Auditing the Auditors: An International Analysis of the Effectiveness of National Inspection. Regimes on Audit Quality

Transcription

1 Auditing the Auditors: An International Analysis of the Effectiveness of National Inspection Regimes on Audit Quality Elizabeth Carson Professor of Accounting University of New South Wales Roger Simnett Scientia Professor of Accounting University of New South Wales Ann Vanstraelen Professor of Accounting and Assurance Services Maastricht University 3 September 2013 Preliminary draft: Please do not quote without permission of the authors. We thank Anna Huggins, Ashna Prasad, Shirley Tsau and Lei Zou for research assistance, and appreciate the comments of participants at workshops at the University of Auckland, the University of New South Wales, the International Symposium on Audit Research and the Accounting and Finance Association of Australia and New Zealand Conference. We also acknowledge the financial support of the Australian Research Council.

2 2

3 Auditing the Auditors: An International Analysis of the Effectiveness of National Inspection Regimes on Audit Quality ABSTRACT: We undertake an examination of the effectiveness of the different forms of public oversight and independent inspection regimes that have been put in place by the various national public oversight bodies of the auditing profession. Using a large sample of companies from 33 countries over the period , we find that audit quality is higher in countries where independent inspections are in place and are member of IFIAR. In order to control for potential endogeneity concerns, we then limit our examination to countries that introduced independent inspection programs during our sample period, and we find that audit quality is higher in the period after inspections are introduced compared to the pre-inspection period. Our results, however, do not support an association between the different ways of organizing public oversight and audit quality. This provides evidence that the existence of public oversight and independent inspections has an impact on audit quality, but not the characteristics of the inspection programs. Overall, our study contributes to the emerging literature on public oversight of the auditing profession. Keywords: public oversight; audit quality; IFIAR; inspections Data Availability: The data are available from sources identified in this study. 3

4 I. INTRODUCTION In the last decade one of the key reform initiatives directed at improving the quality of auditing has been the introduction of public oversight over the audit profession. This is observed at both the national and international levels, and impacts on both the setting of standards for the auditing profession as well as its continuing regulation. For example, over the past decade the setting of auditing standards has been more likely to be undertaken by bodies independent of the audit process (such as the Public Company Accounting Oversight Board (PCAOB) in the U.S., and the Australian Auditing and Assurance Standards Board (AUASB), or independent of the profession (for example, the International Auditing and Assurance Standards Board (IAASB)). The continuing regulation of the auditing profession has also been increasingly removed from the auditing profession to independent national public oversight boards (POBs) (for example, the PCAOB in the U.S. and the Australian Securities and Investments Commission (ASIC) in Australia). The U.S. was one of the front-runners in introducing public oversight of the auditing profession, and the PCAOB model of independent public oversight has formed the basis of many other national POBs (Franzel 2012). The PCAOB has four main responsibilities under the Sarbanes-Oxley (2002) Act: (1) to register public accounting firms; (2) to establish auditing and other professional standards; (3) to conduct and report on regular inspections of registered public accounting firms; and (4) to conduct investigations and disciplinary proceedings of the auditing profession where rules or standards may have been violated. While other countries have followed suit, they have made different choices with regard to the organization of public oversight. For example, most countries outside the U.S. have a policy of adopting or converging with standards 4

5 developed by the IAASB (Simnett and Smith 2005), and the setting of auditing standards is separated from the body undertaking the inspection program. Also, while inspection programs have become the key plank of the strategy and work effort of all national POBs, divergent approaches to these inspection programs have developed. For example, some countries have elected to have direct inspections by the independent POB while other countries have chosen for inspections to be undertaken by professional bodies in the country under the supervision of the POB. Further, the frequency of inspections of audit firms varies across countries and some countries have a longer history of public oversight of the auditing profession compared to others. Internationally, a growing number of independent national audit regulators have decided to join the international organization, the International Forum of Independent Audit Regulators (IFIAR). This organization was founded in September 2006 and aims to promote audit quality, and effective independent audit regulatory activity. IFIAR has a specific focus on inspections of audit firms, and co-operation regarding the inspection of international audit networks (IFIAR 2012). The rationale behind the auditing profession s moves to POBs is that regulators and other market participants expect that independent public oversight is a more effective way to ensure audit quality than self-regulation by the audit profession. Critics are not convinced, however, and question the adequacy of the expertise, knowledge and skills of inspectors of independent POBs (e.g,. Glover et al. 2009). This reflects the often raised concern in relation to public oversight of the auditing profession of the trade-off between expertise and independence (e.g., DeFond 2010) and the effectiveness of such oversight. Given the significant amount of resources that is spent on public oversight, its effectiveness is an important empirical question. There is an emerging literature in the U.S. addressing this question (e.g. Lennox and Pittman 2010; DeFond and Lennox 2011; 5

