Does the Joint Provision of Sustainability Assurance and Financial Audit Improve Financial Audit Quality?

Transcription

1 Does the Joint Provision of Sustainability Assurance and Financial Audit Improve Financial Audit Quality? Abstract: We examine whether the joint provision of corporate social responsibility (CSR) assurance services and financial audit improves audit quality and influences audit fees. We predict that the provision of CSR assurance and financial audit by the same audit firm could lead to knowledge spillovers from the CSR assurance team to the financial audit engagement team, which could lead to more efficient audits and higher audit quality. Using international data for more than 29,000 firm-year observations from 58 countries, we find that audit firms that perform both CSR assurance and financial audits for the same client have higher audit quality (i.e., less frequent restatements and more frequent going concern opinions) at no greater cost to the client. We also find that this effect is stronger in settings where economic bonding is less and knowledge spillovers are more likely to occur. Our results are important especially because of the growing number of countries that require assurance of CSR reports. Keywords: CSR Assurance, Audit Quality, Knowledge spillover, Audit fees, Non-Audit Services 1

2 1. Introduction The stakeholder view of the firm (Freeman, 1984; Magill et al., 2015; Hawn et al., 2018) has gained considerable momentum over the past two decades and seems to have won the battle of ideas (Economist, 2005) over the shareholder view of the firm (Friedman, 1962). Under the stakeholder view, firms are responsible for an array of activities such as preventing environmental externalities, producing safe and healthy products, ensuring the human development of their employees, contributing to government spending by paying taxes, and finding suppliers that respect regulations and ethical principles, i.e., more than just maximizing shareholder value through financial performance. This shift to the stakeholder view is evident as more and more firms respond to stakeholders demand for non-financial information by issuing CSR reports in an attempt to provide information about non-financial performance and the associated risks regarding issues such as environmental performance, social performance and corporate governance (Cahan et al., 2015; Stolowy and Paugam, 2018). In this study, we investigate whether audit firms that provide CSR assurance services and financial statement audits for the same client deliver superior audit quality. We also examine the cost of delivering the audit services. More and more traditional investors, i.e., investors that do not specifically follow a CSR-related investment strategy (or in the finance industry: environmental, social and governance investing) integrate CSR performance into their stock and bond selection decisions (CFA Institute, 2017). One reason is that firms are being held accountable for the negative impact of their activities on various stakeholders (e.g., tax evasion, environmental externalities, unsafe or unhealthy products, improper practices with regard to labor). Therefore, CSR performance could have important financial consequences that affect a firm s stock price and investment strategy. CSR reports help outsiders to assess the extent of risks associated with CSR performance (Steinmeier and Stich, 2018). CSR reports are frequently assured by accounting firms to verify their content and enhance their credibility (Cohen and Simnett, 2015; Peters and Romi, 2015; Ballou et al., 2018; Steinmeier and Stich, 2018). Specific knowledge about CSR performance could create valuable information for auditors to more accurately assess firm-specific risks related to CSR activities. Firms activities lead to risks that have consequences for financial statements such as for provisions booked for specific risks (e.g., environmental risk, litigation risk with employees or suppliers), for impairment of assets (e.g., impairment of fixed assets due to negative environmental externalities), or for tax matters (e.g., deferred taxes). Valuable knowledge about such risks could be gained if the audit firm also providesg assurance services for the client s CSR report. Information transfers from the CSR 2

3 assurance team to the financial audit team may exist and lead to more efficient financial audits and higher audit quality. Such information transfers are unlikely to increase audit fees and, in some cases, could even reduce audit fees. If financial audit teams have access to information from the CSR assurance team they could be more efficient in their efforts to collect audit evidence for assessing whether a client s financial statements fairly represent its financial performance, financial position, and cash flows. Auditors perform various testing procedures based on a firm s information systems and accounts to detect and correct material errors. Information from the CSR assurance team could help auditors target which accounts or information systems to investigate based on issues detected by the CSR assurance team, thus leading to more efficient and effective audits. We predict that, because of information transfers from the CSR assurance service team, audit firms that also provide CSR assurance services for their clients exhibit superior audit quality and that this higher audit quality is delivered at no greater cost to the client (i.e., auditors deliver better quality for the same or lower price). This argument is analogous to the knowledge spillover argument developed in the non-audit service literature, which suggests that non-audit services facilitate the creation and dissemination among financial auditors of client-specific knowledge that helps in conducting the financial audit (Simunic, 1984; Beck et al., 1988; Antle and Demski, 1991; Wu, 2006). It suggests a positive relation between non-audit services and financial audit quality. Some studies corroborate the knowledge spillover argument for catch-all categories of non-audit services or for specific services such as tax or information technology (Kinney et al., 2004; Antle et al., 2006; Knechel and Sharma, 2012; Koh et al., 2013; Bell et al., 2015). 1 However, to date no empirical study has investigated the implications of the joint provision of CSR assurance services and financial audit by the same audit firm for audit quality and audit pricing. Given that the focus and nature of CSR assurance services differs markedly from that of more traditional non-audit services such as tax and information technology, it is unclear whether the joint provision of CSR assurance and financial audit will have different or similar implications for audit outcomes. Unlike tax, transaction, or information technology services, CSR assurance services involve verification of (non-financial) information similar to financial audits in order to ensure that firms disclose credible and reliable CSR reports. These activities can therefore result in synergies between CSR assurance and financial audits. Additionally, CSR assurance services are relevant for understanding CSR risks which are becoming more important and may have implications for financial statements (e.g., provisions). These implications are important because regulators are increasingly likely to require 1 However to date, the empirical research with regard to the relation between non-audit services and financial audit quality is inconclusive (Habib, 2012). DeFond et al. (2002) and Ashbaugh et al. (2003) find no association between non-audit services and audit quality whereas Frankel et al. (2002) document a negative association between non-audit services and financial audits. Davis et al. (2009) and Bell et al. (2015) report a positive association for public firms between non-audit services and financial audits. 3

