Audit Fee Research on Issues Related to Ethics Free

There is now an extensive resource of published research on audit fees, some of which explores potential ethical risks regarding audit fees. This paper reviews research on four issues related to audit fees—fee level, dependence, non-audit fees, and firms that provide non-audit services. This paper is based on a report requested by the International Ethics Standards Board for Accountants (IESBA; hereafter, Ethics Board), which is interested in these issues. In most cases, research has not shown these ethical risks to be significant practical problems. The contribution of the paper is that it addresses issues that are of specific concern to standard-setting bodies, and links the issues to requirements in the Code of Ethics.

The potential risks include auditors reducing fees to attract audit engagements, auditors being dependent on audit fees, auditors providing non-audit services to their audit clients, audit firms that provide extensive non-audit services, and the high level of market concentration in the market for audit services. Although most research studies do not find substantial concerns, there are a few recent studies that do. There is consistent evidence that audit fees for new engagements are lower and that non-audit services affect independence in appearance. There are occasional studies reporting evidence that non-audit services provided by an auditor are associated with a loss of independence indicated by lower audit quality. There has been recent concern about growth in non-audit services to non-audit clients and there is some evidence that audit quality is lower in firms that have more extensive non-audit businesses.

The paper extends beyond previous literature reviews because it links the issues examined to the Code of Ethics and to the concerns of the Ethics Board. Previous studies include Gramling, Jenkins, and Taylor (2010), who examine research about the United States up to mid-2009; DeFond and Zhang (2014), who review archival research related to audit quality published from 1996 to 2013 and include a discussion of auditor independence, among other issues; Sharma (2014, 68), who reviews an illustrative selection of research related to non-audit services (NAS); Tepalagul and Lin (2015), who examine research about auditor independence in a selection of nine research journals; and Simnett, Carson, and Vanstraelen (2016), who review the broad topic of archival research in auditing published in eight journals from 1995 to 2014. These prior reviews are not based upon the Code of Ethics or the other issues that motivated the Ethics Board to examine research in this area. They each have a focus on another issue, either broader or narrower than the topic of my review, and most of them examine a restricted set of research articles. Nevertheless the previous research articles all make a useful contribution to understanding the research in the area. The conclusions of these earlier reviews are compared to the conclusion of this study later in the paper.

The motivation for the paper is the concerns held by the Ethics Board about audit fee issues arising from the IESBA Code of Ethics (International Federation of Accountants 2015) together with concerns raised by stakeholder groups. The relevant issues in the Code of Ethics are self-interest (which includes dependence); intimidation, which includes pressure to reduce work and fees; a fee too low to perform an engagement with professional competence and due care; and overdue fees.¹ I also considered the relationship of audit fees to the Fundamental Principles of the Code of Ethics: integrity, objectivity, professional competence and due care, confidentiality, and professional behavior. The Ethics Board was influenced by the Public Interest Oversight Board (PIOB), the International Organization of Securities Commissions (IOSCO), and national regulators. The PIOB provides independent oversight of the international standard-setting bodies such as the Ethics Board that are supported by the International Federation of Accountants. At the April 2015 meeting of the Ethics Board, the PIOB had urged the Board to revisit auditor independence, including fee-related issues.² The IOSCO had communicated concerns to the Ethics Board about safeguards for fee dependency, about threats to independence from non-audit services provided to audit clients, and about low fees or fee pressures leading to negative impacts on audit quality.³ In addition the Ethics Board reported that “some investors and regulators had expressed concerns relating to firms' business models, specifically the continuing significant growth in the advisory services arms of the larger firms and the related implications for audit quality and independence.”⁴ One of the regulators that has expressed concerns is the Public Company Accounting Oversight Board (PCAOB) in the U.S. (according to Lisic, Myers, Pawlewicz, and Seidel [2016]).

