Impact of Auditor and Audit Committee Report Changes on Audit Quality and Costs: Evidence from the United Kingdom*

Transcription

1 Impact of Auditor and Audit Committee Report Changes on Audit Quality and Costs: Evidence from the United Kingdom* Lauren C. Reid University of Pittsburgh Joseph V. Carcello University of Tennessee Chan Li University of Pittsburgh Terry L. Neal University of Tennessee September 2017 * We greatly appreciate the valuable insights provided by Marek Grabowksi of the United Kingdom s Financial Reporting Council. We also value comments and suggestions from Lee Biggerstaff, Josh Gunn, Jonathan Shipman, Quinn Swanquist, and Rob Whited. In addition, we thank Jillian Damus, Steven Hawkins, Justin Short, Kristin Stack, Jingyu Xu, and Yue Zhang for their research assistance.

2 Impact of Auditor and Audit Committee Report Changes on Audit Quality and Costs: Evidence from the United Kingdom ABSTRACT: While substantial revisions to auditor and audit committee reporting are being discussed internationally, including in the United States, the impact of these reforms on audit quality is unknown. We exploit the United Kingdom s recent auditor and audit committee reporting changes and find that the U.K. s new reporting regime is associated with an improvement in audit quality as proxied by significant decreases in absolute abnormal accruals and the propensity to just meet or beat analyst forecasts, and a significant increase in earnings response coefficients. As for costs, we do not find a significant change in audit fees or audit delay surrounding the implementation of the new reporting regime. Taken together, the results of this study suggest that new auditor and audit committee reporting requirements are associated with a significant improvement in audit quality without detecting a significant increase in costs. Keywords: audit report changes; audit committee report changes; PCAOB; SEC; United Kingdom; audit quality; audit costs. Data Availability: Data are available from public sources identified in the text.

3 1. Introduction The United Kingdom adopted standards on an expanded audit report even earlier, and so we now benefit from a growing body of evidence and experience (Doty 2015a). Recently, United States and international regulators have sought to improve transparency around the audit process by implementing or proposing significant changes to auditor and audit committee reporting. In the United States, for example, the Public Company Accounting Oversight Board (PCAOB) enacted a new auditor reporting standard on June 1, 2017 that requires the auditor to discuss critical audit matters ( CAMs ), auditor tenure, and audit firm independence in the audit report (PCAOB 2017). 1 The U.S. Securities and Exchange Commission (SEC) issued a concept release in 2015 with the goal of improving the audit committee report and in turn, the work performed by audit committees (Heller 2014; SEC 2015; White 2014; Whitehouse 2014). The effects of these reporting changes on audit quality and costs, however, are unknown. Consistent with PCAOB Chairman Doty s opening quote, we examine the quality and cost effects of the recent reporting changes in the United Kingdom (U.K.). Proponents believe that reporting changes will improve audit quality, via a threat of disclosure that provides auditors more bargaining power over management, or by increasing the accountability and/or professional skepticism of the auditor and audit committee (IOSCO 2009; PCAOB 2011; IAASB 2013a; Peecher et al. 2013). Opponents, however, argue that the reforms may not increase audit quality because auditors and audit committees are simply required to report what they have always done (BDO 2013; KPMG 2013). Therefore, whether audit quality increases due to auditor and audit committee reporting changes is an empirical question. This study exploits the exogenous shock 1 Communication of CAMs will be required in audit reports for large accelerated filers with fiscal years ending on or after June 30, Please see detailed discussion on the timeline of implementation in Section 2. 1

4 of the U.K. s regulatory changes to examine whether audit quality is improved by significant revisions to auditor and audit committee reporting requirements. Specifically, we examine the audit quality effects of the Financial Reporting Council s (FRC) changes to the U.K. Corporate Governance Code, its Guidance for Audit Committees, and the International Standards on Auditing (U.K. and Ireland), which became effective for financial years ending on or after September 30, These regulatory revisions require the audit committee report to include significant financial statement issues considered by the committee as well as a discussion of how these issues were addressed (FRC 2012). 2 Furthermore, the FRC s revisions to the International Standard on Auditing 700 (U.K. and Ireland) require auditors to include a discussion of the following in their reports: (1) assessed risks of material misstatement, (2) the materiality threshold used in the engagement, and (3) the scope of the audit (FRC 2013a). We investigate the impact of these reporting reforms on audit quality using three common proxies: (1) absolute abnormal accruals, (2) the propensity to just meet or beat analyst forecasts and (3) earnings response coefficients (ERCs). 3 In addition, given the importance of assessing costs of regulatory revisions and particularly due to the concern that costs will increase as a result of the new reporting requirements (Chalmers 2013; Overend 2013), we examine changes in audit fees and audit delay (i.e., the time between a company s fiscal year-end and the audit report date) surrounding the implementation of the new reports. Using a balanced sample of companies required to comply with the regulatory changes, 2 The report is also required to include a discussion of the evaluation of the external audit and information on how the external auditor is appointed or reappointed. Additionally, the audit committee report must disclose audit firm tenure and the timing of the most recent audit tendering (FRC 2012). 3 As discussed in DeFond and Zhang (2014), audit quality is part of financial reporting quality. Financial reporting quality is also determined by the companies financial reporting systems and innate characteristics. Research often uses the same measures to proxy for both financial reporting quality and audit quality. Given (1) the inherent difficulty in disentangling audit quality and financial reporting quality and (2) the fact that we are studying a change in auditor reporting, we use the term audit quality throughout the paper, but we acknowledge that such change could affect the pre-audited financial reporting quality as well. 2