4 assurance of CSR reports (High-Level Expert Group on Sustainable Finance, 2018). 2 It is therefore important to understand ex ante the implications for audit quality of CSR assurance services provided by the same auditor or by a different assurance provider because many firms will decide whether they should retain their financial auditor for the provision of CSR assurance in the future. It is not ex ante clear that the provision of CSR assurance services would necessarily improve audit quality because CSR assurance may create a financial dependence between the auditor and the client. One argument drawn from the non-audit services literature is that CSR assurance could impair the auditor s independence because the revenues received from the client could create economic incentives to maintain a good and ongoing profitable relationship with the client, and therefore to avoid reporting any discovered misstatements in financial audits (or respond positively to financial reporting demands from the client) (e.g., Frankel et al., 2002). This argument, which is generally referred to as economic bonding, suggests a negative relationship between non-audit services and financial audits. A quintessential example is the highly publicized economic bonding between Arthur Andersen and Enron in which Arthur Andersen was deriving important consulting revenues from Enron (and other clients) while also serving as its statutory auditor (Chaney and Philipich, 2002; Toffler and Reingold, 2003). 3 Economic bonding of auditors that also provide CSR assurance services may occur for example for firms with weak corporate governance or in countries with regulatory environments that do not ensure auditors are sufficiently independent from their clients. We examine the implications for audit quality of the joint provision of CSR assurance and financial audit by the same (Big 4) auditor for an international sample of more than 29,000 observations from 58 countries. We control the variation in audit quality that may result from differences in auditor type by excluding firms that have non-big 4 financial auditors, while allowing only the CSR assurance type provider to vary. We use an international sample for three important reasons. First, it includes considerable variation in economic and business environment conditions that are likely to influence information spillovers from the CSR assurance team to the financial audit engagement team. Some countries are more stakeholder oriented than others (Dhaliwal et al., 2014; Flammer and Kacperczyk, 2016). In more stakeholder-oriented countries, the extent of knowledge spillovers between the CSR assurance team and the audit engagement team could be greater. The high level of variation allows us to conduct more powerful tests. Indeed, 2 For instance, in France the Grenelle II law of 2010 and the Warsmann 4 law of 2012 require public firm to use a third party to verify CSR information included in the management report (rapport de gestion) (see article of the French Code of Commerce) 3 We do not have data about CSR assurance audit fees. However, anecdotal evidence from discussion with auditors in France suggests that CSR assurance fees are between 10,000 and 100,000. Most CSR assurance fees are in the lower range while higher fees are charged larger clients requiring a reasonable assurance of CSR reports. 4

5 information transfer may depend on the relative importance of CSR acvitities in the country, which are affected by the local legal environment (e.g., environmental regulations, labor laws). In some countries, firms are more likely to be held accountable for negative externalities on stakeholders than in others (Dhaliwal et al., 2014). Second, independence and enforcement rules to ensure audit quality differ across countries and therefore may differentially influence the role played by economic bonding for auditors that jointly perform CSR assurance and financial audit. For instance, some countries have stricter audit working environment rules, which ensure that even though an auditor is performing non-audit services her independence is not jeoparadized because of the strong institutional constraints (e.g., independence rules, independent auditor oversight bodies, strict sanctions, auditor inspections, mandatory partner or firm rotations). In other countries, the audit working environment may be more lenient and therefore create an environment where economic bonding is more likely to develop if an auditor provides both CSR assurance and financial audit services. We rely on an international setting to allow differences across institutional settings because we reason that economic bonding and knowledge spillovers are likely to depend on country-specific institutions. Third, using an international sample increases the generalizability of our findings relative to a single-country setting. We find that the joint provision of CSR assurance services and financial audit by the same (Big 4) auditor is positively associated with audit quality. We also find that audit fees do not increase when a client uses the same Big 4 firm to provide CSR assurance services and financial audit. Using earnings restatements and going concern opinions as our output measures of audit quality, we find that the joint provision of CSR assurance and financial audit is associated with less frequent restatements and more frequent going concern opinions. In cross-sectional tests based on firm and country characteristics, we find that the joint provision of CSR assurance and financial audit is associated with higher audit quality for firms or in countries where economic bonding is less likely to occur because of their strong corporate governance and where knowledge spillover effects are more likely to exist because of the stakeholder-orientation of the firm or the country. Our findings hold after we control for several other factors that prior research has identified are related to audit quality and for firm fixed effects, as well as after controlling for endogeneity. Our study makes several contributions to the literature. Prior research on non-audit services shows that some services, such as tax-related non-audit services, actually improve financial audits (Kinney et al., 2004). We contribute to this line of research by exploring an underesearched line of non-audit services, CSR assurance. We study this non-audit service in an international setting and exploit the variation across countries to demonstrate which country or firm characteristics explain information spillovers from the CSR assurance team to the financial audit engagement team. Our 5