6 Carcello et al. 2011); however to date there has been no international comparison of the differing national approaches to public oversight, and in particular the inspection programs, and their association, if any, with resulting audit quality. In this paper we examine the effectiveness of the inspection programs, and their different forms, as undertaken by the national POBs. Specifically, we examine whether inspections (between countries with and without inspection programs) affect audit quality, and whether there is a difference in audit quality between the pre- and post-inspection periods (within member country analysis for countries introducing inspections during the study period). We further examine whether IFIAR membership, first-time inspections, inspection experience, type of inspection system (direct inspections versus oversight of the inspection process) and frequency of inspections affect audit quality. We measure audit quality by the level of abnormal accruals of audit firms client companies. Using a large sample of listed companies from 33 countries during the period , we find that companies domiciled in countries where inspection programs are in place have significantly lower abnormal total accruals and this holds for both incomeincreasing and income-decreasing abnormal accruals. Further, we find that the level of abnormal accruals (both income-increasing and income-decreasing) of audit firms clients in countries that introduced inspection programs during our sample period is significantly lower in the post-inspection period compared to the pre-inspection period. This compares with no change in abnormal accruals over for clients in those countries that did not introduce an inspection program. We further find that companies domiciled in IFIAR member countries have significantly lower abnormal accruals, which is to be attributed to lower income-decreasing accruals. We do not find much support for differences in audit quality that could be explained by differences in design of the oversight system within the group of countries with independent inspections of audit firms in place. Overall, our 6

7 findings contribute to the emerging literature on public oversight of the auditing profession. This study is one of the first that compares the benefits, if any, of the differences in the national regulatory systems at an international level. A better understanding of the effectiveness of the different types of oversight systems can facilitate the design of an optimal and potentially harmonized oversight system which would be congruent with the objectives of IFIAR. Further, our insights may also prove useful for regulators across the world in their ongoing discussions and decisions regarding mutual recognition of public oversight across countries. The remainder of this paper is organized as follows. In Section II, we provide the relevant background for the study and formulate the hypotheses that we test. In Section III, we describe the data, descriptive statistics and research design. In Section IV, we present the results of the study and we conclude in Section V. II. BACKGROUND AND HYPOTHESES DEVELOPMENT The Organization of Public Oversight A number of high-profile corporate financial reporting scandals in which auditors were implicated, including the collapse of Enron and the role of its auditor Arthur Andersen, prompted regulators in the U.S. to end self-regulation of the auditing profession. It was replaced by a system of public oversight of the auditing profession, created by the Sarbanes- Oxley Act in Since 2004, all audit firms that issue an audit report for a SEC-reporting company or substantially participate in the audit are required to be registered with the PCAOB. One of the primary duties of the PCAOB is to perform periodic independent inspections of these firms. In reflecting upon the effectiveness of the PCAOB seven years after inception, Acting Chairman Goelzer (2009) noted, in evaluating the Board s work, the question is not how many inspections have been conducted or how many pages of standards 7

8 have been written. It is whether investors are better off as a result. While that is not a question that is likely to be resolved based on anything I might say today, I believe that evidence is accumulating that they are. Goelzer went on to outline the process by which this benefit was achieved: we know from the visibility that we have through the inspections and remediation process, that the large firms have made important changes to their systems of quality control in response to PCAOB inspections findings. These have included such things as changes related to partner evaluation and compensation to place greater emphasis on audit quality and technical skills; changes to management structures to provide greater separation between the audit quality function and audit business operations; creation of national- or regional-level positions or committees to promote and monitor audit quality; and modifications to internal inspection programs. After the creation of the PCAOB in the U.S., other countries around the world have followed the U.S. example of creating independent POBs by either introducing or expanding their use of inspection programs of audit firms. The European Commission (EC) introduced independent oversight over financial reporting and auditing in its revised Eighth Directive of This was further reinforced in the EC Green Paper Audit Policy (2010) addressing the need to strengthen the current role of audit supervision to ensure the full independence of the public oversight systems from the audit profession in all Member States. In the Green Paper, the EC suggests transforming the current European Group of Auditors Oversight Board (EGAOB) 1 into a so-called Lamfalussy Level 3 Committee to strengthen cooperation between national public oversight bodies, ensure an improved and common approach to 1 The European Group of Auditors Oversight Board (EGAOB) was founded in December 2005 and is composed of high-level representatives from the respective public oversight authorities of EU Member States or alternate representatives from national ministries in the event that such a national oversight system is not yet established. The role of the EGAOB is to provide technical input into the preparation of possible measures of the EC in implementing the 8 th EU Directive on statutory audit, and it facilitates the EC s implementation of the requirements of external quality assurance for statutory auditors and audit firms auditing public interest entities. 8