5 namely firms with a Premium listing of equity shares on the London Stock Exchange, we find evidence of improvements in audit quality. Specifically, both abnormal accruals and the propensity to just meet or beat analyst forecasts significantly decreased from the two years prior to the first two years after the expanded auditor and audit committee disclosures were required. These results suggest that the new reporting regime is associated with a decrease in opportunistic earnings management. As higher quality audits constrain earnings management, audit quality appears to have improved surrounding the introduction of the enhanced auditor and audit committee reporting requirements. Additionally, we find evidence of higher earnings response coefficients around the new reporting regime, which suggests that investors perceive audit quality to be higher after the implementation of the new auditor and audit committee reports. To minimize the threat of contemporaneous events that may confound the analysis, we re-perform the analysis using two control groups: (1) European companies and (2) U.S. companies. We find that U.K. Premium firms, relative to European firms as well as U.S. firms, experienced significantly greater improvements in audit quality from the two years prior to the reporting changes to the first two years after the new reporting requirements. These tests assist in alleviating the concern that the results are driven by other European-specific or worldwide events. To investigate the costs associated with the new reporting regime, we examine whether the additional auditor and audit committee disclosure requirements are associated with changes in audit fees and/or audit delay. We do not detect significant changes in audit fees or the reporting lag as a result of the new reporting requirements, suggesting that the effort and time to generate auditor reports did not significantly change from the pre-period to the post-period. It therefore appears that audit costs did not significantly change in response to the enhanced auditor 3

6 and audit committee reporting regime. However, since the lack of evidence of an association is not necessarily evidence of a lack of association, we cannot conclusively state that there are no significant audit costs associated with the new reporting requirements (DeFond 2010). Overall, this study reveals the benefits associated with the new auditor and audit committee reporting regime in the United Kingdom. Specifically, opportunistic earnings management decreased, which suggests that audit quality improved surrounding the implementation of the new U.K. reporting requirements. We also find that investor perceptions of audit quality improved after the adoption of the enhanced auditor and audit committee reports. In our analysis of associated costs, we fail to find evidence that audit fees and audit delay significantly changed as a result of the enhanced reporting regime. While we cannot conclusively state that costs did not increase, it appears that the reporting requirements may offer audit quality benefits without generating significant costs. This is consistent with the argument that the increase in quality is not due to the performance of additional audit procedures, but instead the quality improvement may be due to (1) the threat of disclosure that gives auditors more leverage over management or (2) the enhanced transparency via actual disclosures that increases the auditor/audit committee s accountability and professional skepticism around the additional disclosed information. In additional analyses, we attempt to tease out whether it is the actual disclosure in the new reports or the threat of disclosure from the new reporting requirements that is driving the documented improvement in audit quality. If the actual disclosure impacts audit quality rather than the threat of disclosure, we would expect the number of risks to have an additional impact on audit quality beyond the increase previously documented. The results show no evidence that the actual number of risks disclosed impacted audit quality above and beyond the increase 4

7 associated with the overall reporting regime change. The lack of evidence, while not conclusive, is consistent with the argument that the improvement in audit quality may be a result of the threat of disclosure rather than the actual disclosure in the new audit reports. One concern of the study is that other contemporaneous changes in the U.K. could affect our results. 4 To alleviate concerns of confounding events in our analysis, we examine whether other changes occurring during the same timeframe that only apply to U.K. Premium listed companies could impact the results of this study. Two such changes include (1) the replacement of the Business Review section of the annual report with a Strategic Report and (2) the requirement of the board of directors to state that the annual report is fair, balanced and understandable (Deloitte 2013; KPMG 2014). In summary, we generally fail to find evidence of differences in audit quality due to changes in the length and coverage of the company s Strategic Report/Business Review. We also did not find evidence that the improvement in audit quality around the new audit report changes was different for companies that included the fair, balanced and understandable (FBU) statement in the prior year annual report and those that did not. These analyses therefore provide comfort that the Strategic Report and FBU statement requirements generally did not impact the results of our study. 5 Given the importance of this regulatory change, it is not surprising that there are other working papers that examine the U.K. setting. Several studies examine the market reaction to the new reporting regime. For instance, Lennox et al. (2017) focus on long-window equity valuation models and find that the risk disclosures reflect uncertainty in accounting estimates, but they do not find significant incremental information content. Guiterrez et al. (2017) investigate short- 4 We also conduct tests using a strict change analysis to account for temporal changes within firms that are not associated with the new reporting regime, and continue to find that the new auditor and audit committee reporting requirements are associated with significant audit quality improvements. 5 We provide more detail of our testing related to confounding events in Section 5, Additional Analyses. 5