6 study has important implications as the financial crisis raised concerns about the quality of auditors work and whether they were sufficiently independent to report accounting misstatements (ACCA, 2011). Responding to these concerns, several jurisdictions issued regulations in an attempt to strengthen financial auditor independence, including limiting certain forms of non-audit services (e.g., European Commission, 2011b; European Commission, 2011a; European Parliament and Council of the European Union, 2014). Our findings suggest that the joint provision of CSR assurance and financial audit can actually provide knowledge spillovers that lead to improved audit quality. The rest of this paper is organized as follows. We review the literature on CSR assurance in Section 2, develop our hypotheses in Section 3, present the sample selection and research design in Section 4, report the results in Section 5, discuss several additional tests in Section 6, and conclude the study in Section Prior studies on CSR assurance Firms that disclose CSR information may voluntarily rely on assurance services from a third party in order to enhance the credibility of the reported information (Simnett et al., 2009a; Brown-Liburd and Zamora, 2015). In the last few years, an increasing number of firms, especially larger ones, have begun issuing assured CSR reports (KPMG, 2013; Ernst & Young and Global Reporting Initiative, 2014). This is evident from the increase in the percentage of the top 250 global firms that issued an assured CSR report from 46 percent in 2011 to 59 percent in 2013 (KPMG, 2013). Of these large global firms, approximately two thirds that issued assured CSR reports used the services of a Big 4 audit firm (KPMG, 2013). Simnett et al. (2009b) investigate the determinants of CSR assurance for an international sample of firms. They find a positive association between firms looking to enhance the credibility of CSR reports and increase their reputation and the use of external CSR assurance. Using a sample of European public firms, Fuhrmann et al. (2017) show that issuers of assured CSR reports with a high level of assurance exhibit lower bid-ask spread relative to matched firms that issue non-assured CSR reports. CSR assurance also relates to another important question, i.e., whether audit firms, in particular Big 4 firms, are able to deliver superior CSR assurance relative to other CSR assurance providers such as CSR specialists (e.g., Bureau Veritas, Lloyd s, DNV, ERM). DeFond and Zhang (2014) argue that the expansion of audit firms to sustainability assurance services raises the question of whether auditors incentives and competencies transfer to non-financial settings. CSR specialists maintain that they have superior industry expertise regarding CSR activities and are 6

7 therefore better equipped to provide assurance services than audit firms. Conversely, Big 4 audit firms claim that they can deliver better CSR report quality assurance because they can benefit from their unique skills and the methodology used in financial audits. Big 4 audit firms reason that their reputation and geographical networks are important to provide high quality CSR assurance services for large clients. These audit firms can acquire or develop internal expertise with regard to CSR activities. Simnett et al. (2009b) find that firms located in stakeholder-oriented countries tend to choose assurance providers from the accounting profession. Ballou et al. (2018) show that auditors identify inaccuracies in CSR reports earlier and prevent future CSR reporting inaccuracies relative to other types of CSR assurance providers. Our study contributes to the body of knowledge on the consequences of CSR report assurance services by examining the implications of the joint provision of CSR report assurance services and financial audits for audit quality and audit pricing. 3. Hypotheses Prior literature that focuses on financial non-audit services such as tax consulting, management advisory services, public securities offering, or information systems consulting argues that nonaudit services may generate valuable knowledge spillovers for financial audits (e.g., Koh et al., 2013; Svanström, 2013; Bell et al., 2015). Similarly, CSR assurance services provided by the same audit firm may facilitate information transfer between the CSR assurance team and the financial audit engagement team. 4 The provision of sustainability assurance may help auditors identify risks related to CSR activities (O Dwyer et al., 2011). These risks are likely to have implications for elements of financial statements such as provisions, impairments, and recognition of tax expense, and for the likelihood of material errors in information systems. Steinmeier and Stich (2018) argue that CSR assurance improves sustainability investment decisions because it increases the set of information available for managerial decision making and thus leads to improved investment efficiency. In addition, Steinmeier and Stich (2018) find that CSR assurance reduces information asymmetry between managers and investors, which in turn facilitates monitoring of sustainability investment decisions. The study also reports some evidence of a stronger effect when the audit firm also provides CSR assurance. If credible CSR information generates valuable information for managers and investors, then auditors who review and verify 4 Discussion with a manager of an auditor providing CSR assurance services in France suggests that these knowledge spillovers exist. For instance the manager talked about inadequate information systems in a subsidiary of a listed firms that was uncovered by the CSR assurance team when reviewing employee information. The information was shared with the financial audit engagement team. 7