9 inspections of audit firms, and foster convergence in the application of the rules. This accentuates the EC s belief that the establishment of an independent oversight board with the vested power to enforce cooperation among member states will ultimately improve audit quality and the level of confidence investors place on financial statement reporting, thereby contributing to financial stability. In a similar vein, at an international level, the International Forum of Independent Audit Regulators (IFIAR) was established in September 2006 by 18 independent audit regulatory organizations (these being the POBs in these countries) around the world. The mission of IFIAR is to create a platform to share knowledge of the audit market environment and practical experience regarding independent audit regulatory activity; to promote collaboration in regulatory activities; and to provide a focal point for contacts with other international organizations that have an interest in audit quality (IFIAR 2013). While membership of IFIAR is not mandated by law, the requirements for becoming a member of IFIAR clearly reinforce the expectation of an improved audit environment and outline how this can be achieved. IFIAR memberships are strictly confined to regulatory agencies that are: (1) independent of the audit profession, which means that the majority of the relevant governing body should be non-practitioners and that the funding of the board should be free of influence by the profession; and (2) engaged in audit regulatory functions in the public interest. The latter requirement refers to the responsibility of the POB in each IFIAR member country to conduct periodic inspections of audit firms undertaking audits of public interest entities (IFIAR 2012). Since its inception, IFIAR has increased its membership base from the initial 18 countries in 2006 to 44 countries at the beginning of Inspection programs of audit firms are the main technique for improving deficiencies in audit quality and driving improvements in the audit process. For example, the Financial Reporting Council ( ) in the United Kingdom describes the objectives and process of 9

10 their inspection program as to monitor and promote improvements in the quality of auditing. As part of our work, we monitor firms compliance with the regulatory framework for auditing, including the Auditing Standards, Ethical Standards and Quality Control Standards for auditors and other requirements under the Audit Regulations. ASIC (2012), the regulator of the auditing profession in Australia, describes the aim of their audit inspection program as being to promote high-quality external audits of financial reports of listed and other public interest entities in Australia. High-quality audits are an important contributor to financial report quality and market confidence. Although considerable resources go into these inspection regimes, very little is known about how effective they have been. In 2012, IFIAR released their first global survey of inspection findings (IFIAR 2012). The survey was designed to identify the level of inspection activity and common inspection findings related to the audits of public companies. The survey identified that more needs to be done to improve the consistency of performance by auditors, and confirmed that IFIAR members are noting audit findings in numerous common areas (inspection themes) including professional skepticism and tone at the top, group audits, revenue recognition, internal control testing, and engagement quality control review across the different jurisdictions. Prior Literature The increasing prominence of the various national audit regulators undertaking public oversight responsibilities for this profession was explored by Simnett and Smith (2005). They compared and contrasted the various structures that have been instigated or proposed by the leading national and international bodies in relation to public oversight of the auditing standard-setting process. These POBs had taken on various dimensions of both the auditing standard-setting responsibility, and the continuing monitoring function, mainly through 10

11 inspections of the auditors and audit firms. At this time a unique combination was that in the U.S., the PCAOB had taken on both the law-maker (developing standards) and the lawenforcer (inspections) aspects of this public oversight. Despite its potential merits, public oversight of the auditing profession has been subject to criticism and skepticism. For example, Glover et al. (2009) criticize the PCAOB for insufficient staff, staff with limited expertise, inadequate transparency of procedures and inspection outcomes, and slow feedback. Small U.S. audit firms also do not appear to see an improvement in audit quality or increased public confidence (Daugherty and Tervo 2010). Thus, as public oversight is contested in some quarters, its effectiveness has become the subject of academic research. Prior research on public oversight has mainly focused on the U.S., as the PCAOB is one of the few public oversight bodies that publicly disclose inspection findings. Early studies provide descriptive evidence on inspection reports that are issued by the PCAOB and the type of quality control defects and audit deficiencies that are found (Hermanson et al. 2007; Hermanson and Houston 2008; Roybark 2009). Subsequently, academic research has looked into the informational value of PCAOB inspection reports. For example, Dee et al. (2011) find a negative stock market reaction to PCAOB sanctions imposed upon one of the Big 4 audit firms. The informational value of PCAOB inspection reports has also been addressed by examining client hiring and firing of audit firms. For example, Lennox and Pittman (2010) find no changes in audit firm market shares following deficient inspection findings. However, Daugherty et al. (2011) find that for triennially inspected audit firms, clients are more likely to dismiss an audit firm with a deficient inspection report and switch to an audit firm with a clean inspection report. Similarly, Abbott et al. (2013) find that clients with effective audit committees or high potential agency conflicts are more likely to switch to a successor auditor without such deficiencies. 11

