Descriptive Statistics (Country-Level Variables) Means for test variables and country-level control variables Big4Share 0.61 0.78 0.36 0.48 0.70 0.56 0.75 0.65 0.63 0.28 0.33 0.29 0.66 0.47 0.42 0.59 0.53 0.52 0.63 0.43 0.64 0.28 0.67 0.56 0.67 0.65 0.37 0.34 0.61 0.51 Big4Concen 0.43 0.27 0.40 0.29 0.44 0.25 0.39 0.33 0.59 0.27 0.39 0.38 0.28 0.31 0.45 0.30 0.29 0.35 0.28 0.33 0.32 0.63 0.28 0.49 0.33 0.32 0.36 0.27 0.25 0.35 See Appendix for variable definitions. FIN_DEVEL 0.55 1.10 0.19 0.67 0.35 1.08 0.89 0.57 1.59 0.86 0.50 0.70 3.45 0.66 0.57 0.49 1.44 0.22 1.21 0.39 0.42 0.50 1.94 0.79 1.08 2.43 0.52 1.49 1.41 0.97 LRATE 0.23 0.08 0.06 0.07 0.64 0.06 0.10 0.07 0.05 0.07 0.10 0.09 0.06 0.04 0.11 0.06 0.07 0.13 0.04 0.09 0.07 0.11 0.06 0.04 0.05 0.04 0.07 0.05 0.06 0.10 LNGNI 8.52 9.95 10.19 10.16 8.15 10.05 8.45 10.40 10.20 10.13 10.16 9.55 10.16 10.13 9.77 9.99 8.26 8.67 10.23 9.62 10.61 6.95 10.01 9.73 10.30 10.63 7.66 10.20 10.48 9.63 COMMON 0 1 0 0 0 1 0 0 0 0 0 0 1 1 1 0 1 0 0 1 0 0 1 0 0 0 1 1 1 0.38 ANTIDIR 4 4 2 0 3 5 5 2 3 3 1 2 5 4 3 1 4 1 2 4 4 3 4 4 3 2 2 5 5 3.10 DISC 0.50 0.75 0.25 0.42 0.25 0.92 0.58 0.58 0.50 0.75 0.42 0.33 0.92 0.67 0.67 0.67 0.92 0.58 0.50 0.67 0.58 0.83 1 0.50 0.58 0.67 0.92 0.83 1 0.65 LIAB 0.22 0.66 0.11 0.44 0.33 1 0.33 0.55 0.66 0.22 0 0.49 0.66 0.44 0.66 0.22 0.66 0.11 0.89 0.44 0.39 1 0.66 0.66 0.28 0.44 0.22 0.66 1 0.50 PUBENF 0.58 0.90 0.17 0.15 0.58 0.80 0.60 0.37 0.32 0.77 0.22 0.32 0.86 0.37 0.63 0.48 0.77 0.35 0.47 0.33 0.32 0.83 0.87 0.33 0.50 0.33 0.72 0.68 0.90 0.54 SEC_STAFF 3.50 34.40 10.00 13.80 2.70 38.90 9.90 10.90 11.20 5.90 4.40 12.20 59.60 23.30 18.80 7.30 22.40 5.20 23.50 8.90 20.80 4.30 77.70 8.50 7.20 8.90 6.50 19.00 23.80 17.36 INV_PROT -1.23 0.31 -2.18 -2.08 -1.58 0.79 -0.86 -1.37 -1.28 -0.93 -2.23 -1.70 0.82 -0.50 -0.30 -1.56 0.26 -1.86 -1.05 -0.65 -1.14 -0.31 0.91 -1.15 -1.30 -1.40 -0.50 0.21 0.84 0.00 RULE_OF_LAW -0.49 1.77 1.83 1.45 -0.32 1.75 1.18 1.85 1.88 1.34 1.70 0.77 1.11 1.59 0.89 0.78 0.43 -0.41 1.75 1.81 1.90 -0.46 1.59 1.29 1.83 1.96 0.20 1.73 1.60 1.18 40 TABLE 3 Panel B Big4Concen PwC-ML FIN_DEVEL LRATE LNGNI COMMON ANTIDIR DISC LIAB PUBENF SEC_STAFF INV_PROT RULE_OF_LA W Descriptive Statistics (Country-Level Variables) Correlation between test variables and country-level control variables Big4Share Big4Concen -0.120 0.263 0.319 0.124 0.142 0.015 0.184 0.238 0.391 0.161 0.249 0.231 0.268 1 -0.333 -0.333 0.196 -0.327 -0.555 -0.484 -0.632 -0.505 -0.603 -0.312 -0.605 -0.252 PwCML 1 0.372 0.045 0.120 0.520 0.547 0.497 0.597 0.521 0.215 0.569 0.050 FIN_DEVEL LRATE LNGNI COMMON 1 -0.219 0.228 0.438 0.448 0.492 0.416 0.425 0.499 0.516 0.239 1 -0.369 -0.209 -0.144 -0.352 -0.199 -0.087 -0.183 -0.226 -0.467 1 0.053 0.262 0.182 0.358 0.112 0.129 0.210 0.838 1 0.799 0.864 0.718 0.813 0.541 0.901 0.089 All pairwise correlations were significant at the 0.10 level (2-tail). See Appendix for variable definitions. ANTIDI R 1 0.763 0.794 0.781 0.410 0.887 0.268 DISC