8 CSR information may be able to obtain a richer set of information about their client using first-hand knowledge that leads to a better and more efficient financial audit. The improved efficiency may also lead to a higher level of audit quality being provided to the client at no greater cost. Financial audit teams collect audit evidence to assess whether the financial statements give a true and fair view of the firm s performance and position and are in accordance with GAAP. If auditors have access to information from the CSR assurance team, they may be able to work more efficiently to collect audit evidence by reallocating resources based on information gained from the CSR assurance team. If the audit engagement team already has information about potential risks, then it can more efficiently allocate its costly, scarce resources. For instance, the CSR assurance team may review a client s payroll and staff numbers to assess the accuracy of the social dimension of reported CSR information and may find problems in the internal information system, which could be shared with the audit engagement team. The auditors could then directly focus on the implications for the financial audit of the discovered problems in the client s information system instead of investigating the origin of the problem. Similarly, the CSR assurance team may share knowledge with the audit engagement team on environmental matters such as carbon emission that may have implications for financial statements (e.g., provisions for environmental risk). However, the non-audit services literature also suggests that non-audit services may create economic bonding between the auditor and the client, which could ultimately lead to lower audit quality because the auditor has pressure to accept client demands (Simunic, 1984; Beck et al., 1988). Economic bonding is more likely if the firm has weak corporate governance (Larcker and Richardson, 2004) or if the client is a high growth firm (Reynolds et al., 2004). Causholli et al. (2014) suggest that future non-audit services fees impair audit quality because, prior to SOX, auditors had an incentive to compromise current audit quality in order to sell more non-audit services in future years because their compensation package is structured to reward non-audit services revenue. It is therefore also possible that CSR assurance services create an economic bond between the auditor and the client and thus impair audit quality. Given these competing arguments, we state the following hypothesis (in the null form): H1: The joint provision of CSR assurance services and financial audit by the same audit firm is unrelated to audit quality. If, as discussed earlier, information collected during the CSR assurance process is relevant to the financial audit and helps the financial auditors to better focus their audit efforts, it will facilitate more cost effective financial audits. If so, auditors could provide higher financial audit quality at no 8

9 greater or perhaps even lower cost. If, on the other hand, the provision of CSR assurance services strengthens the economic bond between the auditor and the client, then it is will likely lead to higher audit fees. Under the bonding argument the client accepts to be charged higher audit fees in exchange for leniency with regard to financial reporting (e.g., Asthana and Boone, 2012; He et al., 2017). Clients having an economic bond with their auditor may, for instance, report more aggressive earnings. The auditor could be lenient to preserve a profitable relation with its client. Given that the information spillover and the economic bonding arguments suggest opposite effect on audit fees, we hypothesize the following: H2: The joint provision of CSR assurance services and financial audit by the same audit firm is unrelated to audit fees. Regulation and the audit working environment have an effect on the economic bonding associated with non-audit services (Eilifsen and Knivsfla, 2013). Several institutional factors may mitigate the economic bonding effect for auditors that provide CSR assurance services to their clients. For instance, the importance to auditors of protecting their reputation may limit the risk of economic bonding (Weber et al., 2008). Other forces related to the audit working environment such as litigation risk, oversight by independent bodies or regular inspections and sanctions, mandatory rotation of audit firms or engagement partners, and limits on non-audit fees are also likely to limit the potential for economic bonding. Similarly, the strength of corporate governance is likely to weaken the economic bond between the auditor and the client and thus positively affect audit quality (He et al., 2017). In addition, the extent of potential information transfers from the CSR assurance team to the audit engagement team is likely to vary across institutional settings and across firms. The importance of the CSR assurance process to the financial audit varies with firm-specific factors. For instance, the CSR assurance team may obtain useful information for the audit engagement team for firms that are more concerned about stakeholders. Similarly, regulations and litigation risks related to CSR activities are likely to be affected by country characteristics (Dhaliwal et al., 2014). Some countries have stricter regulations about the environment and employees than others because they are more stakeholder oriented. As a result, the effect of knowledge spillovers on audit quality is likely to vary based on firm and country characteristics. Therefore, we state the following hypothesis: H3: Country institutional characteristics and firm characteristics influence the implications of the joint provision of CSR assurance services and financial audit by the same audit firm for audit quality. 9