8 term windows and do not report an investor reaction to the new U.K. reporting regime. Guiterrez et al. (2017) also examine the impact of the new reports on audit quality and audit fees. The authors find some evidence that companies with high numbers of risks and long reports pay higher fees but no evidence of an impact on audit quality as measured by abnormal accruals. There are a couple of important differences to note between our study and Guiterrez et al. (2017). First, our main analysis uses a balanced panel where each firm is used as its own control in addition to employing European and U.S. control groups. 6 Second, we triangulate our audit quality analysis by using two additional measures (the propensity to just meet or beat analyst forecasts and earnings response coefficients) in addition to abnormal accruals. 7 The findings of this study make several important contributions. First, the paper directly examines the effects of the auditor and audit committee reporting changes in the United Kingdom, which is an important market to examine. 8 Second, this study allows the examination of the effects of regulatory changes prior to the implementation of similar requirements in other countries. Specifically, since the United Kingdom and the United States share many regulatory and cultural commonalities, the analysis of the change in the U.K. s reporting requirements creates a rare opportunity to examine the impact of standard changes that are not yet effective in the United States. Third, this paper contributes to the audit quality literature by providing evidence that is consistent with enhanced auditor and audit committee reporting requirements 6 Guitierrez et al. (2017) use firms with different fiscal year-ends as a control group as well as firms listed on the Alternative Investment Market (AIM) market, which is comprised of smaller, growing companies that are significantly different from the large, established companies with a Premium listing of equity. 7 As discussed in DeFond and Zhang (2014), there are four output audit quality measurement categories: material misstatement, auditor communication (such as going concern opinions), financial reporting quality (such as abnormal accruals and meet/beat analyst forecasts), and perception-based quality (such as ERCs). They recommend using several measures across multiple categories to measure audit quality. We are unable to use either material misstatements or going concern opinions because of the rare number of accounting misstatements or going concern opinions in the U.K. See Section 3 for more discussion of our measures. 8 The United Kingdom ranked fourth in market capitalization of all countries with over $3 trillion in market capitalization in 2012 behind the United States, China, and Japan (WorldBank 2014). 6

9 increasing audit quality without a significant increase in audit costs. The remainder of the paper is organized as follows. Section 2 provides background on auditor and audit committee reporting changes and develops hypotheses. Section 3 outlines the research method while Section 4 describes the results. Section 5 describes additional analyses and the final section concludes. 2. Background and Hypotheses Development U.S. Discussion of Reporting Changes The PCAOB issued a new auditing standard on June 1, 2017 to change the audit report with the goal of providing more entity- and engagement-specific information about the audit process (PCAOB 2017). The standard requires the auditor to discuss critical audit matters ( CAMs ), auditor tenure, and audit firm independence in the audit report. CAMs are defined as issues communicated to the audit committee that relate to material financial statement accounts and that involve challenging, subjective, or complex auditor judgment (PCAOB 2017). The standard instructs auditors to consider various factors in identifying CAMs, including the auditor s assessment of the risks of material misstatement, including significant risks (PCAOB 2017, 22). 9 The PCAOB s new rule outlines the following implementation timeline: (1) compliance with the new audit report format, including disclosure of tenure and auditor independence for audits of all companies with fiscal years ending on or after December 15, 2017; (2) communication of CAMs for audits of large accelerated filers with fiscal years ending on or after 9 Changes to the audit report have either occurred or are being contemplated by standard setters from around the world. The IAASB recently adopted a new audit report that requires the disclosure of key audit matters ( KAMs ) similar to the PCAOB s CAMs (IAASB 2015). The European Parliament has also proposed auditor reporting changes, endorsing a draft agreement that would require auditors to provide detailed information on the audit process (Cohn 2014). 7

10 June 30, 2019; and (3) communication of CAMs for audits of all other companies with fiscal years ending on or after December 15, Stakeholders have also voiced the need for additional and more valuable audit committee disclosures (CAQ 2013; CII 2013; EY 2013a, 2013b; NACD 2013; TapestryNetworks 2013). In response to these requests, the SEC issued a concept release in 2015 to investigate possible revisions to the audit committee report (SEC 2015). Potential changes to audit committee disclosures include the provision of information regarding (1) the committee s evaluation of the auditor, (2) communications between the committee and the auditor, (3) the auditor reappointment process and (4) the qualifications of the auditor (SEC 2015). United Kingdom Reporting Regime Changes The FRC made significant changes to the U.K. Corporate Governance Code, its Guidance for Audit Committees, and the International Standards on Auditing (U.K. and Ireland) effective for fiscal years ending on or after September 30, The FRC s modified governance policies instruct audit committees to discuss significant issues considered by the committee and how they addressed these issues, including (and focusing on) the issues communicated to the committee by the auditor (FRC 2012). The FRC s revisions to the International Standard on Auditing (U.K. and Ireland) 700 require auditors to expand their report to include a discussion of material misstatement risks, materiality, and the scope of the audit (FRC 2013a). Specifically, the auditor must disclose the risks that had the greatest effect on the overall audit strategy, the allocation of resources in the audit, and directing the efforts of the engagement team (FRC 2013a). Auditors should also explain how they determined and applied materiality during the audit, and describe how the scope of the audit sufficiently addressed the material misstatement risks they identified (FRC 2013a). Please refer to Appendix 1 for excerpts from an example of the new auditor and 8