12 A number of studies suggest that public oversight bodies are able to discriminate between various levels of audit quality. For example, Gunny and Zhang (2013) find that clients of triennially inspected audit firms with deficiencies identified by the PCAOB have higher levels of abnormal accruals. A similar result is reported by Van Opijnen et al. (2011) in a Dutch setting. Furthermore, there is some support within individual countries as to the effectiveness of national public oversight. For example, Carcello et al. (2011) find that absolute abnormal accruals decrease following PCAOB inspections. Similarly, Fuentes et al. (2010) find that earnings quality has improved in Spain since the Spanish public oversight body started its inspections. Knechel et al. (2012) find that audit fees increase in response to deficient inspection results. Apart from the exceptions noted above, research on public oversight bodies outside the U.S. is very limited. The purpose of our study is to compare the effectiveness internationally of public oversight of the auditing profession and to provide a comparison of the relative effectiveness of the differing aspects of the organization of the various public oversight systems. Development of Hypotheses This study examines the effectiveness of public oversight of the auditing profession from an international perspective. In addition, we examine a number of specific choices that countries have made with regard to the organization of public oversight. First, we look into countries that have an inspection program in place and compare these with countries without inspections of audit firms. Secondly, in order to address potential endogeneity concerns, we consider whether the introduction of inspection processes in countries that introduced inspection programs during our sample period is followed by an increase in audit quality. In additional analyses, we look into the choice of a public oversight body of a country to become 12

13 a member of IFIAR. As discussed above, IFIAR membership is restricted to those public oversight bodies that meet certain criteria. Further, we examine whether there is any association between the characteristics of the various inspection regimes and audit quality. We infer audit quality by examining the accruals properties of audit firms client companies. In line with prior research, we consider that a high quality audit will be more effective at constraining opportunistic reporting by managers, which is reflected in lower levels of abnormal accruals (e.g. Carey and Simnett 2006; Francis and Wang 2008). We formulate the following hypotheses: Inspections For the reasons outlined earlier, we expect that companies domiciled in countries that have independent inspections in place have higher audit quality compared to companies domiciled in countries that do not perform independent inspections of audit firms. We therefore hypothesize that: H1: Companies domiciled in countries where audit firms are subject to independent inspections are associated with higher audit quality compared to companies domiciled in countries where audit firms are not subject to independent inspections. Pre-/post-inspection We understand that the above analysis could potentially suffer from endogeneity issues. Thus we examine whether, for the smaller group of countries in which public oversight was initiated during the sample period, there is a difference in audit quality for pre- versus post-inspection audits. We expect that the benefits of introducing an inspection process will result in audit quality being higher in the post-inspection period. This is tested by the following hypothesis: H2: Audit quality after the introduction of the inspection process will be higher than audit quality before the inspection process began. 13

14 III. DATA, DESCRIPTIVE STATISTICS, AND RESEARCH DESIGN Sample Selection and Description To collect data for this study, we use the Worldscope database which provides financial data covering 91 countries. The initial scope of the sample is to include all 40 member countries of IFIAR as at the end of , and another 51 non-ifiar member countries that are covered in Worldscope. However, data for some countries are lacking or insufficient for this type of international research and are therefore excluded from the sample. Furthermore, the data is then filtered to only contain samples that do not have any missing variables that are required to calculate the dependent and independent variables. Through the process of filtering, several countries whose number of complete observations fell below 50 per year are eliminated. Finally, consistent with prior studies, financial institutions with Standard Industrial Codes are excluded from the sample because this industry is subject to specific accounting requirements that significantly affect the calculation of discretionary accruals. The final dataset contains a total of 33 countries, seven of which do not perform independent inspections and eight of which are non-ifiar members at the end of Our sample period is , but we also collect data for 2005 for the calculation of the discretionary accruals of In total, we have 39,278 firm-year observations over the period The year 2006 is an appropriate year to start to undertake this analysis, because as well as IFIAR coming into existence in this year, it is also after the introduction of International Financial Reporting Standards (IFRS) for most countries, and it is the year that the European Commission published its revised EU 8th Directive, requiring member countries to install public oversight (with an implementation deadline of two years maximum). For the testing of Hypothesis 2, we restrict the sample to the 11 countries where public oversight and 2 Data collection occurred in 2012; at that time, data was only available up until the end of