10 4. Data and research design Sample We obtain data on CSR reporting and assurance from Thomson Reuters Asset 4, CSR assurance provider names, financial auditor names, audit fees, data on restatements and auditors opinions from Thomson Reuters Eikon, and firm-specific variables from Thomson Reuters Datastream. We begin our sample selection process with all firms covered by Thomson Reuters Asset 4 from 2002 to 2016 that have data available on CSR reporting and auditing. We begin our sample period in 2002 because Thomson Reuters Asset 4 initiated coverage in Table 1 summarizes the sample selection process for the financial quality sample (left column) and audit fees sample (right column). Importantly, we restrict the analysis to firms that use a Big 4 audit firm (i.e., Deloitte, Ernst & Young, KPMG, and PwC) for their financial statements to allow isolating the effect of variation in the assurance of CSR reports and financial audit by the same Big 4 audit firm (Lim and Tan, 2008). Including non-big 4 audit firms would lead to greater heterogeneity in financial audit quality and client characteristics which would complicate the analysis. This procedure also facilitates the identification of the auditor pairs based on the provision of CSR assurance services and financial audit. Both models start with 33,647 firm-year observations with available data on CSR reporting, CSR assurance, CSR assurance provider and financial auditor. We delete 4,078 observations with missing Datastream data variables on firm s characteristics. 5 To obtain the audit quality sample, we delete 258 firm-year observations with missing data on earnings restatement and/or auditor opinion. The audit quality sample includes 29,311 firm-year observations (5,235 firms). To obtain the audit fee sample we delete 8,107 observations with unavailable data on audit fees in Eikon. The audit fees sample includes 21,462 firm-year observations (4,701 firms). [Insert Table 1 About Here] We report sample distribution by industry, country and year for the two samples in Table 2. Panel A presents the distribution of observations for 58 countries and year for the audit quality sample. Most observations are from the US (9,870), followed by Japan (2,995), the UK (2,683), Australia (1,929), Canada (1,812) and Hong Kong (1,006). Panel B of Table 2 presents the distribution of observations across ICB industry and year. In the audit quality sample, most of observations are clustered in Financials (6,243), Industrials (5,373), Consumer Services (4,087), and Consumer Goods (2,983). Similarly, Panels C and D present the distribution of observations by 5 We include financial firms in our sample. Results are similar if we exclude financial institutions from our sample. 10

11 country-year and industry-year for the audit fees sample. Again, most observations are from the US (7,563), the UK (2,277), Japan (2,136), Australia (1,792), Canada (1,466) and Hong Kong (896). The industry distribution for the audit fees sample is as follows: Financials (4,427), Industrials (4,043), Consumer Services (3,072), and Consumer Goods (2,201). [Insert Table 2 About Here] Joint provision of CSR assurance and financial audit We examine the influence of the joint provision of CSR assurance and financial audit on audit quality and audit fees using a large international panel dataset. Our sample comprises four groups of firms providing different level of CSR information and using or not using CSR assurance services: (1) firms that do not issue a CSR report, (2) firms issuing a CSR report but that do not use CSR assurance, (3) firms issuing a CSR report assured by a different assurance provider than their financial auditor, and (4) firms issuing a CSR report assured by the same provider of the financial statements audit. Consequently, our empirical design considers the effects on audit quality and audit fees of three key elements for each firm-year observation: whether the firm issues a CSR report, whether the CSR report is assured or not, and whether the firm s financial auditor assures the CSR report. In order to identify whether a firm appoints the same provider for CSR assurance and financial audit, we obtain the code of the auditor of the financial statements for each firm-year observation from Eikon (TR.BSAuditorCode). Because the auditors code is standardized in two to four alphabetical digits, we match auditors code with auditors full name provided by Eikon. We use a computer-based procedure to identify Big 4 firms auditing the financial statement and manually check the resulting classification. 6 Further, we obtain the full name of the CSR assurance provider for each firm-year observation from Eikon (TR.CSRReportingExternalAuditName). We adopt the same procedure illustrated for financial auditors and identify Big 4 firms providing CSR assurance services. We classify all non-big 4 CSR assurance providers under the Others category. Finally, we generate auditor and CSR report assurance provider pairs by identifying the CSR assurance and the financial auditor for each observation. In identifying firms that use their financial auditor to assure their CSR report, we notice that Asset 4 incorrectly or incompletely classifies some firm-year observations: Asset 4 classifies some firms as using CSR report assurance services despite being coded as not issuing a CSR report (417 6 The computer-based procedure identifies whether the auditor s full name contains the following substrings: ERNST, E&Y, YOUNG, KPMG, PRICE, PwC, DELOITTE, TOUCHE. If one or more of these substrings occur, the procedure classifies the observation under the appropriate Big 4. 11