The main issues that I examined are:

• Issue 1: Level of audit fees for individual audit engagements;

• Issue 2: Relative size of fees to the partner, office, or firm and the extent to which partner remuneration is dependent upon fees from a particular client;

• Issue 3: Ratio of NAS fees to audit fees; and

• Issue 4: Provision of audit services by a firm that also has a significant non-audit services business.

These four issues are all relevant current issues that are of concern to the Ethics Board and that relate to issues in the Code of Ethics.

In summary, the findings show that there are numerous studies on some issues, but few on others. The level of audit fees appears to have increased substantially until about 2007, but less so since. Non-audit fees have reduced since the early 21st century. The concentration of the market for audit services is high. While concentration has in the past been considered not to be a problem, some recent studies suggest that it is now becoming a concern, at least in the U.K. and Australia. Research about the size of audit fees, and possible dependence of an auditor upon a client, shows that there are some cases where higher fees are associated with worse audits. There are also occasional studies suggesting that higher non-audit fees reduce independence (indicated by audit quality) in some conditions. There is considerable evidence that higher non-audit fees reduce independence in appearance. There is not much research about audit services provided by firms that have a substantial non-audit services business, but what there is suggests some indication that high levels of NAS business are associated with lower audit quality.

I conducted a search in May 2016 using the Accounting and Tax database (part of ProQuest). I added further papers including official and commercial reports and relevant recent unpublished papers. The initial search found 187 papers, but these papers were not all relevant enough to be included in the discussion. The remainder were papers on closely related topics but not sufficiently relevant to be included, or were reports of fee levels or fee issues without including research, or were comments on issues without research included. Of that initial search, 78 studies were relevant enough to be included in the paper. Of the remaining 78 papers, 32 relate to the level of fees, 18 to relative fees and dependence, 23 to the ratio of non-audit to audit fees, and 5 to firms with a non-audit business. Table 1 summarizes the papers and the issues.

The paper includes a discussion of the primary, generally accepted findings of the research, as well as exceptional problem areas found in some studies. However, I have highlighted indications of problems where these are evident, as these are the results that are most likely to be of concern. Table 2 supports Table 1 by listing specific papers that raised issues of potential concern for the issues discussed.

The review covers the issues that the Ethics Board considers to be of concern and that relate to audit fees: the review is restricted to studies of audit fees. To ensure that it is current, and at the request of the Board, it covers a limited period of time (2006–2016); it is intended to be brief. The need for an immediate report and the limited number of studies in some areas precluded a quantitative meta-analysis of the underlying studies. I have avoided making conclusions based on counting the number of papers because that approach can be misleading—in many cases some research papers need to be given more weight than others because they use larger samples or are of higher quality.

The research results reported are in most cases published in leading academic journals and can be regarded as rigorous. An exception is the unpublished studies referred to at some points, especially under Issue 4, which is a fast-moving current issue for which the only studies available are as yet unpublished. The research articles use multivariate statistical techniques to account for differences among the audits being examined. The papers have been through a rigorous process of peer review and are subject to extensive tests of issues such as endogeneity, and sensitivity tests for issues such as the measures used for the variables. For example, where there are underlying issues such as greater risk or complexity that could affect results, the researchers have controlled for those effects. Nevertheless, the issue of whether the studies show causation or simply correlation can never be entirely eliminated. There are also numerous issues for which the outcome is simply not known. The significant results reported represent warning signals about problems that could be of concern.

The ethical issues concerned with audit fees are whether overall fees are adequate to perform a proper audit, whether audit firms are providing a competitive market, and whether there is “lowballing” of new audit engagements. The Code of Ethics notes that a self-interest threat to the Fundamental Principle of professional competence and due care is created if the audit fee is so low that it may be difficult to perform the engagement in accordance with applicable technical and professional standards.⁵ The adverse impact of low audit fees was a particular concern raised by the PIOB and IOSCO. Publications on the issue of the level of audit fees for individual audit engagements include the overall level of audit fees or audit fee increases in a country. I also include in this category studies of the market share of audit firms and the “lowballing” of audit fees.