15 inspections started during the period , and use the year 2006 as the baseline where, by design, no country has an inspection regime in place. Table 1 provides an overview of the number of firm-year observations per country included in the sample in the period and its inspection regime, IFIAR membership status, and the year inspections by independent POBs commenced. We gather information on the frequency, experience, and type of inspection system using the country profile documentation available on the IFIAR website, the official websites of the POBs and/or through follow up direct contact with these boards. This information, for listed entities in each country, is summarized in Table 1. Specification of Empirical Models To test our hypotheses, we infer audit quality from reduced managerial discretion resulting in less abnormal accruals, all other things being equal. As commonly used in the literature, we measure abnormal accruals by means of the performance-adjusted crosssectional modified Jones model (Jones 1991; Dechow et al. 1995; Kothari et al. 2005). This model has also been widely used in the international auditing literature to measure audit quality (e.g., Kwon et al. 2007; Francis et al. 2013). Specifically, we estimate: TA i, t 0 1 country 1 Assets / year 2 ( REV i, t AR i, t i, t 1 / industry fixed effects ) PPE 3 i, t ROA 4 i, t [1] Where: TAi,t is total accruals in year t (defined as the change in non-cash current assets minus change in current liabilities excluding current portion of long-term debt, minus depreciation and amortisation) scaled by lagged total assets; ASSETSi,t-1 is a firm s total assets in year t-1; ΔREVi,t is sales in year t less sales in year t-1; ΔARi,t is accounts receivable in year t less accounts receivable in year t-1; PPEi,t is net property, plant and equipment in year t; and ROAi,t is return on assets in year t. A company s unadjusted abnormal accruals are set equal to the firm-specific residuals estimated from the above model of expected (normal) accruals. 15

16 As an alternative measure for abnormal accruals, we use abnormal working capital accruals based on the expectation model in DeFond and Park (2001). A common argument in the literature is that non-working capital accruals are less susceptible to manipulation than working capital accruals (e.g., DeFond and Jiambalvo 1994; Teoh et al. 1998). Specifically, we calculate abnormal working capital accruals as realized working capital minus expected working capital, where expected working capital is assumed to be a fixed proportion of sales (Dechow et al. 1998; DeFond and Park 2001). Or formally: AWCA i, t WC i, t ( WC i, t 1 / S i, t 1) * S i, t [2] Where: AWCAt is abnormal working capital accruals in year t; WCt and WCt-1 are noncash working capital in year t and year t-1, respectively, where non-cash working capital is computed as the difference between (current assets minus cash and short-term investments) and (current liabilities minus short-term debt); and St and St-1 are sales in year t and year t-1. To test our hypotheses, we use the following model: ATA i, t or AWCA i, t Size MB 6 i, t i, t Leverage PPE _ Growth i, t i, t Performance Auditor Growth Rule of Law Loss Variables_ of _ Interest Industry Fixed Effects Year Fixed Effects i, t 9 i, t 4 i, t i, t 5 i, t [3] Where: ATAi,t is the absolute value of performance-adjusted abnormal total accruals measured by modified Jones model in Equation 1; AWCAi,t is the absolute value of abnormal working capital accruals in Equation 2; Sizei,t is the natural logarithm of total assets (in millions); Leveraget is the ratio of long-term debt to total assets; Performancei,t is the operating cash flow scaled by total assets; Growthi,t is the percentage of year-to-year growth in sales; Lossi,t is an indicator variable for loss in the current year; MBi,t is a company s market value of equity scaled by book value of equity; PPE_Growthi,t is a company s one year growth in gross property, plant and equipment from year t-1 to year t. Auditori,t is an 16

17 indicator variable for Big N auditors; Rule of Lawi,t is a World Bank Governance Indicator developed by Kaufmann et al. (2010 4) and represents capturing 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 ; and Industry is a vector of industry indicator variables and Year is a vector of year indicator variables. The variables of interest are: Inspection, which is an indicator variable taking the value of 1 when the observation relates to a country where independent inspections are in place; Post_Inspect, which is an indicator variable taking the value of 1 when the observation relates to the post-inspection period; IFIAR, which is an indicator variable taking the value of 1 if the observation relates to a country whose regulator is a current member of IFIAR; First_time, which is an indicator variable taking the value of 1 when it is the first year that inspections are performed; Ln_exp, which is the natural logarithm of the number of years of inspection experience; Type_oversight, which represents the type of oversight system taking the value of 2 in case of direct oversight, 0 in case of oversight through the profession under supervision of the oversight board, and 1 in case of a combination; and Freq_Inspect, which is an indicator variable taking the value of 1 in cases where large audit firms are inspected on an annual basis. IV. RESULTS Descriptives Table 1 presents the 33 countries included in the sample, the number of observations per country, IFIAR membership and other public oversight features. The most highly represented countries in the sample in terms of number of observations are Japan (20.33%) and the U.S. (16.02%). All other countries individually represent less than 7% of the total sample. To control for the discrepancy in the number of observations and as our focus is at the 17