LIAB 1 0.758 0.899 0.516 0.936 0.103 1 0.745 0.418 0.864 0.284 PUBEN F 1 0.508 0.925 0.013 SEC_STAFF INV_PROT 1 0.621 0.182 1 0.206 41 TABLE 4 Descriptive Statistics (Client-Level Variables) Panel A Descriptive statistics for dependent variables, test variables and clientlevel control variables Variable Mean Median Std. Dev Earnings-informativeness analysis (N=16,617) RETURN 0.133 0.018 0.650 0.013 0.160 0.111 0.070 0.207 Big4Share 0.534 0.554 0.153 Big4Concen 0.287 0.266 0.075 EARN 0.013 0.031 0.192 SIZE 5.594 5.622 2.177 LEV 0.632 0.580 0.411 ROA -0.010 0.031 0.210 GROWTH 0.055 0.109 0.750 BETA 0.385 0.142 0.512 TANGIBLE 0.349 0.283 0.287 TOT_ACC -0.011 -0.010 0.130 Cost of Equity analysis (N=11,823) IndEP 0.047 Cost of Debt analysis (N=31,181) CostDebt Test variables Control variables See Appendix for variable definitions. 42 TABLE 4 Panel B Descriptive Statistics (Client-Level Variables) Correlations between dependent variables, test variables and client-level control variables RETURN IndEP CostDebt Big4Share IndEP -0.049 1 CostDebt 0.007 0.011 1 Big4Share -0.035 -0.098 -0.037 1 Big4Concen -0.040 0.049 0.062 -0.112 1 EARN 0.196 0.632 -0.022 -0.023 0.016 1 SIZE -0.013 0.099 -0.072 0.044 -0.066 0.139 1 LEV 0.045 0.076 -0.051 -0.012 0.005 0.020 0.136 1 ROA 0.169 0.142 -0.034 0.010 -0.003 0.566 0.222 -0.011 1 GROWTH 0.142 0.018 0.017 0.049 -0.155 0.238 -0.017 0.107 0.242 1 BETA 0.016 -0.079 0.003 0.309 -0.320 -0.041 0.219 -0.029 -0.071 0.132 1 TANGIBLE 0.036 -0.001 -0.074 0.027 -0.003 0.067 -0.068 0.243 0.101 0.097 -0.115 1 TOT_ACC 0.044 0.002 -0.008 0.029 -0.011 0.142 -0.019 -0.101 0.195 0.063 -0.024 -0.161 Bold text indicates two-tail significance at the .10 level or less. See Appendix for variable definitions. Big4Concen EARN SIZE LEV ROA GROWTH BETA TANGIBLE 43 TABLE 5 Effects of Big 4 Market Share: Full Sample Panel A Big 4 Market Share and Earnings Informativeness Dependent variable = RETURN Predicted Variables Sign Intercept ? EARN + Big4Share ? EARN*Big4Share ? SIZE ? LEV ? ROA ? GROWTH ? BETA ? TOT_ACC ? FIN_DEVEL ? LNGNI ? INV_PROT ? RULE_OF_LAW ? EARN *SIZE + EARN *LEV + EARN *ROA + EARN *GROWTH + EARN *BETA + EARN *TOT_ACC EARN *FIN_DEVEL ? EARN *LNGNI ? EARN *INV_PROT + EARN *RULE_OF_LAW + Industry fixed effects Included Year fixed effects Included N 16617 R-square 0.181 Coeff. Estimate 0.017 1.841 -0.027 0.530 -0.021 0.047 0.597 0.050 0.060 0.038 0.005 -0.003 0.002 -0.043 0.020 -0.155 2.250 0.117 -0.011 0.117 -0.179 -0.150 0.188 0.098 Robust t-stat 0.16 1.54 -0.86 2.06 -8.00 2.76 8.69 7.02 5.05 0.67 0.53 -0.26 0.33 -2.43 1.00 -1.31 8.48 2.46 -0.13 0.38 -1.96 -1.15 3.57 0.71 P-value 0.876 0.062* 0.391 0.040** 0.000*** 0.006*** 0.000*** 0.000*** 0.000*** 0.501 0.593 0.793 0.742 0.015** 0.156 0.095* 0.000*** 0.007*** 0.449 0.352 0.050** 0.248 0.000*** 0.241 Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 44 TABLE 5 Effects of Big 4 Market Share: Full Sample (continued) Panel B Big 4 Market Share and Cost of Equity