The first issue examined is the level of fees, including whether fees for individual engagements are too high or too low. The PIOB and IOSCO had both advised the Ethics Board that they were concerned about the possible adverse impact of low audit fees on audit quality. The Consultative Advisory Group (CAG) of the Ethics Board had raised the issue of downward pressure on audit fees created by new requirements for mandatory audit firm rotation and tendering in some locations.⁶

Some papers provide relatively raw data that do not take into account changes in the underlying population of companies or changes in the extent of audit work required. However, there are research studies of the level of fees that take these changes into account and that examine fee changes after significant events that affected auditing in some countries.

In the U.S., reports on the overall level of audit fees show that they increased substantially in 2004 and have continued to increase slowly, or decrease, after that (Whalen, Hannen, and Lussier 2015a). Research studies confirm an increase in audit fees in the U.S. subsequent to SOX (e.g., Griffin and Lont 2007; Ghosh and Pawlewicz 2009). There are also recent studies reporting fee pressure, including fee cuts for many clients in the U.S. in the period 2006–2009 (Ettredge, Fuerherm, and Li 2014; Krishnan and Zhang 2014). During the recession of 2007–2009, Ettredge et al. (2014) found that fee pressure in 2008 (but not 2007 or 2009) was associated with lower audit quality, as measured by a greater number of restatements. A study of bank audits also found that reduced audit fees were associated with lower quality, but only when the auditor was non-Big 4, the client was exempt from an internal control audit, and the fee cut was more than 25 percent (Krishnan and Zhang 2014). There is therefore some evidence that pressure on audit fees is associated with lower audit quality, although it does not have this affect in every period or setting.

In the U.K., trends appear to be similar to the U.S. In annual reviews of fees in 2005, 2006 and 2007, audit fees were reported to have “surged” (Neveling 2006), had “double digit” increases (Accountancy Age 2006a), and had a “huge” increase (Accountancy Age 2006b). In Canada, audit fees have increased very little since 2007—by a steady amount of approximately 3.2 percent per year over the period (Whalen, Hannen, and Lussier 2015b). Research studies on New Zealand data by Griffin, Lont, and Sun (2009) and Higgins, Lont, and Scott (2016) report an increase in audit fees, which they attribute to the introduction of IFRS.

There is at least one study warning that audit fees could be too low in a particular market. Behn, Lee, and Jin (2009) found evidence in Korea of the discounting of initial fee engagements, and that audit fees per hour were decreasing. They reported “the regulatory and legislative authorities became concerned that price-based competition could cause deterioration in audit quality” and that “public agencies and the popular press suspected that increasing competition for audit services could lead to underpricing” (Behn et al. 2009, 132–133). Their data were from 1999–2004, however, and more recent studies of Korean data have not reported the same issue (Kwon, Lim, and Simnett 2014; Jung, Kim, and Chung 2016).

To summarize, there is an ethical issue whereby low fees could impair professional competency and due care. The research findings show that audit fees increased in the early part of the 21st century. There is occasional evidence that shows associations between low fees and reduced audit quality, but this issue has yet to be investigated very thoroughly.

An issue of potential concern is monopolistic pricing, as in most countries the market for audit services is highly concentrated, with only a few audit firms auditing most clients. Competition is not referred to in the Code of Ethics, although accountants are required to comply with relevant laws and not to bring the profession into disrepute.⁷ This issue was also not specifically included in the terms of reference for this investigation, although it seems relevant to the topic of “the level of fees for individual engagements.”

Concentration of the market for audit services has been controversial since the 1970s and 1980s (Simunic 1980). It is argued that reduced competition lowers the auditor's incentives to provide high quality (DeFond and Zhang 2014, 311). Although this issue has been studied, evidence that there is a lack of competition has not been found (according to Simunic [2014]). That is the accepted view, although there are some studies showing evidence of lack of competition in the U.K. and Australia (Oxera 2006; Carson, Simnett, Soo, and Wright 2012; Ding and Jia 2012).