12 observations). Similarly, Eikon does not contain the name of the CSR assurance provider for some firm-year observations that are classified in Asset 4 as using assurance services for their CSR report, and vice versa (675 observations). We delete these observations from the sample. We define UCSRREPORTING, SAMEAUDIT, and DIFFERENTAUDIT, as follows. If a firm does not issue a CSR report, we code zero UCSRREPORTING, SAMEAUDIT, and DIFFERENTAUDIT. If a firm issues an unassured CSR report, we code UCSRREPORTING = 1, and we code zero SAMEAUDIT, and DIFFERENTAUDIT. For firms issuing an assured CSR report and using an assurance provider other than the financial statements auditor, we code DIFFERENTAUDIT = 1, and we code zero SAMEAUDIT and UCSRREPORTING. Finally, we code SAMEAUDIT = 1 if the firm issues an assured CSR report and use the same audit firm for both the financial statements and the CSR report and zero DIFFERENTAUDIT and UCSRREPORTING. Firms that are inactive in terms of CSR information and assurance serve as the control group. Figure 1 presents the distribution of our two samples across the four groups based on the CSR information level. [Insert Figure 1 About Here] Panel A and Panel B of Table 3 report the distribution across the four groups based on CSR information level by year, for the audit quality sample (Panel A) and audit fees sample (Panel B). We only comment the descriptive statistics presented in Panel A because Panel B presents a similar distribution for the audit fees sample. First, we notice a positive trend in the issuance of CSR reports. Panel A shows that 92.1% of firms did not issue a CSR report in 2002, this percentage decreases to 69.9% in Regarding the use of CSR assurance, in 2002, only 12 firms used assurance services for their CSR report (ten firms used a different assurance provider than their financial auditor and two firms used their financial auditor) and 41 firms issued an unassured CSR report. In 2016, 625 firms issued assured CSR reports (363 used a different assurance provider and 262 used their financial auditor). In 2016, firms issuing assured CSR reports account for 23.8% (( ) / 2,628) of firms. Among firms that use CSR assurance services, we find a positive time trend in the use of the same Big 4 auditor for CSR assurance over the period. In 2007 (when the number of firms using CSR assurance starts to increase), 28.5% of firms (= 78 / ( )) use their Big 4 auditor to assure their CSR reports whereas, in 2016, 42.0% of firms (= 262 / ( )) use their Big 4 auditor to assure the CSR report. Panels C and D of Table 3 show the sample distributions based on the providers of financial audits and CSR assurance. Panel C presents the distribution of providers of financial audits and 12

13 CSR assurance for the audit quality sample and Panel D presents corresponding statistics for the audit fees sample. Again, we only comment Panel C because Panel D shows similar descriptive statistics. Approximately half the firms use non-big 4 firms for CSR report assurance. We find that 3,084 observations (48.7% of observations issuing an assured CSR reports) use non-big 4 assurance providers. Big 4 auditors have relatively similar market shares for CSR report assurance services with market shares ranging between 10.3% (Deloitte) and 15.1% (PwC) for each Big 4 auditor. Additionally, we find that conditional to the use of a Big 4 auditor to provide CSR assurance services, it is more likely for firms to use CSR assurance services from the same Big 4 auditor auditing the financial statements relative to using other Big 4 auditors (the number of observations in the diagonals of Panel C of Table 3 is larger relative to the number of observations for each of the other Big 4 auditors). [Insert Table 3 About Here] Audit quality and Audit fees We use two direct measures of audit quality: earnings restatements and going concern opinions. Higher audit quality is generally associated with fewer earnings restatements and a higher likelihood of going concern opinions issued by auditors (e.g., Bills et al., 2016; Hardies et al., 2016; Lamoreaux, 2016; Pincus et al., 2017). 7 Compared to other measures of audit quality such as abnormal accruals, restatements and going concern opinions are direct measures of audit quality that are less subject to measurement error which is a particular concern in an international setting. Restatements show that the auditor signed off on materially misstated financial statements (Francis et al., 2013). We obtain earnings restatement data from Eikon (TR.EarningsRestatement) and we code RESTATEMENT = 1 if a firm has a material earnings restatement, and 0 otherwise. We also download auditor opinions from Eikon (TR.BSAuditorOpinionCode). We code GCOPINION = 1 if a firm receives an audit opinion different from an unqualified opinion, and 0 if the firm receives an unqualified opinion. We obtain audit fees from Thomson Reuters Eikon, which contains data on audit fees, auditrelated fees, and non-audit fees. Audit fees (TR.AuditFees) includes amounts payed for the provision of the audit of financial statements in accordance with the accounting standards of the public company accounting. Audit-related fees comprise services related to the main mission of audit of the financial statements. Finally, non-audit fees are fees paid for other services (e.g., consulting services, tax services, transaction services). Based on the description provided in Eikon, 7 We do not use discretionary accruals to measure audit quality because this measure is relatively noisy. 13

14 our measure for financial audit fees is the natural logarithm of audit fees (Paterson and Valencia, 2011). 8 Empirical models We adopt a model commonly used in the auditing literature (e.g., DeFond and Zhang, 2014) and estimate OLS Model (1) to test the association between audit quality and the joint provision of CSR assurance and financial audit by the same (Big 4) auditor as follows: AUDITQUALITY t = β 0 + β 1 SAMEAUDIT t + β 2 DIFFERENTAUDIT t + β 3 UCSRREPORTING t where: AUDITQUALITY SAMEAUDIT + β 4 SIZE t + β 5 LOSS t + β 6 LEV t + β 7 SALESG t + β 8 MTB t + β 9 ROA t + β 10 INT t + β 11 CFO t + β 12 SDEAR t + β 13 FIX t + β 14 CLOSELY t + β 15 BONDING t + β 16 STAKE t + Firm FE + Year FE +ɛ (1) = RESTATEMENT or GCOPINION; = dummy variable equal to 1 if a firm issues a CSR report audited by the same assurance provider appointed for the financial statement, 0 otherwise; DIFFERENTAUDIT = dummy variable equal to 1 if a firm issues a CSR report audited by a different assurance provider appointed for the financial statement, 0 otherwise; UCSRREPORTING = dummy variable equal to 1 if a firm issues a non-audited CSR report, 0 SIZE otherwise; = natural logarithm of firm s total assets expressed in US dollars; LOSS = dummy variable equal to 1 if a firm reports a negative net income, 0 LEV SALESG MTB ROA INTA CFO otherwise; = ratio between total debt and book value of equity at fiscal year-end; = change year over year in total value of sales; = ratio between total market capitalization and book value of equity at fiscal year-end; = ratio between net income and total assets at fiscal year-end; = ratio between total value of intangibles and total assets at fiscal year-end; = ratio between net funds from operation and total assets at fiscal year-end; SDEAR = natural logarithm of the standard deviation of net income on the period [t; t- 4]; 8 In Section 6 Additional analyses, we replicate our analysis by replacing audit fees with the sum of audit fees, audit related fees, and non-audit fees. Despite a reduction in sample size due to missing data unavailability, results are consistent with our main inferences. 14