11 audit committee reports. 10 Consistent with the quote from Chairman Doty that opened the paper, examining the effects of the United Kingdom s regulatory changes may provide useful feedback to other nations standard setters. The PCAOB has indicated on numerous occasions how closely it is monitoring and considering the effects of the new U.K. auditor reporting regime as the Board develops its own rule (Doty 2015b; PCAOB 2016). For example, PCAOB Chairman Doty stated at the 2015 FEE Audit Conference that I am encouraged by the FRC's March 2015 report on compliance and user benefits in the first year of extended auditor's reports and The FRC experience is useful for the PCAOB's consideration and analysis (Doty 2015b). Moreover, PricewaterhouseCoopers has stated, whilst the new reports may be a few years off for the rest of the world, what those reports will look like is becoming increasingly clear (PwC 2013). By testing the new regime implemented in the United Kingdom, we are able to analyze the potential benefits and costs of these revised standards before comparable requirements are implemented in other regimes. This sort of live field-testing is really important at this point [as] standard setters need robust evidence of the practical implications of what they are proposing (PwC 2013). Our field-testing of the United Kingdom is particularly relevant to the reporting changes made in the United States given the similar levels of disclosure and securities regulation in these two countries as well as their comparable legal environments (Hail and Leuz 2006; La Porta et al. 1997). Additionally, there is overlap between the new reporting regime changes in the United States and the U.K. s standards (PCAOB 2017). Specifically, both standards require the auditor to consider and disclose risks of material misstatement. The FRC s standard, however, includes the disclosure of materiality and scope, which has not been included 10 For the complete audit committee and auditor reports, please refer to pages and , respectively, of the Rolls-Royce 2013 annual report found at Annual %20Report tcm pdf. 9

12 in the U.S. s new standard. Hypotheses Development Proponents of changing the audit report argue that audit quality will increase as a result of new auditor and audit committee disclosure requirements for at least two reasons. First, the threat of disclosure may provide the auditor and audit committee with greater leverage over management, which results in management adopting more acceptable financial reporting approaches and disclosures (PCAOB 2016, 76, 82). In particular, for accounts involving greater estimation risk, the auditor may be able to achieve concessions related to more aggressive management estimates and judgments in exchange for not explicitly highlighting the financial statement area in the report (i.e., because once the estimate is less aggressive the account may no longer represent a heightened area of misstatement risk). In fact, companies recognize that the new audit report will enhance auditors leverage over management (Wells Fargo 2016). In addition, prominent investors also recognize that an expanded auditor report may enhance the auditor s leverage. Paul Haaga, the former chairman of the Capital Group, stated at a PCAOB Roundtable, The mere fact that there s more to say than pass or fail would give the auditors a stronger hand. They would win more arguments, and we think that would be a good thing. And that could be a good thing even if they didn t say anything at all in the emphasis paragraphs. Simply the ability to say something there is an additional tool (our emphasis) (PCAOB 2011). Similarly, the ability of the audit committee to report on financial statement risks, among other items, provides the committee with a greater opportunity to constrain management behavior and improve audit quality. Second, the enhanced transparency via the actual disclosures made in the reports may increase the accountability of auditors and audit committee members to financial statement users 10

13 (FRC 2013b; IAASB 2013b). Peecher et al. (2013) discuss improved audit reports as a mechanism that could be used to enhance accountability, and in turn, audit quality. In particular, the reporting changes may lead to an enhanced focus by audit committees and external auditors on the most significant areas in the financial reporting process, with the objective of reducing financial reporting risk in these disclosed areas. By focusing attention on key areas, audit committees and auditors are likely to perform their procedures around these issues with heightened professional skepticism, among other contributors to audit quality (IAASB 2013a). Based on the above arguments, we test the following hypothesis: H1: Audit quality improved in the United Kingdom after the new reporting regime became effective. We acknowledge that some professionals have argued that revising the audit committee and auditor reporting requirements will not impact audit quality (BDO 2013). Since the only new requirement is for the auditor and audit committee to report what they have done during the year, this may not impact the audit process. If this is the case, we will not see any improvement in audit quality surrounding this regulatory change. As it is important for regulators to also assess the costs associated with new policies, we examine the potential costs that companies face in order to comply with the revised reporting requirements. While we are unable to quantify the costs directly related to the work performed by the audit committee, we are able to investigate if there was an increase in external audit costs by examining both audit fees and audit delay (i.e., the time between a company s fiscal year end and the audit report date). Audit fees as well as audit delay may increase in response to the new requirements for several reasons. First, the inclusion of more detail about the audit process could lead to additional review by senior members of the engagement team (Chalmers 2013; Overend 2013). Second, some of the larger audit firms have noted that they expect to incur training and 11