Dependent variable = IndEP Predicted Variables Sign Intercept ? Big4Share ? SIZE LEV + ROA GROWTH BETA + TOT_ACC + FIN_DEVEL ? LNGNI ? INV_PROT RULE_OF_LAW Industry fixed effects Included Year fixed effects Included N 11823 R-square 0.095 Coeff. Estimate -0.349 -0.102 0.001 0.005 -0.028 0.010 -0.027 0.019 -0.007 0.041 -0.010 -0.016 Robust t-stat. -8.22 -8.67 1.07 0.41 -1.19 3.85 -8.01 0.91 -1.90 7.34 -5.28 -1.86 P-value 0.000*** 0.000*** 0.142 0.340 0.117 0.000*** 0.000*** 0.182 0.057* 0.000*** 0.000*** 0.031** Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 45 TABLE 5 Effects of Big 4 Market Share: Full Sample (continued) Panel C Big 4 Market Share and Cost of Debt Dependent variable = CostDebt Predicted Variables Sign Intercept ? Big4Share ? SIZE LEV + ROA TANGIBLE TOT_ACC + FIN_DEVEL ? LRATE + LNGNI ? INV_PROT RULE_OF_LAW Industry fixed effects Included Year fixed effects Included N 31811 R-square 0.055 Coeff. Estimate 0.039 -0.025 -0.009 -0.007 -0.052 -0.031 -0.009 -0.010 0.502 0.010 -0.005 0.013 Robust t-stat. 1.30 -1.84 -9.59 -1.72 -4.42 -5.01 -0.68 -3.57 11.03 2.48 -2.63 2.09 P-value 0.193 0.066* 0.000*** 0.042** 0.000*** 0.000*** 0.248 0.000*** 0.000*** 0.013** 0.009*** 0.036** Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 46 TABLE 6 Effects of Within-Big 4 Market Concentration: Full Sample Panel A Within-Big 4 Market Concentration and Earnings Informativeness Dependent variable = RETURN Predicted Variables Sign Intercept ? EARN + Big4Concen ? EARN* Big4Concen ? SIZE ? LEV ? ROA ? GROWTH ? BETA ? TOT_ACC ? FIN_DEVEL ? LNGNI ? INV_PROT ? RULE_OF_LAW ? EARN *SIZE + EARN *LEV + EARN *ROA + EARN *GROWTH + EARN *BETA + EARN *TOT_ACC EARN *FIN_DEVEL ? EARN *LNGNI ? EARN *INV_PROT + EARN *RULE_OF_LAW + Industry fixed effects Included Year fixed effects Included N 16617 R-square 0.181 Coeff. Estimate 0.112 2.255 -0.140 -1.076 -0.021 0.048 0.618 0.049 0.053 0.033 0.003 -0.011 -0.003 -0.034 0.022 -0.149 2.309 0.117 0.011 0.132 -0.178 -0.117 0.181 0.039 Robust t-stat. 0.94 1.81 -2.47 -1.79 -7.87 2.84 9.08 6.82 4.53 0.59 0.31 -0.85 -0.58 -1.94 1.09 -1.27 8.77 2.46 0.13 0.43 -1.9 -0.92 3.46 0.29 P-value 0.348 0.035** 0.014** 0.074* 0.000*** 0.004*** 0.000*** 0.000*** 0.000*** 0.554 0.759 0.395 0.560 0.052* 0.137 0.102 0.000*** 0.007*** 0.447 0.332 0.057* 0.178 0.001*** 0.384 Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 47 TABLE 6 Panel B Effects of Within-Big 4 Market Concentration: Full Sample Within-Big 4 Market Concentration and Cost of Equity Dependent variable = IndEP Predicted Variables Sign Intercept ? Big4Concen ? SIZE LEV + ROA GROWTH BETA + TOT_ACC + FIN_DEVEL ? LNGNI ? INV_PROT RULE_OF_LAW Industry fixed effects Included Year fixed effects Included N 11823 R-square 0.096 Coeff. Estimate -0.409 0.051 0.002 0.000 -0.025 0.010 -0.033 0.019 -0.014 0.032 -0.010 -0.001 Robust t-stat. -9.16 1.81 1.46 0.02 -1.06 4.02 -9.77 0.93 -3.78 5.68 -5.11 -0.17 P-value 0.000*** 0.070* 0.072* 0.492 0.144 0.000*** 0.000*** 0.177 