15 FIX = ratio between total property plant and equipment (net) and total assets at fiscal year-end; CLOSELY = represents shares held by insiders (e.g., managers, employees) divided total common shares outstanding; BONDING = governance score provided by Asset 4; STAKE = the average of social and environmental performance scores provided by Asset 4; Prior auditing literature suggest the inclusion of several control variables associated with audit quality (Becker et al., 1998; Ashbaugh et al., 2003; Myers et al., 2003; Butler et al., 2004; Menon and Williams, 2004; Gul et al., 2009; Lennox and Li, 2012; Prawitt et al., 2012; DeFond and Zhang, 2014). Specifically, we control for client size (SIZE), the extent to which the business model depends on intangible assets (INTA), the risk of bankruptcy (LOSS, LEV), sales growth (SALESG), growth options (MTB), liquidity (CFO), business risk (SDEAR), tangibility (FIX), ownership structure (CLOSELY), and firm performance (ROA). 9 We also control for the firm level bonding using the strength of corporate governance (BONDING) and stakeholder orientation using social and environmental scores (STAKE). One particular concern in our setting is that higher quality firms, i.e., firms with better innate characteristics, may be more likely to retain their Big 4 financial auditor to also provide assurance for their CSR report. These better firms would also be more likely to exhibit better audit quality. If this is the case then we would not be able to attribute lower restatements and more frequent going concern opinions to the joint provision of financial audit and CSR assurance by the same auditor. To mitigate this concern and reduce difficulties coming from omitted variables, we include firms fixed effect, which control for time-invariant firm s characteristics. This allows to reduce, although not eliminate, the concern related to omitted variables such as firm quality. We also include year fixed effect. Consistent with our research design, to test H1, we examine the difference between coefficients β 1 and β 2 in Model (1). If Big 4 auditors also providing CSR report assurance services to their clients benefit from knowledge spillovers from the CSR assurance team and deliver superior audit quality we expect to find that β 1 < β 2 for RESTATEMENT and β 1 > β 2 for GCOPINION. Next, we test the association between audit fees and the joint provision of CSR assurance and financial audit by estimating the following OLS Model (2): 9 Untabulated analysis including additional controls for prior year restatement and going concern opinion produces similar results. 15

16 AUDITFEES = β 0 + β 1 SAMEAUDIT + β 2 DIFFERENTAUDIT + β 3 UCSRREPORTING + β 4 SIZE + β 5 LOSS + β 6 LEV + β 7 SALESG + β 8 MTB + β 9 ROA + β 10 INTA + β 11 CFO t + β 12 SDEAR t + β 13 FIX t + β 14 CLOSELY t + β 15 BONDING t + β 16 STAKE t + Firm FE + Year FE +ɛ (2) where: AUDITFEES = Natural logarithm of audit fees; All the other variables are as defined earlier. In Model (2), we include the same firm-specific factors as in Model (1) that are likely to explain the amount of audit fees. Firm size (SIZE) is an important determinant of audit fees as well as other measures that capture audit risks (LEV, LOSS, SALESG) and complexity of the audit (SDEAR, MTB, INTA) and firm performance (ROA, CFO). To test H2, we examine the difference between β 1 and β 2 in Model (2). If Big 4 auditors also providing CSR report assurance services benefit from knowledge spillovers from the CSR assurance team and conduct more efficient financial audits we expect to find that β 1 < β 2 or β 1 is no different from β 2. Alternatively, if we expect that auditors are able to generate higher audit quality (less frequent restatements and be more likely to issue a going concern opinion) at no greater cost to their clients, we expect that β 1 is not statistically different from β 2. If we find higher audit quality in Model (1) and find in Model (2) that β 1 > β2 one could argue that auditor performing both financial audit and CSR assurance receive additional compensation for their work which could indicate that higher auditor effort drives higher audit quality. Overall, consistent the knowledge spillover argument suggests that β 2 is either lower or no different than β 1. To examine the influence of economic bonding and knowledge spillovers and between the CSR assurance team and the audit engagement team (H3 and H4) we estimate Model (1) and Model (2) within groups based on firm or country levels of economic bonding and stakeholder orientation. First, we perform a cross-sectional analysis for firm-level of bonding using the industry-year median governance score from Eikon (TR.GovScore). We reason that economic bonding is more likely to occur in the low governance group than in the high governance group. Therefore, according to H3 we expect to find more limited effect on audit quality of the joint provision of CSR assurance and financial audit. We also use the Brown et al. (2014) measure of audit environment and perform a similar cross sectional analysis to investigate the moderating effect of country-level variation in audit quality on the relation between the joint provision of CSR assurance and financial audit and audit quality. 16