18 country-level, we use weighted regressions where we equally weight countries rather than equal weight client observations, consistent with the approach used by Choi and Wong (2007). During the sample period ( ), 25 countries (75.75%) are members of IFIAR. 3 There are seven countries (21.2%) in our sample that do not have a public oversight system in place during the sample period 4, and 13 countries (39.39%) that installed a public oversight system during the sample period. The year when public oversight inspections started in the countries in our sample ranges from 1978 to Table 1 further demonstrates that within the group of countries with a public oversight system in place, there is much variation in the organization of public oversight. There is variation in the type of oversight system, which can be direct inspection (16 countries), primarily direct inspection (4 countries), through oversight only (4 countries), or a combination of direct inspections and oversight (2 countries). Also, the frequency of oversight varies ranges from on demand (Brazil) or from time to time (Malaysia), to 3 years for PIE audit firms (which is the minimum requirement for EU countries) up to annually for large audit firms (Australia, Canada, Germany, Switzerland, Taiwan, Thailand, United Kingdom, and United States). - INSERT TABLE 1- Table 2 presents the descriptive statistics of the dependent and control variables, the variables of interest, and the distribution across industries. Table 2, Panel A, shows that the mean (median) abnormal total accruals in absolute value is (0.037). The mean (median) total assets is 3,866 million USD (373 million USD), while the ratio of long-term debt to total assets has a mean (median) of 16.7% (12.7%). The mean (median) absolute value of operating 3 As can be seen from Table 1, two other countries in our sample joined IFIAR after It is noted that these seven countries did not have an oversight body in place in 2011, but some of these countries recently have or are in the process of installing public oversight. Hong Kong, India, Indonesia and Russia have, or are in the process of establishing, a public oversight system and are considered in the transitional period for European Union recognition (information as at January 2011); Israel: no public oversight system (information as at January 2011); China: European Commission declared the oversight system of China to be equivalent to those of the EU (information as at January 2011); Chile: no public oversight system. 18

19 cash flow scaled by total assets is (0.075). The mean (median) growth rate measured as the percentage change in sales is 13.6% (3.5%), and the mean (median) growth rate of gross property plant and equipment is 15.8% (2.2%). The mean (median) market to book value is 1.4 (1.2). In the sample, 25.4% of the observations have a negative income, and 80.3% are audited by a Big 4 auditor 5. The rule of law has a mean (median) value of (1.352). Table 2, Panel B, shows that the majority of the observations (53.96%) relate to the manufacturing industry (SIC 20-30), followed by 12.39% in the utilities industry (SIC 40-49), and 11.77% in the wholesale trade industry (SIC 50-59). Table 2, Panel C, presents the mean (median) values of the variables of interest in the analyses and the corresponding number of observations. For the testing of Hypothesis 2, we concentrate on those countries where public oversight was introduced in the period , and use the year 2006 as the benchmark as, at that stage, no country had public oversight in place. In total, 11 countries started their inspections in the period , which corresponds with 10,718 observations of which 58.2% (41.8%) relate to the pre- (post) inspection period. In the additional analysis, for the testing of IFIAR membership, we focus on the period , and exclude the year 2006 since IFIAR was founded in September For this analysis, we have 31,833 observations, 78.7% of which belong to an IFIAR member country. Our additional analysis of characteristics of inspections uses a sample of those countries where inspections were in place before or during the sample period, which represents 28,415 observations. Of these observations, 8.2% relate to a country where inspections are performed for the first time. The mean (median) years of experience with inspections is 4.5 (4) years. With regard to the type of oversight system, the median observation relates to a country where direct or primarily direct inspection is in place. Note 5 We note that the data for the type of auditor are retrieved from Worldscope which only offers static data for this data item. As we do not have access to earlier versions of Worldscope, we cannot control for changes during the sample period. 19

20 that in the analyses we classify primarily direct inspection under direct inspection since our sample is comprised of listed companies and all listed companies are subject to direct inspection in those countries where primarily direct inspection applies. -INSERT TABLE 2- Multivariate Analysis Table 3 presents the results of the testing of Hypothesis 1 on the association between inspections and audit quality (measured by abnormal total accruals). The results (Column A) show a significant negative coefficient for the Inspection variable (p<0.01), implying that companies domiciled in countries with independent inspections have significantly lower levels of abnormal accruals. This would suggest that audit quality is higher in those countries as auditors appear to act as a stronger constraint on earnings management. Significant control variables are size, PPE growth, loss, auditor, and rule of law, all of which are in the expected direction. We repeat the analysis for income-increasing (Column B) and income-decreasing (Column C) accruals, and find that inspections are associated with significantly less incomeincreasing (p<0.10) and income-decreasing accruals (p<0.01). We repeat these analyses with abnormal working capital accruals as the dependent variable. The results (not tabulated for parsimony) are qualitatively similar: inspections are associated with significantly less abnormal working capital accruals (p<0.01). Similarly, inspections are associated with significantly less income-increasing abnormal working capital accruals (p<0.05), and significantly less income-decreasing abnormal working capital accruals (p<0.01). -INSERT TABLE 3- In Table 4, the results of the analysis of Hypothesis 2 relating to audit quality pre- and post-inspections are presented. The coefficient of the post inspection variable (Column A) is significantly negative (p<0.01). This implies that the level of abnormal accruals of companies 20