Papers on the market share of audit firms commonly find a premium for the Big N firms, although this is mainly a U.S. phenomenon according to Hay and Knechel (2017). It is widely considered, based on earlier research, that the large market share held by the Big 4 firms in most countries is not inconsistent with competition, and that higher fees represent a premium for higher quality.

Nevertheless, some recent papers show that there are concerns about competition. Notable official studies in the period were by Oxera (2006) for the U.K. Department of Trade and Industry and the Financial Reporting Council and by the Government Accountability Office (GAO 2008) in the U.S. Oxera (2006, i) concluded that in the U.K. market, “higher concentration has led to higher audit fees.” The GAO (2008) found that “Continued concentration in [the] audit market for large public companies does not call for immediate action.” However, the report also acknowledges that the available data do not allow for a conclusion that the auditing market is competitive (GAO 2008).

There are other studies that find evidence of reduced competition. Ding and Jia (2012) report a significant increase in fees for Big N firms in the U.K. after the Price Waterhouse-Coopers & Lybrand merger in 1998 and conclude that there was increased market power. Carson et al. (2012), in a study of Australian data, found an increase in the Big N firm premium over the years 1996 to 2007, during which the firms decreased from the Big 6 to the Big 4, although this increase did not apply to the largest client firms. A later paper comments that “the decrease in the number of the largest audit firms from the Big 6 to the Big 4 has likely reduced the level of competition in the audit market for public companies as reflected by increases in audit fee premiums, but that this impact is not consistent across all client segments” (Carson, Botica Redmayne, and Liao 2014, 302).

There are many other settings for which there are no recent published studies. In general, the evidence does not indicate the lack of competition is an issue, but there are a few studies suggesting that it might be in some markets.

Whether auditors charge a fee that is too low in order to attract clients is an ethical issue. The Code of Ethics states that “the fact that one professional accountant in public practice may charge a fee lower than another is not in itself unethical” but that “a self-interest threat to professional competence and due care is created if the fee is so low that it may be difficult to perform the engagement in accordance with applicable technical and professional standards for that price.”⁸

Lowballing is a term used for auditors charging a fee below cost in the early years of an audit engagement in order to win the client. Since the auditor's cost is not usually observable, researchers often examine fee cutting, i.e., charging a lower fee after a switch of auditor. Fee cutting is a proxy measure for lowballing, although fee cutting is not itself an issue to the same extent as lowballing. Related issues include rotation of auditors and long tenure (when clients do not change auditors, generating a risk of familiarity).

There is evidence that fee cutting occurs. Audit fees charged by firms for new engagements are lower than for comparable engagements where there has not been a change of auditor. A recent study by Griffin and Lont (2011) found lower fees both before and after switches in the U.S. H. Huang, Raghunandan, T. Huang, and Chiou (2015) found evidence of lower fees and lower quality after switches in China, and Stanley, Brandon, and McMillan (2015) found evidence of fee discounting and lower quality after switches in the U.S. Huang et al. (2015) found that switches were followed by lower audit fees and associated sanctions for problem audits and greater earnings management (controlling for variables related to the client and audit firm). Stanley et al. (2015) reported lower fees after a switch, accompanied by a greater tendency for clients to use discretionary accruals to meet analyst forecasts. A recent unpublished paper (Kácer and Wilson 2016) on U.K. data reports a very large study of 792,905 company-year observations. They report initial audit fee cutting.⁹ There is consistent evidence overall of fee cutting, i.e., that fees are lower for new engagements. However, there is only very limited evidence that this is associated with reduced audit quality.

On the other hand, there is contrary evidence (but not from audit fee studies) showing higher audit quality after a switch (Lennox, Wu, and Zhang 2014; Kim, H. Lee, and J. Lee 2015). Lennox et al. (2014) found evidence that first-year audits were of higher quality in China, and Kim et al. (2015) also found evidence of higher quality at first, after auditor switches in Korea.