14 implementation costs associated with setting up additional quality control processes around the new, more informative and tailored auditors report (Chalmers 2013). Third, the auditors will likely spend additional time discussing audit matters with management and the audit committee. Any additional time and effort expended by the auditors will likely trickle down to the company in the form of audit fees and/or delay in the issuance of the audit report. Thus, our second hypothesis is: H2: Audit costs increased in the United Kingdom after the new reporting regime became effective. The prediction of H2 also has tension because it is possible that audit costs will not significantly increase as a result of the new requirements. For one, if the revised regulations do not substantially increase the amount of work performed by the auditors, we are unlikely to find a significant change in audit fees and audit delay. Others have argued that most of the information contained in the report is not new information, and can be extracted from the summary memo the auditor already prepares for the audit committee. As PCAOB Board Member Steve Harris questioned, since all that investors are asking for is what auditors already know, why can t this be done easily and cost effectively? (PCAOB 2011). If this is the case, then audit costs are unlikely to significantly change. 3. Methodology Audit Quality In order to capture audit quality, three proxies are employed: (1) absolute abnormal accruals (ABS_ACC), (2) the propensity to just meet or beat analyst forecasts (MEET), and (3) earnings response coefficients (ERCs). The first two measures are designed to capture opportunistic earnings management, which increases as abnormal accruals and the propensity to 12

15 just meet or beat analyst forecasts increase. Higher quality audits constrain earnings management. Therefore, improvements in audit quality would be evidenced by lower abnormal accruals and a lower propensity to just meet or beat analyst forecasts. The third proxy, ERCs, captures investor perceptions of quality by measuring the degree to which investors respond to unexpected earnings. The greater the investor confidence in the earnings reported by management, the larger the stock price reaction to unexpected earnings. Thus, higher perceived audit quality would be indicated by a higher earnings response coefficient. 11 We examine the relation between the new reporting regime and absolute abnormal accruals using the following model: ABS_ACC t = β 0 + β 1 POST + β 2 SIZE t + β 3 ROA t + β 4 LOSS t + β 5 MB t + β 6 LEVERAGE t + β 7 PRIOR_ACC t + β 8 CFO t + β 9 VOLATILITY t + β 10 BIG4 t + IND_FE + ε it (1) The dependent variable in this regression is performance-adjusted absolute abnormal accruals (ABS_ACC), which is estimated using the modified Jones (1991) approach within two-digit ICB industry groups and we require a minimum of 15 observations per two-digit ICB industry (Dechow et al. 1995; Kothari et al. 2005; Carcello and Li 2013). 12 Our variable of interest, POST, equals one if the fiscal year is the first two years of the new reporting regime and zero otherwise. As described above, we predict β 1 will be negative and significant as we expect the 11 We are unable to use material misstatements as an audit quality measure because misstatements unrelated to accounting standard changes are very rare in the U.K. We consulted with senior officials at the FRC to confirm this and they reported only nine material misstatements in their financial statement review of 252 companies in the fiscal year At this misstatement rate of 3.57 percent, our largest sample of 326 companies would only have 12 misstatements. We therefore do not hand collect misstatement data. We however did hand collect audit opinion data and found that only two companies in our sample received going concern opinions (we also note that these two companies received going concern opinions in both t and t-1). We therefore are also unable to use going concern opinions as an audit quality measure. 12 We calculate total accruals (TA) using the following equation: TA = (Change in Current Assets Change in Cash Change in Current Liabilities + Change in Current Debt Depreciation) / Lag of Total Assets. We then estimate the following regression: TA = β 0 + β 1 (1/Lag of Total Assets) + β 2 (Change in Revenue/Lag of Total Assets Change in Receivables/Lag of Total Assets) + β3(ppe/lag of Total Assets) + ε. We then match each firm-year observation by two-digit industry code, year, and closest ROA. ABS_ACC equals the absolute value of the residuals from the second regression minus the matched abnormal accrual (in order to adjust for performance). As discussed later in the sample selection, we delete financial firms and utilities firms from the accruals sample. We also winsorize all continuous variables at the 1 st and 99 th percentile when calculating abnormal accruals. 13

16 new reporting requirements to improve audit quality. In line with prior studies (Ashbaugh et al. 2003; Carcello and Li 2013), we control for various firm-level characteristics that have been shown to impact abnormal accruals. These control variables include total assets (SIZE), profitability (ROA and LOSS), market-to-book ratio (MB), leverage (LEVERAGE), the prior year s accruals (PRIOR_ACC), cash flow from operations (CFO), sales volatility (VOLATILITY), and the use of a Big 4 auditor (BIG4). Finally, we include industry fixed effects to account for differences in abnormal accruals across industries (IND_FE). As an additional test of the association between audit quality and the new reporting regime, we use the propensity to just meet or beat analyst forecasts (MEET) as a proxy of quality and estimate the following logistic model: MEET t = β 0 + β 1 POST + β 2 LN_MVE t + β 3 ROA t + β 4 LOSS t + β 5 MB t + β 6 LEVERAGE t + β 7 CFO t + β 8 VOLATILITY t + β 9 BIG4 t + β 10 NUM_ANALYST + β 11 DISP t + β 12 HORIZON t + IND_FE + ε it (2) The dependent variable in this regression, MEET, equals one if the difference between the firm s annual earnings per share and the most recent mean consensus analyst earnings forecast is greater than zero and less than or equal to one cent (Reichelt and Wang 2010). Once again, our variable of interest is POST and we predict its coefficient will be negative and significant. Following prior literature (Davis et al. 2009; Prawitt et al. 2009; Reichelt and Wang 2010), we control for the market value of the firm (LN_MVE), profitability (ROA and LOSS), market-tobook ratio (MB), leverage (LEVERAGE), cash flow from operations (CFO), sales volatility (VOLATILITY), the use of a Big 4 auditor (BIG4), analyst coverage (NUM_ANALYST), analyst forecast dispersion (DISP), and the lag between the last analyst forecasting date before the earnings announcement and the earnings announcement date (HORIZON). We also include industry fixed effects (IND_FE). 14