0.000*** 0.000*** 0.000*** 0.216 Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 48 TABLE 6 Panel C Effects of Within-Big 4 Market Concentration: Full Sample Within-Big 4 Market Concentration and Cost of Debt Dependent variable = CostDebt Predicted Variables Sign Intercept ? Big4Concen ? SIZE LEV + ROA TANGIBLE TOT_ACC + FIN_DEVEL ? LRATE + LNGNI ? INV_PROT RULE_OF_LAW Industry fixed effects Included Year fixed effects Included N 31811 R-square 0.056 Coeff. Estimate 0.008 0.067 -0.008 -0.007 -0.052 -0.033 -0.009 -0.013 0.486 0.010 -0.003 0.013 Robust t-stat. 0.23 1.79 -9.53 -1.63 -4.41 -5.21 -0.71 -4.57 10.70 2.54 -1.16 2.16 P-value 0.821 0.074* 0.000*** 0.051* 0.000*** 0.000*** 0.238 0.000*** 0.000*** 0.011** 0.122 0.030** Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. 49 TABLE 7 Effects of Big 4 Market Share: PwC-ML Countries vs. Other Countries Panel A Big 4 Market Share and Earnings Informativeness Dependent variable = RETURN Variables Intercept EARN Big4Share EARN*Big4Share SIZE LEV ROA GROWTH BETA TOT_ACC FIN_DEVEL LNGNI INV_PROT RULE_OF_LAW EARN *SIZE EARN *LEV EARN *ROA EARN *GROWTH EARN *BETA EARN *TOT_ACC EARN *FIN_DEVEL EARN *LNGNI EARN *INV_PROT EARN *RULE_OF_LAW Industry fixed effects PwC-ML = 1 Predicted Coeff. Robust Sign Estimate t-stat. ? 0.084 0.54 + 2.285 1.52 ? -0.003 -0.07 ? 0.693† 1.94 ? -0.023 -6.59 ? 0.047 2.23 ? 0.445 5.11 ? 0.074 6.67 ? 0.073 4.60 ? 0.062 0.81 ? -0.004 -0.31 ? -0.008 -0.41 ? 0.023 3.38 ? -0.035 -1.29 + 0.029 1.16 + -0.146 -1.04 + 2.050 6.65 + 0.208 3.43 + -0.016 -0.15 0.181 0.50 ? -0.198 -1.57 ? -0.209 -1.17 + 0.146 2.24 + 0.124 0.46 Included P-value 0.592 0.064* 0.944 0.052* 0.000*** 0.026** 0.000*** 0.000*** 0.000*** 0.415 0.755 0.684 0.001*** 0.198 0.123 0.198 0.000*** 0.001*** 0.441 0.308 0.116 0.242 0.012** 0.324 Predicted Sign ? + ? ? ? ? ? ? ? ? ? ? ? ? + + + + + ? ? + + Included PwC-ML = 0 Coeff. Robust Estimate t-stat. 0.316 1.72 0.392 0.16 -0.079 -1.76 0.420† 1.12 -0.017 -3.88 0.039 1.46 0.854 7.51 0.033 3.58 0.041 1.71 0.002 0.03 0.034 1.61 -0.030 -1.38 -0.017 -1.44 -0.035 -1.27 -0.006 -0.18 -0.139 -0.67 2.469 4.45 -0.004 -0.06 0.097 0.52 0.194 0.31 -0.108 -0.51 0.009 0.03 0.231 2.07 0.050 0.30 P-value 0.085* 0.437 0.079* 0.264 0.000*** 0.145 0.000*** 0.000*** 0.087* 0.977 0.107 0.169 0.150 0.205 0.428 0.251 0.000*** 0.477 0.301 0.373 0.609 0.973 0.019** 0.383 50 Year fixed effects N R-square Included 10764 0.152 Included 5853 0.272 † F-stat on the difference in coefficient of EARN*Big4Share between PwC-ML countries (0.693) and other countries (0.420) = 5.35 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). 