21 domiciled in countries that installed public oversight during the period is significantly lower in the post-inspection regime compared to the pre-inspection regime. Significant control variables are size, leverage, PPE growth, and loss, all in the expected directions. This result holds for income-increasing (Column B, p<0.10) and for incomedecreasing accruals (Column C, p<0.01). A univariate t-test shows that the mean value of absolute abnormal accruals is in the pre-inspection regime (n=6,236) and drops to in the post-inspection regime (n=4,482) (p<0.01). There is a potential concern that all environments have experienced an increase in audit quality over this period, and so we compare these results with companies in our sample from countries in which public oversight was not installed either before or during the period For this sample we find a mean absolute value of abnormal accruals of in the benchmark year 2006 (n=1,117), and in the years (n=5,276), which suggests a slight increase in abnormal accruals. In a multivariate analysis (not reported), the coefficient of the benchmark year 2006 is not significant. Collectively, this supports the view that the implementation of an inspection regime has resulted in an increase in audit quality, as over the same period there is no similar improvement in audit quality where there are no inspection processes in place. We repeat the analyses of Table 4 with abnormal working capital accruals as the dependent variable. Untabulated results show that post-inspection these accruals are significantly lower in absolute value (p<0.01), which is attributed to income-decreasing (p<0.01) abnormal working capital accruals being significantly lower, while income-increasing abnormal working capital accruals are negative but not significant. We note that if we run unweighted regressions, we obtain significant lower abnormal working capital accruals in absolute value (p<0.01), both for income-increasing (p<0.05) and income-decreasing (p<0.01) abnormal working capital accruals. -INSERT TABLE 4-21

22 Additional Analyses IFIAR Membership As outlined earlier, IFIAR membership is confined to regulatory agencies that are independent of the audit profession and conduct periodic independent inspections of audit firms undertaking audits of public interest entities. Both of these characteristics of a POB should result in a higher audit quality in that country, and thus we expect that IFIAR membership has a positive association with audit quality. Since IFIAR was founded in September 2006, the period of study for this analysis starts in Table 5 presents the results of the analysis on the association between IFIAR membership and audit quality. The results (Column A) show a significant negative coefficient for IFIAR (p<0.01), implying that companies domiciled in IFIAR member countries have significantly lower levels of abnormal accruals. This would suggest that audit quality is higher in those countries as auditors appear to act as a stronger constraint on earnings management. We repeat the analysis for income-increasing (Column B) and income-decreasing (Column C) accruals, and find that IFIAR membership is associated both with significantly less incomedecreasing accruals (p<0.01), and negative but insignificant income-increasing abnormal accruals. We repeat these analyses with abnormal working capital accruals as the dependent variable. The results (not tabulated for parsimony) are qualitatively similar: IFIAR membership is associated with significantly less abnormal working capital accruals (p<0.05). For IFIAR membership, we find a significant negative association with income-decreasing abnormal working capital accruals (p<0.05), and a negative but insignificant association with income-increasing abnormal working capital accruals. -INSERT TABLE 5-22

23 Unweighted Regression Analyses We re-ran the models (untabulated for parsimony) presented in Tables 3 and 5 using unweighted regressions and subsequently excluding alternatively Japan and the U.S., which are the dominant countries in the sample. In all cases (full sample, with and without Japan, with and without the U.S.), we observe a significant negative coefficient for Inspection (as in Table 3) for the absolute value of abnormal total accruals, income-increasing absolute abnormal total accruals, and income-decreasing absolute abnormal total accruals. For IFIAR, we observe a significant negative coefficient for the full sample and without the U.S. for the absolute value of abnormal total accruals, income-increasing absolute abnormal total accruals, and income-decreasing absolute abnormal total accruals. Without Japan, the coefficient for IFIAR is only significantly positive for income-decreasing abnormal accruals. Unweighted regressions (untabulated for parsimony) for the models presented in Table 4 also give a significant negative coefficient for the variable Post_Inspect for the absolute value of abnormal total accruals, income-increasing absolute total accruals, and income-decreasing absolute total accruals. Note that Japan and the U.S. are not included in the sample used for testing of hypothesis 2 as shown in Table 4 and therefore their inclusion or exclusion will not influence the results reported. Characteristics of inspection process In examining the impact of the major characteristics of the inspection process on audit quality we focus on those countries where independent inspections of audit firms are in place, and consider a number of choices made by regulators in the design of inspection processes. Specifically, we consider three major characteristics of the inspection regime, being the period of time since inspections commenced (outlining the number of years of inspection 23