A recent study by Bell, Causholli, and Knechel (2015) examined both fees and audit firms' records of effort (measured by audit hours). They found that a change in auditor was followed by low fees at first, but not by low effort. They also found that audit quality declines eventually with long tenure, but that this effect applies only to private clients, not to listed companies where there is more regulation and greater risk (Bell et al. 2015, 465). The evidence in Bell et al. (2015) suggests that there is a greater concern about long tenure when there are fewer other controls, such as in the market for private company audits. Among earlier studies, J. Myers, L. Myers, and Omer (2003) found longer tenure was associated with higher quality. There are a number of other studies on the auditor rotation issue that do not examine audit fees, and so are not captured here. Review articles summarizing these studies include Casterella and Johnston (2013), Ewelt-Knauer, Gold, and Pott (2013), and Jenkins and Vermeer (2013).

There are two sets of concerns. First, that auditors might lowball, so that the quality of initial engagements might be low; and second, that auditors with long tenure might also reduce quality. Thus there are concerns about professional competency and due care. There is consistent evidence that fees are lower after a change of auditors, and mixed evidence about whether quality is lower. This issue has also been recognized by the Ethics Board through its paper on downward fee pressure (IESBA 2016).

The issue concerning relative size of fees and dependence is whether high audit fees or dependence by an auditor on one client for a high proportion of their revenue constitutes a self-interest threat. This issue of fee dependency is referred to in the Code of Ethics as a self-interest or intimidation threat, and could occur when total fees from an audit client represent either a large proportion of the total fees of the audit firm,¹⁰ a large proportion of the revenue from an individual partner's clients, or the revenue of an individual office of a firm.¹¹ The IOSCO expressed a particular concern that “the safeguards for fee dependency do not appear to be commensurate with the potential threats to independence that the Code seeks to prevent.”¹²

The research studies tend to show that this threat does not appear to affect auditors, although there are a few studies that show high client importance or dependence is associated with reduced independence as indicated by lower audit quality.¹³ Higher fees and dependence (measured by higher fees) are not usually found to be associated with reduced auditor independence. There is evidence that higher audit quality is associated with more important clients, as measured by fees (Gaver and Paterson 2007; Hunt and Lulseged 2007; Blankley, Hurtt, and MacGregor 2012). In China, Chen, Sun, and Wu (2010) report lower quality where there is fee dependence before 2000, but higher quality after 2000 (measured by modified opinions). Li (2009) found that dependence at the firm level in the U.S. had no effect on going concern opinions before SOX; however, in 2003 dependence did make a difference. Later studies by Feldmann and Read (2010) and Kao, Li, and Zhang (2014) show that this change was only a temporary effect and that there is no effect in the longer term.

Evidence of the impact of fees on lower audit quality is provided by Choi, Kim, and Zang (2010), showing that higher abnormal audit fees in the U.S. are associated with lower quality (measured by discretionary accruals). A study in Taiwan shows that economically important clients at the partner level are associated with lower quality (measured by accruals) for small firms but not for Big N firms (Chi, Douthett, and Lisic 2012). More evidence, in this case from the U.S., is reported by Blay and Geiger (2013), who show that distressed firms are less likely to be issued going concern opinions if future total fees and NAS fees are higher. However, there is a question of the interpretation of higher audit fees. Some studies, such as Choi et al. (2010), interpret higher fees as abnormal profits to the auditor, but there are others that interpret higher fees as indicating greater risk and more audit work (Hribar, Kravet, and Wilson 2014). Doogar, Sivadasan, and Solomon (2015) conclude, by showing that these higher fees continue after a switch of auditor, that higher fees represent audit costs, and that they do not represent an adverse impact of auditor independence (Doogar et al. 2015, 1278).