51 TABLE 7 Panel B Effects of Big 4 Market Share: PwC-ML Countries vs. Other Countries (continued) Big 4 Market Share and Cost of Equity Dependent variable = IndEP Variables Intercept Big4Share SIZE LEV ROA GROWTH BETA TOT_ACC FIN_DEVEL LNGNI INV_PROT RULE_OF_LAW Industry fixed effects Year fixed effects N R-square Predicted Sign ? ? + + + ? ? Included Included 7641 0.094 PwC-ML = 1 Coeff. Robust Estimate t-stat. -0.204 -3.81 -0.108† -7.06 0.001 0.55 -0.003 -0.21 -0.038 -1.49 0.007 2.42 -0.027 -7.34 0.062 2.65 -0.003 -0.58 0.028 3.86 -0.010 -3.85 0.022 1.98 P-value 0.000*** 0.000*** 0.292 0.417 0.063* 0.008*** 0.000*** 0.004*** 0.563 0.000*** 0.000*** 0.024** Predicted Sign ? ? + + + ? ? Included Included 4182 0.092 PwC-ML = 0 Coeff. Robust Estimate t-stat. -0.456 -6.95 -0.092† -5.20 0.001 0.46 0.024 1.20 -0.069 -1.44 0.015 3.55 -0.019 -2.61 -0.051 -1.25 -0.027 -4.37 0.063 7.40 -0.001 -0.41 -0.051 -3.85 P-value 0.000*** 0.000*** 0.324 0.116 0.075* 0.000*** 0.005*** 0.105 0.000*** 0.000*** 0.341 0.000*** † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (-0.108) and other countries (-0.092) = 4.00 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). 52 TABLE 7 Panel C Effects of Big 4 Market Share: PwC-ML Countries vs. Other Countries (continued) Big 4 Market Share and Cost of Debt Dependent Variable = CostDebt Variables Intercept Big4Share SIZE LEV ROA TANGIBLE TOT_ACC FIN_DEVEL LRATE LNGNI INV_PROT RULE_OF_LAW Industry fixed effects Year fixed effects N R-square Predicted Sign ? ? + + ? + ? Included Included 21413 0.069 Coeff. Estimate 0.009 -0.043† -0.009 -0.009 -0.060 -0.031 -0.021 -0.013 0.503 0.015 -0.009 0.010 PwC-ML = 1 Robust P-value t-stat. 0.20 0.845 -2.40 0.017** -8.85 0.000*** -1.81 0.036** -4.52 0.000*** -4.31 0.000*** -1.40 0.081* -4.09 0.000*** 9.96 0.000*** 2.54 0.011** -2.97 0.003*** 1.18 0.118 Predicted Sign ? ? + + ? + ? Included Included 10398 0.045 PwC-ML = 0 Coeff. Robust Estimate t-stat. 0.088 1.58 -0.014† -0.78 -0.010 -6.30 -0.003 -0.47 -0.028 -1.27 -0.038 -3.54 0.010 0.42 0.001 0.12 0.403 3.51 0.000 0.04 -0.015 -3.08 0.014 1.44 P-value 0.115 0.436 0.000*** 0.320 0.102 0.000*** 0.336 0.906 0.000*** 0.970 0.001*** 0.074* † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (-0.043) and other countries (-0.014) = 4.79 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). TABLE 8 Panel A Effects of Within-Big 4 Market Concentration: PwC-ML Countries vs. Other Countries Within-Big 4 Market Concentration and Earnings Informativeness Dependent variable = RETURN Variables Intercept EARN Big4Concen EARN*Big4Concen SIZE LEV ROA GROWTH BETA TOT_ACC FIN_DEVEL LNGNI INV_PROT RULE_OF_LAW EARN *SIZE EARN *LEV EARN *ROA EARN *GROWTH EARN *BETA EARN *TOT_ACC EARN *FIN_DEVEL EARN *LNGNI EARN *INV_PROT EARN *RULE_OF_LAW Industry fixed effects Year fixed effects N R-square Predicte d Sign ? + ? ? ? ? ? ? ? ? ? ? ? ? + + + + + ? ? + + Included Included 10764 0.152 PwC-ML = 1 Coeff. Robust Estimate t-stat. 0.022 0.13 1.961 1.58 0.082 0.10 0.030† 0.93 -0.023 0.00 0.047 0.02 0.468 0.09 0.076 0.01 0.074 0.02 0.064 0.08 -0.005 0.01 -0.003 0.02 0.025 0.01 -0.040 0.03 0.031 0.02 -0.152 0.14 2.100 0.30 0.216 0.06 0.032 0.11 0.163 0.36 -0.136 0.14 -0.123 0.18 0.178 0.07 0.005 0.27 P-value 0.900 0.106 0.399 0.974 0.000*** 0.026** 0.000*** 0.000*** 0.000*** 0.402 0.706 0.873 0.001 0.135 