17 In order to provide triangulated evidence, we also use a perception-based quality measure earnings response coefficients (DeFond and Zhang 2014). We estimate the following regression model: CAR t = β 0 + β 1 POST + β 2 UE + β 3 POST x UE + β 4 LN_MVE t + β 5 ROA t + β 6 LOSS t +β 7 MB t + β 8 LEVERAGE t + β 9 CFO t + β 10 VOLATILITY t + β 11 DISP t + β 12 HORIZON t + β 13 BIG4 t + β 14 LN_MVE t x UE + β 15 ROA t x UE + β 16 LOSS t x UE +β 17 MB t x UE + β 18 LEVERAGE t x UE + β 19 CFO t x UE + β 20 VOLATILITY t x UE + β 21 DISP t x UE + β 22 HORIZON t x UE + β 23 BIG4 t x UE + IND_FE + ε it (3) The dependent variable in this regression, CAR, equals the cumulative market adjusted return for the two days beginning on the company s earnings announcement date. Unexpected earnings (UE) is calculated as the difference between the actual earnings per share of the firm and the mean consensus forecast scaled by the stock price (Baber et al. 2014). The interaction of UE and POST measures the change in the earnings response coefficient from the prior reporting regime to the new reporting regime. Since we predict that investor perceptions of audit quality will improve around the implementation of the new reporting regime, we expect β 3 to be positive and significant. We control for the market value of the firm (LN_MVE), profitability (ROA and LOSS), market-to-book ratio (MB), leverage (LEVERAGE), cash flow from operations (CFO), sales volatility (VOLATILITY), analyst forecast dispersion (DISP), the lag between the last analyst forecast date before the earnings announcement and the earnings announcement date (HORIZON), and the use of a Big 4 auditor (BIG4). We also interact each of our control variables with UE and include industry fixed effects (IND_FE). Audit Costs To study the costs associated with the new auditor and audit committee reporting standards, we examine (1) audit fees and (2) audit delay. We first estimate the following audit fee model: 15

18 LN_FEE t = β 0 + β 1 POST + β 2 SIZE t + β 3 ROA t + β 4 LOSS t + β 5 MB t + β 6 LEVERAGE t + β 7 CFO t + β 8 VOLATILITY t + β 9 BIG4 t + β 10 INV t + β 11 REC t + β 12 BUSY t + IND_FE+ ε it (4) The dependent variable, LN_FEE, represents the natural logarithm of audit fees. As described earlier, we expect an increase in audit costs, thus we predict that the sign of β 1 will be positive. Following prior literature (e.g., Simunic 1980; Carcello and Li 2013), we control for firm total assets (SIZE), profitability (ROA and LOSS), market-to-book ratio (MB), leverage (LEVERAGE), cash flow from operations (CFO), sales volatility (VOLATILITY), the use of a Big 4 auditor (BIG4), inventory and receivables intensity (INV and REC), and auditor busy season (BUSY). As in models (1) (3), we also include industry fixed effects to capture differences in audit fees across industries (IND_FE). Costs of the new regulatory regime may also come in the form of a longer delay between a firm s fiscal year end and the issuance of the audit report. Therefore, as an additional test of H2, we estimate the following audit delay model: DELAY t = β 0 + β 1 POST + β 2 SIZE t + β 3 ROA t + β 4 LOSS t + β 5 MB t + β 6 LEVERAGE t + β 7 CFO t + β 8 VOLATILITY t + β 9 BIG4 t + β 10 INV t + β 11 REC t + β 12 BUSY t + β 13 LN_FEE t + IND_FE+ ε it (5) DELAY equals the number of calendar days between a firm s fiscal year end and the date of its audit report (Ashton et al. 1987; Ettredge et al. 2006; Abbott et al. 2012). Given that our dependent variable is an over-dispersed count variable, we use a negative binomial regression to estimate our delay model. 13 In addition to the control variables used in model (4), we control for audit fees (LN_FEE) as in Ettredge et al. (2006). A complete list of variable definitions for all the models is provided in Appendix 2. Sample 13 Negative-binomial models are appropriate when the variance of the dependent variable is much larger than the mean value as is the case with DELAY. Our inferences are unchanged when a standard OLS model is used instead. 16