Partner income is examined by Knechel, Niemi, and Zerni (2013), who show that, for three of the Big 4 firms, individual partner income is affected by gaining clients, losing clients, and audit failures. Their data are from Sweden. They conclude that whether this situation applies varies according to the firm's remuneration policy.¹⁴

Evidence that higher fees affect independence in appearance is shown in Ghosh, Kallapur, and Moon (2009). They show that earnings response coefficients are less—suggesting that analysts find the announcements less convincing—when there are higher audit fees (measured by relative client importance), but not when there are high non-audit fees.¹⁵

Dependence as defined in the Code of Ethics has not been widely examined. There are few studies that examine the 15 percent threshold for fee dependence. More research on what the appropriate threshold should be at the level of firm, office, and partner could be useful. Overall, there is some evidence of concerns about high fees as a measure of dependence, but there are also studies showing high fees are beneficial (in the U.S. in most of the periods examined and in Italy), and others showing that high levels of fees make no difference. The evidence generally suggests that auditor independence is not reduced when there are high relative fees, but there is also some opposing evidence.

Since the scandals of the early 2000s, the issue of high non-audit fees in relation to audit fees has been topical and widely researched. The concern is potential loss of independence. The Fundamental Principles in the Code of Ethics specify that “a professional accountant shall not perform a professional activity or service if a circumstance or relationship biases or unduly influences the accountant's professional judgment with respect to that service.”¹⁶ Sections 290.154 to 290.216 of the Code of Ethics discuss the effects of particular types of service. The Ethics Board was concerned about this issue because the IOSCO had suggested that the Code of Ethics should include guidance about when the quantum of non-audit services might threaten independence.¹⁷

NAS to audit clients declined quickly after 2002, and then continued to decline slowly (e.g., Whalen et al. 2015a). This is not surprising. It is partly due to new regulations in many countries, but also because directors and auditors responded to public concerns about non-audit services (e.g., Abidin, Beattie, and Goodacre 2010; DeFond and Zhang 2014, 312). There has been extensive audit fee research on the impact of high non-audit fees in relation to audit fees. There is no strong evidence that NAS are a problem, but occasionally studies showing apparent loss of independence are appearing.

While some observers expect to see that auditors reduce their audit fee as a “loss-leader,” this effect has not been found. Instead, high non-audit fees are usually associated with high audit fees as well.

Research prior to the period of this review is important so far as this issue is concerned. Frankel, Johnson, and Nelson (2002) reported that there was evidence of auditors losing their independence when there were high non-audit fees in proportion to audit fees, using data from soon after when U.S. companies were first required to disclose audit fees. That study was controversial. It was challenged by further research (Ashbaugh, LaFond, and Mayhew 2003; Chung and Kallapur 2003; Larcker and Richardson 2004). It became generally accepted that non-audit fees did not impair independence (Sharma 2014, 68). As a result, there is less motivation for researchers to examine this issue again.

More recent evidence suggests there may be problems in some circumstances. In the research published during the period under review, Bell et al. (2015, 465) find that NAS are associated with higher audit quality and higher fees for public companies, but lower quality and lower fees for private companies. This is likely to be due to “market and regulatory mechanisms” (Bell et al. 2015, 465) that keep watch on public clients.¹⁸ There are recent studies that find results suggesting loss of independence. Some of these studies use U.S. data from before SOX (Srinidhi and Gul 2007; Gul, Jaggi, and Krishnan 2007; Krishnan, Su, and Zhang 2011; Omer, Bedard, and Falsetta 2006; Dickins 2008; Causholli, Chambers, and Payne 2014; Bell et al. 2015). However, there are others that find evidence of loss of independence using more recent U.S. data (Rice and Weber 2012; Ye 2012), or data from Australia (Ye, Carson, and Simnett 2011), or New Zealand (Wang and Hay 2013). More details of these studies are provided in Table 2. In contrast, in Italy, Ianniello (2012) found some positive associations between NAS and qualified opinions, suggesting higher non-audit services led to higher audit quality.