0.107 0.116 0.000*** 0.000*** 0.381 0.325 0.315 0.482 0.003*** 0.493 Predicted Sign ? + ? ? ? ? ? ? ? ? ? ? ? ? + + + + + ? ? + + Included Included 5853 0.275 PwC-ML = 0 Coeff. Robust Estimate t-stat. 0.575 2.96 2.656 0.98 -0.263 -3.24 -2.216† -2.49 -0.017 -3.79 0.045 1.70 0.879 7.83 0.029 3.11 0.028 1.19 -0.016 -0.20 0.024 1.18 -0.054 -2.52 -0.024 -2.04 -0.017 -0.63 0.001 0.04 -0.112 -0.53 2.625 4.80 -0.005 -0.06 0.105 0.57 0.344 0.57 -0.016 -0.08 -0.133 -0.50 0.139 1.19 0.060 0.39 P-value 0.003*** 0.163 0.001*** 0.013** 0.000*** 0.090* 0.000*** 0.002*** 0.234 0.844 0.239 0.012** 0.041** 0.530 0.484 0.296 0.000*** 0.472 0.285 0.284 0.939 0.618 0.116 0.348 † F-stat on the difference in coefficient of EARN*Big4Concen between PwC-ML countries (0.030) and other countries (-2.216) = 6.92 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the largest market share based on client sales (or assets). TABLE 8 Effects of Within-Big 4 Market Concentration: PwC-ML Countries vs. Other Countries Panel B Within-Big 4 Market Concentration and Cost of Equity Dependent variable = IndEP PwC-ML = 1 PwC-ML = 0 Predicted Coeff. Robust Predicted Coeff. Robust Variables P-value Sign Estimate t-stat. Sign Estimate t-stat. Intercept ? -0.181 -3.29 0.001*** ? -0.597 -8.23 Big4Concen ? -0.012† -0.33 0.745 ? 0.099† 2.28 SIZE 0.001 1.41 0.074* 0.002 0.71 LEV + -0.008 -0.56 0.287 + 0.021 1.03 ROA 0.007 0.26 0.394 -0.063 -1.28 GROWTH 0.004 1.32 0.092* 0.014 3.34 BETA + -0.029 -7.77 0.000*** + -0.022 -2.90 TOT_ACC + 0.055 2.43 0.007*** + -0.049 -1.20 FIN_DEVEL ? -0.004 -0.94 0.345 ? -0.037 -6.76 LNGNI ? 0.005 0.84 0.401 ? 0.057 6.21 INV_PROT -0.013 -4.61 0.000*** 0.002 0.69 RULE_OF_LAW 0.046 4.80 0.000*** -0.044 -3.24 Industry fixed effects Included Included Year fixed effects Included Included N 7641 4182 R-square 0.095 0.097 Pvalue 0.000*** 0.023** 0.238 0.151 0.100* 0.001*** 0.002*** 0.115 0.000*** 0.000*** 0.246 0.001*** † F-stat on the difference in coefficient of Big4Concen between PwC-ML countries (-0.012) and other countries (0.099) = 7.30 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). TABLE 8 Panel C Effects of Within-Big 4 Market Concentration: PwC-ML Countries vs. Other Countries Within-Big 4 Market Concentration and Cost of Debt Dependent variable = CostDebt Variables Intercept Big4Concen SIZE LEV ROA TANGIBLE TOT_ACC FIN_DEVEL LRATE LNGNI INV_PROT RULE_OF_LAW Industry fixed effects Year fixed effects N R-square Predicted Sign ? ? + + ? + ? Included Included 21413 0.068 PwC-ML = 1 Coeff. Robust Estimate t-stat. 0.010 0.20 0.041† 0.71 -0.009 -8.79 -0.009 -1.79 -0.059 -4.46 -0.032 -4.42 -0.021 -1.43 -0.016 -5.11 0.489 9.60 0.011 1.91 -0.007 -1.76 0.016 1.88 P-value 0.845 0.478 0.000*** 0.073* 0.000*** 0.000*** 0.153 0.000*** 0.000*** 0.056* 0.078* 0.061* Predicted Sign ? ? + + ? + ? Included Included 10398 0.046 PwC-ML = 0 Coeff. Robust Estimate t-stat. 0.001 0.05 0.101† 1.96 -0.009 -6.28 -0.003 -0.38 -0.028 -1.27 -0.041 -3.77 0.010 0.41 -0.001 -0.21 0.410 3.52 0.006 0.85 -0.011 -2.01 0.011 1.11 P-value 0.975 0.050** 0.000*** 0.354 0.102 0.000*** 0.340 0.831 0.000*** 0.394 0.022** 0.134 † F-stat on the difference in coefficient of Big4Concen between PwC-ML countries (0.041) and other countries (0.101) = 4.56 (p-value=0.000). Industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. See Appendix for variable definitions. PwC-ML countries are countries where PwC has the largest market share based on client sales (or assets). TABLE 9 Results Based on Factual Audit Quality Measures Panel A Effects of Big 4 Market Share: PwC-ML Countries vs. Other Countries Dependent variable =TOT_ACC PwC-ML = 1 Robust Predicted Coeff. Variables Sign Estimate t-stat. P-value Big4Share ? -0.03 -3.81 0.000*** R-square 0.287 PwC-ML = 0 Robust Predicted Coeff. Sign Estimate t-stat. P-value ? 0.020 2.00 0.046** 0.364 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (-0.030) and other countries (0.020) =5.67 (p-value=0.000). Dependent variable =AB_ACC PwC-ML = 1 PwC-ML = 0 Robust Robust PPredicted Coeff. Predicted Coeff. Variables Sign Estimate t-stat. P-value Sign Estimate t-stat. value Big4Share ? -0.072 -2.97 0.007*** ? -0.046 -1.00 0.327 R-square 0.108 0.109 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (-0.072) and other countries (0.046) =3.87 (p-value=0.049). Dependent variable = Prob.(PROFIT) PwC-ML = 1 PwC-ML = 0 Predicted Coeff. Predicted Coeff. PChiChiVariables Sign Estimate Square P-value Sign Estimate Square value Big4Share ? -0.794 10.470 0.001*** ? -0.251 0.705 0.401 Pseudo Rsquare 0.381 0.322 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (-0.794) and other countries (0.251) =5.66 (p-value=0.059). The PwC-ML =1 sub-sample contains 11,359 observations, and the PwC-ML =0 sub-sample contains 5,187 observations. We measure the variables following Francis et al. (2012)’s definitions, which are described as follows: (1) TOT_ACC is net income before extraordinary items less cash flows from operations, scaled by lagged total assets; (2) AB_ACC is the residual from a regression predicting non-discretionary accruals as calculated in Dechow et al (1995) and Kothari et al (2005). The model specification used is: NDAt = α + β1 (1 / ASSETSit-1) + β2 (∆SALESit - ∆ARit) + β3 (PPEit) + β4 (ROAit) + Year Fixed Effects + Industry Fixed Effects + Country Fixed Effects + ε where NDA is non-discretionary accruals (net income less cash flows, scaled by lagged assets), ASSETS is a firm’s total assets, SALES is sales, AR is accounts receivable, PPE is gross property, plant and equipment and ROA is return on assets. AB_ACC is the residual from this regression performed over separate industry-year groupings where industries are based on 2-digit SIC codes; and (3) PROFIT is 1 when a firm’s net income for the year is above zero, and 0 otherwise. See Appendix for definitions of other variables. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). Control variables and industry and year fixed effects are not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations. Panel B Effects of Within-Big 4 Market Concentration: PwC-ML Countries vs. Other Countries Dependent variable = TOT_ACC PwC-ML = 1 PwC-ML = 0 Robust PRobust PPredicted Coeff. Predicted Coeff. Variables Sign Estimate t-stat. value Sign Estimate t-stat. value Big4Concen ? 0.037 1.11 0.272 ? 0.043 1.69 0.092* R-square 0.260 0.332 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (0.037) and other countries (0.043) =3.22 (p-value=0.073). Dependent variable = AB_ACC PwC-ML = 1 PwC-ML = 0 Robust PRobust Predicted Coeff. Predicted Coeff. Variables Sign Estimate t-stat. value Sign Estimate t-stat. P-value Big4Concen ? 0.062 0.82 0.418 ? 0.225 2.52 0.012** R-square 0.118 0.102 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (0.062) and other countries (0.225) =12.87 (p-value=0.000). Dependent variable = Prob.(PROFIT) PwC-ML = 1 PwC-ML = 0 Predicted Coeff. PPredicted Coeff. ChiChiVariables Sign Estimate Square value Sign Estimate Square P-value Big4Concen ? 1.599 2.461 0.117 ? 3.394 7.139 0.008*** Pseudo Rsquare 0.384 0.330 † F-stat on the difference in coefficient of Big4Share between PwC-ML countries (1.599) and other countries (3.394) =9.69 (p-value=0.000). The PwC-ML =1 sub-sample contains 11,359 observations, and the PwC-ML =0 sub-sample contains 5,187 observations. We measure the variables following Francis et al. (2012)’s definitions, which are described as follows: (1) TOT_ACC is net income before extraordinary items less cash flows from operations, scaled by lagged total assets; (2) AB_ACC is the residual from a regression predicting non-discretionary accruals as calculated in Dechow et al (1995) and Kothari et al (2005). The model specification used is: NDAt = α + β1 (1 / ASSETSit-1) + β2 (∆SALESit - ∆ARit) + β3 (PPEit) + β4 (ROAit) + Year Fixed Effects + Industry Fixed Effects + Country Fixed Effects + ε where NDA is non-discretionary accruals (net income less cash flows, scaled by lagged assets), ASSETS is a firm’s total assets, SALES is sales, AR is accounts receivable, PPE is gross property, plant and equipment and ROA is return on assets. AB_ACC is the residual from this regression performed over separate industry-year groupings where industries are based on 2-digit SIC codes; and (3) PROFIT is 1 when a firm’s net income for the year is above zero, and 0 otherwise. See Appendix for definitions of other variables. PwC-ML countries are countries where PwC has the single largest market share based on client sales (or assets). Coefficients on control variables, year and industry fixed effects are included but not reported for brevity. T-statistics are calculated based on robust standard errors clustered at the firm level. Reported p-values are based on one-tailed tests where the sign is predicted and two-tailed tests otherwise. *, ** and *** represent significance at level of 10%, 5% and 1% respectively. N denotes the number of firm-year observations.