19 Only entities with a Premium listing of equity shares on the London Stock Exchange are required to comply with the auditor and audit committee reporting changes in the United Kingdom (FRC 2012, 2013a). We obtain data related to the first two years of implementation (t and t+1) as well as the prior period (t-1 and t-2) for each firm. 14 Based on the London Stock Exchange s record of listings, there are 4,670 firm-years associated with a Premium listing. Of these, however, 1,652 firm-year observations are associated with investment funds. Due to the unique nature of investments funds, we exclude these listings and only test commercial equity firm-year observations (N=3,018). In addition, we exclude 1,194 firm-years for companies that did not comply with the new reporting regime in the post-period. 15 We also eliminate 374 firmyears related to other financial and utility companies from our abnormal accrual analysis as is common in prior studies using this measure. <Insert Table 1 Here> Panels A, B, C, D, and E of Table 1 provide detail regarding the sample construction for the abnormal accrual, meet or beat, ERC, audit fee, and audit delay analyses, respectively. Datastream is used to collect the financial information for the Premium commercial equity firms including analyst forecast data, which is available from I/B/E/S on Datastream. Reliable audit fee data for U.K. companies is unavailable electronically. We therefore hand collect this information directly from the annual reports of the sample firms. Each sample is generated using the following process: (1) exclude firm-years missing data necessary to compute the dependent variable in the analysis, (2) delete firm-year observations without the data required to calculate 14 We are aware of only three companies that voluntarily adopted the audit report requirements in year t-1: Vodafone Group, British Sky Broadcasting Group PLC, and Ashmore Group PLC. In untabulated tests, we exclude these firms from the analysis and the results remain unchanged. Please see Section 5 for a discussion of voluntary adoption of the audit committee report requirements in year t We checked compliance by examining the audit committee and auditor reports in each company s annual report in the post-period. It appears that some of these companies may have changed their type of listing (i.e., no longer Premium listing on the main market but instead listed on the AIM), experienced a merger or acquisition during the sample period, or delisted during the post-period. 17

20 control variables, (3) exclude observations for firms headquartered outside of the United Kingdom, and (4) exclude observations missing complete data for both the pre and post periods. 16 This final data requirement allows the comparison of firms in the pre-period to the same firms in the post-period. The use of this balanced panel design reduces the threat of firmlevel correlated omitted variables (Doyle and Magilke 2013). The final sample used in the abnormal accrual, meet or beat, ERC, audit fee, and audit delay analyses are comprised of 1,088, 888, 884, 1,304, and 884 firm-year observations, respectively. Control Groups Even though a balanced panel accounts for time-invariant firm-specific characteristics that may confound the analysis, it is possible that other events occurring around the same time as the new reporting standards could impact the results. We therefore use two control groups: other European companies and U.S. companies. 17 The use of other European companies alleviates concerns that the results may be driven by a correlated omitted variable related to Europeanspecific or global events outside of the reporting revisions, particularly given the use of international accounting and auditing standards by the U.K. and other European countries. Using U.S. firms as a control group also mitigates the concern that contemporaneous global changes might be correlated omitted variables. Furthermore, since we are unable to gather audit fee and audit delay data for the other European companies due to the lack of accurate downloadable data and the difficulty of accurately hand collecting this data from filings in foreign languages, U.S. firms serve as our control group in the cost analyses. 16 We exclude observations for firms headquartered outside of the U.K. in order to ensure that our analysis is performed on a fairly homogenous set of firms that are affected by similar economic factors. Our inferences remain unchanged if these observations are included in the analysis. 17 We also considered using companies with a standard listing of equity on the London Stock Exchange. However, most of the standard share listings are related to the same companies as premium share listings because the standard shares can be used as a different class of stock for the same company. As a result, there are only 16, 8, and 8 firmyear observations related to standard listed companies (that do not also have a premium listing) with balanced data for the accrual, meet/beat, and ERC analyses, respectively. 18

21 We gather European data from Datastream and U.S. data from Compustat, Audit Analytics, and I/B/E/S. We then match our U.K. sample firms with firms in each control group based on year, industry, size, and return on assets. 18 Using these matched samples, we re-perform the main tests including an indicator for U.K. firms (UK) and interacting UK with POST. 4. Results Descriptive Statistics Panel A of Table 2 provides descriptive statistics of our firm-year observations. All continuous variables are winsorized at the 1 st percent and 99 th percent level. Panels B F compare the means of the variables used in each audit quality and audit cost analysis for the preperiod (POST=0) and the post-period (POST=1). Panel B reports that the absolute value of mean abnormal accruals is significantly lower in the post-period at compared to the pre-period at (p<0.01). This univariate result provides preliminary evidence that abnormal accruals decreased upon the implementation of additional reporting requirements for auditors and audit committees. Panel B also reveals that firms have a lower return on assets, greater incidence of experiencing a loss, higher market-to-book ratio, lower prior accruals, and lower sales volatility in the post-period compared to the pre-period. These differences indicate the importance of performing a multivariate regression analysis. 19 Panel C reports that the firm s propensity to just meet or beat analyst forecasts significantly decreased from to from the pre-period to the post-period (p<0.01), 18 We run a first-stage model with UK as the dependent variable, with year, two-digit ICB industry, firm size and ROA as independent variables. Then, we match the U.K. firms with the European (U.S.) firms based on the closest predicted value from the first-stage model. We compare the size and ROA between U.K. firms and matched European (U.S.) firms and find there is no significant difference between the two groups. We obtain similar results when using unmatched samples. 19 We also re-perform all of our analyses using just the year prior and year after the reporting change (i.e., t-1 and t) and do not find any differences in our control variables between the two years. Furthermore, our overall inferences do not change using this shortened timeframe. 19

