Management Going Concern Reporting: Impact on Investors and Auditors. Jagan Krishnan Jayanthi Krishnan

Transcription

1 Management Going Concern Reporting: Impact on Investors and Auditors Jagan Krishnan Jayanthi Krishnan Eunju (Ivy) Lee Temple University Department of Accounting Fox School of Business 1801 Liacouras Walk Philadelphia, PA Revised: September 2018 Preliminary version We appreciate the comments of workshop participants at Temple University.

2 Management Going Concern Reporting: Impact on Investors and Auditors ABSTRACT The Financial Accounting Standards Board (FASB) s Accounting Standards Update (ASU) required, effective for fiscal years ending after December 15, 2016, managements to evaluate whether there is substantial doubt about the firm s ability to continue as a going concern, and provide disclosures in financial statement footnotes. Prior to this, information about a firm s going concern status came from its auditor, which issues a going concern modified (AGC) or clean audit opinion on the financial statements. We examine two research questions. First, is the new information provided by management valued by investors? We find that the earnings response coefficients for firms with clean audit opinions (including those that disclose going concern issues), but not for firms with AGCs, increased in the first year of the standard. Second, we examine whether there was a change in auditors reporting strategy. We find, after controlling for changes in client characteristics, that auditors became more conservative in the issuance of AGCs in the first year of the standard. Keywords: ASU , FASB, Going Concern, Audit Opinion 1

3 I. INTRODUCTION Information about a company s future survivability is important for investors and the financial markets. Unless liquidation is imminent, US GAAP requires that corporate financial statements be prepared under the going concern presumption that the company will continue to operate. 1 However, even if liquidation is not imminent, a firm can face going concern uncertainties that would be of interest to investors. Until recently, the firm s managers played a relatively passive role in disclosing such uncertainties. Auditing standards require the firm s independent auditor to evaluate whether there is substantial doubt about its client s going concern ability and, in the event of an affirmative assessment, issue a going concern modified audit opinion (henceforth AGC) (AS 2415, PCAOB 2017). When such an opinion is issued, the SEC requires the firm to disclose the associated financial difficulties. In 2014, the Financial Accounting Standards Board (FASB) issued Accounting Standards Update (ASU) , introducing an important change to firms going concern reporting. Management must now evaluate whether there is substantial doubt about the firm s ability to continue as a going concern, and provide disclosures in financial statement footnotes. We examine the impact of this standard in the first year of adoption. 2 The FASB expects the standard to improve financial reporting quality by reducing diversity in the timing and content of existing footnote disclosures for all entities (emphasis added). Our first research question, motivated by the FASB s expectation, is the following: do investors perceive earnings 1 ASU does not define the concept of going concern presumption. In its 2013 Exposure draft for the standard, the Board defined it as the presumption that an entity will continue to operate such that it will be able to realize its assets and meet its obligations in the ordinary course of business (FASB 2013). 2 ASU applies to public and private entities. We focus on public entities only. 2

4 of companies to be more credible in the first year of adoption of ASU (which included both auditor and management evaluations) compared with the previous year (which included only auditor evaluations)? We then examine a second question: did the new standard change the auditor s behavior in issuing an AGC? Although the FASB s main focus in implementing this standard is on the benefits to users of financial statements, the auditor will need to assess these new client disclosures about going concern, while also making its own AGC decision (PCAOB 2014, Staff Audit Practice Alert No. 13). We posit therefore that auditor behavior can also be affected by the standard. Further motivation for this research question is provided by commenters on the exposure draft that preceded the new standard, many of whom analyzed the potential effect on auditors. Also, the PCAOB is now reviewing and considering revisions to the auditing standards for the going concern opinion, in light of the new FASB accounting standard. The going concern evaluation requires considerable judgement (Carcello, Hermanson, and Huss 1995; DeFond, Raghunandan, and Subramanyam 2002; Knechel and Vanstraelen 2007; DeFond and Zhang 2014). The FASB articulates two sources of benefits to investors from ASU First, management can provide additional information about the entity s going concern status than that provided by auditors, particularly when the auditor does not issue a going concern opinion in the presence of mitigating factors. To the extent that this is new information which investors can use, it can influence investor perceptions about the companies reporting such information. The second benefit can arise for all companies as noted in the FASB quote above. ASU defines management s responsibility to evaluate its going concern status and provides guidance about the evaluation and disclosures. Management has private information about the company operations, its strategy, its negotiation with lenders, and other critical matters, which are relevant to such evaluation. The FASB expects that, because of the 3