24 experience), the type of inspection system, being inspections directly undertaken by the independent POB or undertaken by professional bodies under the POB s supervision, and the frequency of inspections. This allows an examination as to whether variation in the characteristics of inspection regimes is associated with audit quality. We examine whether a learning effect arises from the cumulative experience of being inspected. We also consider the differential impact of the two main types of inspection systems on audit quality. We consider whether inspections of audit firms directly performed by inspectors of independent POBs have a stronger association with audit quality compared to inspections undertaken by a professional body under the supervision of the POB. We argue that inspection directly undertaken by the POBs will result in a stricter and more rigorous type of inspection system. The rationale for this analysis is consistent with the reasons why there has been a move in the last decade from self-regulation by the audit profession to oversight by independent POBs. In our third additional analysis we consider the frequency of inspections conducted. We expect that more frequent inspections are associated with higher audit quality. The reasoning behind this is that the incentives of audit firms to be lenient towards their client companies may decrease as a function of the likelihood of being caught for malpractice, which may increase with the frequency of inspections. The results of these analyses relating to features of inspection regimes and audit quality are as follows. None of the features that we consider (number of years of inspection experience, type of oversight system, and frequency of inspections) are significantly associated with the absolute value of abnormal total accruals. Similarly, none of these features significantly affects income-increasing abnormal total accruals. We also do not find any support for a significant association between features of inspection regimes and audit quality using abnormal working capital accruals, except that direct inspections significantly increase income-decreasing abnormal working capital accruals (p<0.10). 24

25 V. CONCLUSION One of the most profound changes in audit regulation of the past decade has been the move to independent public oversight of the auditing profession, putting an end to the traditional model of self-regulation. As the potential merits of public oversight are contested by some, research on its effectiveness is warranted. Currently, the emerging literature on public oversight has mainly focused on the U.S., since it is one of the few countries where the public oversight body publicly discloses inspection findings. The issue is much broader than the U.S., and the move to independent public oversight bodies has been a world-wide trend, emphasized by the formation of IFIAR in September 2006, growing to a membership of 44 countries in early However, apart from a couple of studies relating to individual countries other than the U.S., to the best of our knowledge, there is no international study on the effectiveness of independent public oversight which we examine here. In our examination of the effectiveness of public oversight in its different forms at an international level, we make use of a large sample of non-financial companies located in 33 countries over the period We observe that the level of abnormal accruals is significantly lower for companies domiciled in countries where independent inspections of audit firms are conducted and in IFIAR member countries. This suggests that audit quality is stronger in countries where a national independent inspection regime is in place as well as in IFIAR member countries, where the POB is bound by a number of requirements that are expected to lead to an improved audit environment. We further concentrate our analysis on the key plank of the strategy and work effort of the independent oversight bodies, which is the inspection of audit firms. We find that the level of abnormal accruals is significantly lower in the period after the introduction of inspections compared to the pre-inspection regime. This provides evidence that audit quality improves 25

26 after installing public oversight and conducting inspections. We also consider the various characteristics of the inspection regimes to identify whether variation in the performance characteristics of these inspections is associated with audit quality. We do not find much difference in audit quality that could be explained by different features of the public oversight system. Collectively, our results contribute to the emerging literature and debate on public oversight by providing insights from an international setting. Our findings may prove useful for regulators in the further development, design and mutual recognition of public oversight systems. They do suggest that one of the major impacts on audit quality is having an inspection program, and that the specific characteristics of the inspection programs are less important. One of the potential limitations of our study is endogeneity concerns, especially with regards to the analysis of impact of countries with inspections and IFIAR membership. Our pre-/post-analysis, where we find that the level of abnormal accruals is significantly lower in the post-inspection regime compared to the pre-inspection regime, and that this finding does not hold for audits not subject to inspection regimes over a similar period, helps to alleviate this concern. However, we cannot categorically attribute the observed improvement in audit quality directly to the performance of inspections. Although this is the major initiative of the independent oversight boards aimed at audit quality, it may be that other common factors associated with the countries that have created independent oversight boards that have instigated active inspection regimes have also contributed. Alternatively, it may be not that inspections themselves are improving audit quality but that the expectation of having audit work subjected to external scrutiny and public disclosure may improve audit practices and impact audit quality 6. A further limitation is that we are restricted to one measure of audit 6 The Consultative Committee of Accountancy Bodies in the United Kingdom conducted a series of interviews with stakeholders regarding perceptions of the impact of inspections and transparency reports in 2011: Some 26