22 providing initial evidence of a lower propensity to just meet or beat analyst forecasts surrounding the introduction of enhanced auditor and audit committee reporting requirements. The observations in the meet or beat analysis also have a higher market value of equity, higher market-to-book ratio, lower sales volatility, lower forecast dispersion, and fewer days between the forecasting date and earnings announcement date in the post-period compared to the preperiod. Panel D reveals that the statistical changes in the ERC sample firms from the pre-period to the post-period are a higher market value of equity, a higher market-to-book ratio, lower sales volatility, lower forecast dispersion, and fewer days between the forecasting date and the earnings announcement date. Taken together, Panels B and C provide some initial evidence that audit quality increased (as proxied by lower abnormal accruals and a decrease in the propensity to just meet or beat analyst forecasts) from the two years prior to the new reporting requirements to the first two years after the implementation. Panels E and F present the univariate results for the audit fee and audit delay analyses, respectively. These panels report that there is not a significant change in audit fees or audit delay from the pre-period to the post-period. The only significant differences between the pre-period and post-period are an increase in loss firms, increase in the market-to-book ratio, and a decrease in sales volatility. Thus, at the univariate level, we fail to find any significant change in audit cost (as proxied by audit fees and audit delay) from the two years before the new reporting requirements to the first two years of implementation. <Insert Table 2 Here> Regression Results Tables 3 and 4 present the regression results for the audit quality and audit cost analyses, 20

23 respectively. The primary model for each analysis is provided in column 1 of each panel. In Table 3, we present the results using the European control group and the U.S. control group in columns 2 and 3, respectively. In Table 4, the results using the U.S. control group are provided in column 2. Abnormal Accrual Analysis Panel A of Table 3 reports the results of model (1). Column 1 shows that the coefficient on POST is negative and significant at the p<0.01 level. This result provides evidence that abnormal accruals significantly decreased in the post-period compared to the pre-period for the U.K. companies required to provide the new disclosures. Thus, it appears that audit quality improved under the new reporting regime. Consistent with prior literature, we also find that large firms are associated with lower abnormal accruals. Columns 2 and 3 report that the coefficient on the interaction of POST and UK is significantly negative (p<0.01 and p<0.05, respectively). This result suggests that the relation between the new reporting requirements and improved audit quality is stronger for Premiumlisted U.K. companies compared to both European and U.S. companies. In fact, there is no evidence suggesting an improvement in audit quality for either European or U.S. companies given the positive and significant coefficient on POST in column 2 and the insignificant coefficient on POST in column 3. Furthermore, the joint test (untabulated) of POST + POST x UK is negative and significant (p<0.01) with both control groups, indicating an improvement in audit quality around the new reporting requirements for U.K. companies. <Insert Table 3> Meet or Beat Analysis Panel B of Table 3 provides the results of the estimation of model (2). Column 1 reports 21

24 that the coefficient on POST is significantly negative at the p<0.01 level, suggesting that the propensity for Premium firms to just meet or beat analyst forecasts is lower in the post-period compared to the pre-period. 20 This finding suggests that audit quality improved under the enhanced auditor and audit committee reporting requirements. Results for control variables suggest that firms with larger market values have a lower propensity to just meet or beat analyst forecasts while firms with greater analyst coverage have a higher propensity to just meet or beat analyst forecasts. Columns 2 and 3 report the results using the European and U.S. control groups and show a negative and significant coefficient on POST UK (p<0.05 and p<0.10, respectively). These findings suggest that, consistent with the accrual analysis in Panel A, the relation between the new reporting requirements and improved audit quality is stronger for Premium-listed U.K. companies compared to European and U.S. companies. In addition, the insignificant coefficients on POST indicate that there was no improvement in audit quality for the control companies. Furthermore, the coefficient (untabulated) of the joint test of POST + POST x UK is negative and significant (p<0.01) in both analyses, which again suggests an increase in quality from the preperiod to the post-period for U.K. firms. ERC Analysis Panel C of Table 3 reports the results of the estimation of model (3). Column 1 reveals a positive and significant coefficient on the interaction of POST UE (p<0.05 level). This result suggests that investors perceive audit quality to be higher in the post-period compared to the preperiod. Columns 2 and 3 reveal that the interaction of POST UE UK is positive and significant (p<0.10 and p<0.05, respectively), suggesting that Premium-listed U.K. firms 20 In untabulated tests, we also focus on the group of firms having analysts forecast errors within +/- $0.01 (i.e., companies that could have managed earnings N=184 firm-years) and the coefficient on POST is negative and significant (p-value<0.01). 22