17 Second, we divide the sample into two groups: high stakeholder orientation and low stakeholder orientation. We reason that knowledge spillovers from CSR report assurance teams to financial audit engagement teams is more likely to occur for stakeholder-oriented firms for which CSR risks have more implications for financial statements relative to firms that are less stakeholderoriented. We expect to find that firms using their financial auditor to assure their CSR report that are more stakeholder-oriented (High Stakeholder) exhibit higher audit quality relative to firms that are less stakeholder oriented (Low Stakeholder). We also expect that knowledge spillovers are more likely to be more important in high stakeholder-oriented firms which could lead to more efficient audits and lower audit fees for auditors performing jointly the financial audit and the CSR assurance. This is less likely to occur in low stakeholder-oriented firms. We measure stakeholder orientation with the average of social score and environmental scores from Eikon (TR.SocScore and TR.EnvScore) and allocate firms into High Stakeholder and Low Stakeholder groups based on the industry-year median value of the social and environmental score. According to H4, we expect the difference between β 1 and β 2 in Model (1) to be larger for firms that are more stakeholder oriented relative to firms that are less stakeholder oriented. In Model (2), we expect β 1 < β 2 in high stakeholder oriented firms but not in low stakeholder-oriented firms. We also examine the effect of a country s level of stakeholder orientation by its citizens percentage preference between the two following statements in the World Values Surveys: (1) Protecting the environment should be given priority, even if it causes slower economic growth and some loss of jobs; (2) Economic growth and creating jobs should be the top priority, even if the environment suffers to some extent. The higher the values, the more preference for protecting the environment and, therefore, the more stakeholder oriented is a country. Conversely, we consider a country oriented to economic growth, at the expense of environment, to be more shareholder oriented. We repeat our cross-sectional analysis using this country measure of stakeholder orientation. 5. Findings Table 4 presents the descriptive statistics of our sample. Panel A shows the univariate descriptive statistics for the full sample, Panel B shows descriptive statistics per type of CSR information provided (no CSR report, unassured CSR report, assured CSR report by a different assurance provider and assured CSR report by the same audit firm). Panel C presents the correlation between the variables. [Insert Table 4 About Here] 17

18 Table 5 provides the results of our main test. We find that firms for which the auditor performs both the financial audit and provides a CSR assurance are associated with less frequent restatements (see test at the bottom of Table 5 indicating that β 1 < β 2, p-value < 0.01) and financial auditors are more likely to issue going concern opinions (see test at the bottom of Table 5 indicating that β 1 > β 2, p-value < 0.01). This indicate that firms that have the same financial auditor for their financial audit are associated with higher audit quality. Importantly, we find that this higher audit quality is not driven by higher audit effort, which would result in higher audit fees. Firms for which the financial auditor performs financial audit and CSR assurance do not exhibit higher audit fees than firms that use a different CSR assurance provider (see test at the bottom of Table 5 indicating that β 1 is not statistically different from β 2, p-value > 0.10). We find that the joint provision of financial audit and CSR assurance by the same audit firm lead to higher audit quality at no greater cost to the client. [Insert Table 5 About Here] In Table 6, we examine cross-sectional variation of the relation between the joint provision of financial audit and CSR assurance, and audit quality and audit fees based on the likelihood of firmlevel economic bonding between a firm and its auditor (proxied with corporate governance) and the extent of potential knowledge spillovers (proxied with the extent of stakeholder engagement). Panel A of Table 6 shows that in the low bonding group firms are less likely to restate earnings if their auditor also assure CSR reports (see test at the bottom of Table 6 for RESTATEMENT and Low BONDING indicating that β 1 < β 2, p-value < 0.01) whereas we do not find that this is the case in the high bonding group (see test at the bottom of Table 6 for RESTATEMENT and High BONDING indicating that β 1 is not different from β 2, p-value > 0.10). We also find that auditors are more likely to issue going concern opinions if the auditor performs both the financial audit and provide assurance on the CSR report in the low Bonding group (see test at the bottom of Table 6 for GCOPINION and Low BONDING indicating that β 1 > β 2, p-value < 0.01). We also find that firms having their auditor doing both the financial audit and providing the CSR assurance charge lower audit fees in the low bonding group (see test at the bottom of Table 6 for AUDITFEES and Low BONDING indicating that β 1 < β 2, p-value < 0.05) but not in the high bonding group (see test at the bottom of Table 6 for AUDITFEES and High BONDING indicating that β 1 is not different from β 2, p-value > 0.10). Panel B of Table 6 shows that higher audit quality for firms that use the same auditor for their financial statements and their CSR assurance is stronger for firms that are more stakeholder oriented. In high stakeholder-oriented firms, firms using the same auditor for their financial statements and CSR assurance are less likely to restate earnings (see test at the bottom of Table 7 18