The research evidence also shows that NAS have an effect on independence in appearance. Numerous studies show that higher non-audit fees impact share price, or related measures such as the earnings response coefficient (ERC). There are eight studies, all using U.S. data, in the period 2006–2016. At least one (Ghosh et al. 2009) uses data subsequent to SOX. Higgs and Skantz (2006) use an event study methodology with unexpected returns regressed on unexpected earnings. They add measures for unexpected audit and non-audit fees with interaction terms to test whether ERCs are associated with audit and non-audit fee residuals (Higgs and Skantz 2006, 7).

Sharma (2014) sums up this evidence by stating, “investors see a problem even if audit quality studies don't.”¹⁹ DeFond and Zhang (2014, 309) propose that these results may mean that investors have ways of detecting the effects of reduced independence that are not captured by research.

The issue of concern is objectivity, including independence of mind and independence in appearance. There is mixed evidence about the effect of NAS on auditor independence of mind, but consistent evidence from numerous studies that NAS reduce the appearance of independence.²⁰ There is some recent evidence that there is reduced independence of mind in some circumstances.

The Ethics Board had noted that some investors and regulators have expressed concerns about the “continuing significant growth in the advisory services arms of the larger firms.”²¹ Firms providing NAS to non-audit clients is an issue of increasing concern (U.S. Department of the Treasury 2008; Hermanson 2009; Dey, Robin, and Tessoni 2012; Chi et al. 2012; Lisic et al. 2016; Beardsley, Lassila, and Omer 2016). The concern is that if audit firms become predominantly focused on consulting services, then they will allocate resources and talent to their non-audit businesses and their culture will place less focus on auditing.

NAS provided to non-audit clients by U.S. audit firms have grown considerably in recent years and now represent substantial proportions of total revenue (Dey et al. 2012; Lisic et al. 2016). There is not very much research about how high levels of NAS to non-audit clients affect audit quality. There are two unpublished studies that show some evidence that audits are of lower quality when firms provide more NAS to non-audit clients (Lisic et al. 2016; Beardsley et al. 2016) and that independence in appearance is reduced (Lisic et al. 2016). Beardsley et al. (2016) find that when offices of a firm are experiencing fee pressure, they are likely to increase non-audit services; when they do increase non-audit services, there is an increase in misstatements. Fee pressure is measured by a drop in revenue of one or more deciles for a particular office of a firm, relative to the other offices. Misstatements are measured by subsequent restatements. Lisic et al. (2016, 28) report a higher likelihood of misstatement among clients that have a lower litigation risk when their auditors generate more revenue from consulting. A misstatement is measured by subsequent restatements, litigation risk is measured using two models from previous studies, and revenues from consulting are measured by non-audit services fees divided by audit fees for the whole firm. Independence in appearance is measured by the earnings response coefficient.

The issue of firms that have non-audit services businesses has raised concerns about professional competence and due care. There is some evidence that high levels of NAS provided to non-audit clients are detrimental to audit quality, but this evidence is preliminary as yet.

Other recent literature reviews that cover research related to the issues examined here include Gramling et al. (2010), DeFond and Zhang (2014), Sharma (2014), Tepalagul and Lin (2015), and Simnett et al. (2016). Their conclusions are consistent with the comments in this paper, that while ethical issues are not strongly evident, there are nevertheless some concerns. Specific excerpts include:

• Gramling et al. (2010, 547): “[research] does not provide strong evidence that the prohibitions and requirements embedded in the current independence rules for public company auditors are strongly correlated with independence in fact. However, some prohibitions and requirements do appear to improve perceived independence—at least in the pre-SOX era.”

• DeFond and Zhang (2014, 312): “evidence is persuasive that audit committee and Section 404 provisions, but not banning NAS, improves audit quality … evidence is limited on whether low-balling or opinion shopping improves audit quality.”

• Sharma (2014, 80): “although this paradigm often concludes that NAS do not threaten auditor independence, the literature remains unclear.”

• Tepalagul and Lin (2015, 112): “there is limited evidence that auditor independence is compromised in the presence of client importance and NAS.”

• Simnett et al. (2016, 11): “further examination of the potential threat to auditor independence, both actual and in appearance, arising from the provision of NAS would be beneficial to justify the current restrictions on the provision of NAS.”

This report is an overview and concentrates on research that became available in a particular period. Studies before 2006 can be important, and research within the period is influenced by them. Conflicting results between papers are a further complication. The results also vary according to setting or period.

Publication bias can also be important—authors, reviewers for journals, and editors are much more interested in papers that have significant or surprising results, and there may be studies that are never seen because they are abandoned by the authors or rejected by the journals for not providing significant results (Stanley, Doucouliagos, and Jarrell 2008; Hay and Knechel 2017). There are techniques for taking into account differences in setting and period, that weight studies according to size, control for publication bias, and that can be used to resolve conflicting results. An example is meta-regression (Hay, Knechel, and Wong 2006; Hay 2013). This technique is time consuming and complex, but could be considered for future work to assess research on some of the questions arising, especially after more research studies become available.

The state of research shows that there are studies examining important issues, but in most cases the research is not yet sufficient to allow firm conclusions to be drawn. Areas for future research could include:

1. Dependence: It would be useful to examine the extent of fee dependence at the firm, office, and partner level, and the impact of high dependence. The Code of Ethics says that when audit fees from a single public interest audit client are more than 15 percent of the firm's total revenue, then special procedures are required (¶290.219). Questions to be investigated could include the following: Why 15 percent? Have different thresholds been tested? What should the threshold be for an individual office or partner? Research could examine different thresholds for offices and partners and whether there is an association with auditor independence.

2. Audit fees, audit quality, and inspection reports: Is there an association between fee levels for an auditor and the outcome of inspection reports by regulators such as the PCAOB? Research on whether low fees are associated with inadequate audit quality (or whether high fees are associated with reduced independence) has used indirect proxy measures, such as earnings management, for quality and independence.

3. NAS and the appearance of independence around the world: Research on the impact of auditors providing NAS on the appearance of independence has nearly always used evidence from the U.S. It would be useful to conduct research in different settings that have differing legal environments and less-developed financial markets.

4. Firm data: One recent study (Bell et al. 2015) was able to obtain a much deeper understanding because the authors had access to internal data from a major firm, including the firm's assessment of audit quality, and records of time and cost, as well as fees. They examined the issues of non-audit services, lowballing, and long tenure. They were able to assess whether more or less effort was made. The Ethics Board may be able to prevail on firms to allow researchers access to confidential data like these, subject to controls such as anonymity.

5. Meta-analysis: Given the conflicting results in several areas, it would be useful, when sufficient studies are available, to conduct a meta-analysis to take stock of the overall results of research such as the effect of non-audit services. This might clarify some of the results that are discussed in more general terms in the paper. Areas where that might be useful include the lowballing issue and the issues of independence of mind and appearance when there are non-audit services.

There are risks to auditor independence that are supported by the research evidence, risks that are not supported, and risks where evidence is mixed. There is no evidence of auditors using the audit as a loss-leader to obtain more lucrative consulting work. There are few signs of audit fees being too low to be able to conduct an adequate audit. Nevertheless, there is evidence of some issues of concern, including non-audit services associated with indications of reduced independence, and non-audit services leading to reduced independence in appearance. There is some concern about the audit services provided by firms that have substantial non-audit service businesses. In general, audit fee research does not convey a message that there are widespread ethical problems, but there are some risk areas, and there are issues that should be further investigated. The Ethics Board is continuing to investigate the issue, taking into account the findings of the study, and proceeding to conduct further fact finding through stakeholder outreach.²²