5 judgment needed in the GC assessment, firms may need to implement and document underlying processes and controls (FASB 2014, 31). Thus, regardless of the final outcome of the going concern judgments by the auditor, the working of the standard can engender positive investor perceptions of change in firms financial reporting quality for all firms. Under ASU , management performs an active role in the evaluation and public disclosure of going concern issues, compared with its previous passive role of responding behind the scenes to its auditor s inquiries about such issues. 3 Two factors are relevant here. First, such active involvement likely means that management is more rigorous than before in its assessment. Since management has superior private information about the firm, this can potentially provide both investors and auditors with greater confidence than before in their own assessments. For investors, while the auditor s opinion on the firm s going concern is informative, another layer of information provided by management can enhance the information environment. Second, while ASU incorporates many of the disclosure considerations in the auditing standard, it introduces some important differences. These include, as we discuss in more detail later, a formal definition of substantial doubt, a longer forward-looking window than that used by auditors, and the requirement that, unlike auditors, management must disclose relevant information even when the initial substantial doubt is expected to be alleviated by mitigating factors. Investors, who previously knew only the auditor s binary going concern assessment, are now provided more details about the company s going concern status. An example of such additional details is provided in a recent much-publicized Form 10-K filing by 3 In addition, SEC rules require disclosures by management when the auditor issues a going concern opinion. More relevant to our analyses, some managers with no auditor going concern modified opinions also disclosed going-concern-related issues prior to the new standard, in the Management, Discussion, and Analysis (MD&A) (Mayew, Sethuraman, and Venkatachalam 2015) and Risk Factors sections of their 10-Ks, but these disclosures were voluntary in nature. 4

6 Sears for the fiscal year ended January 28, Sears disclosed that conditions suggesting substantial doubt about going concern were mitigated by management plans. This disclosure was accompanied by a clean audit report from Sears auditors, Deloitte (Steele 2017). We hand-collect data on management disclosures for firm-years before and after the effective date of the standard. Not surprisingly, there is considerable overlap between auditors and managements going concern assessments when the auditor issues a going concern opinion. However, there is variation in managements disclosures about going concern for clean firms that have not received going concern opinions from their auditors. In most cases, clean firms have no disclosures relating to going concern (suggesting the absence of any adverse issues), or state explicitly that they have no going concern issues. We expect that this provides a dual assurance about clean firms, the auditor s clean opinion being confirmed by management. This is different from the pre-standard period when disclosures were voluntary and therefore, their absence did not necessarily indicate the absence of going concern problems. Some firms disclose (under ASU ) issues relating to going concern even in the absence of an auditor s going concern report. These provide information about management plans and factors that mitigate initial substantial doubt judgments. We compare investors responsiveness to earnings measured by the response of stock returns to earnings surprises (the earnings response coefficient, ERC) - in the pre-standard year (henceforth the PRE period ) and the post-standard year (the POST period ). We document the following findings. First, there is no change in the ERCs for firms with going concern audit opinions. This is consistent with the conjecture that, due to the SEC disclosure requirements for AGC firms, the ASU disclosures do not provide new information. Second, ERCs for firms with clean audit reports (i.e., no AGC) and with management confirmation of the absence 5

7 of going concern issues, increase in the POST period. We conjecture that this is due to the dual assurance being offered in the POST period, as opposed to the single (auditor) assurance that was offered in the PRE period. Third, we also find, interestingly, an increase in ERC for the clean (audit opinion) firms whose management discloses going concern-related issues in the 10- K. This suggests that investors find the information to be useful in interpreting earnings even though the information indicates that there was an initial assessment about substantial doubt. One potential explanation is the usefulness of information about management s plans. ASU requires management to disclose plans that alleviate the initial substantial doubt about going concern. The FASB asserts that information about management s plans could give financial statement users the opportunity to evaluate the likely success of those plans in mitigating the conditions or events that raised substantial doubt (FASB 2014, BC36). The increase in ERC we note above seems to support this conjecture. In our second research question, we examine whether auditors change their going concern reporting strategy after the adoption of ASU Following Francis and Krishnan (2002) and Geiger, Raghunandan, and Rama (2005), we decompose the change in probability of issuing a going concern opinion from the PRE to POST periods into two components: the change attributable to auditors reporting strategy and change attributable to client risk characteristics. 4 We find an average increase of 1.22% in the probability of auditors issuance of a going concern report. However, we are interested in whether auditors reporting strategy changed from the PRE to POST periods. Decomposition of the overall change in probability into its two components 4 This methodology (i.e., the decomposition of a change in probability into components) is drawn from studies in labor economics (e.g., Farber 1990). 6

8 reveals that there is a 0.16% increase on average due to change in auditors reporting strategy, after controlling for changes in client characteristics. This suggests that auditors became more conservative in issuing a going concern report. 5 Moreover, the number is economically significant, given a going concern rate of 5.5% in the PRE period. Interestingly, this increased conservatism is more marked for firms with no management assertions about going concern issues. We offer two explanations for this increased conservatism: (a) auditors face more scrutiny from regulators and more attention from financial statement users in the first year of adoption of the standard, and (b) auditors have less pressure to avoid issuing a going concern opinion to keep the client when management is also responsible for going concern disclosures. Overall, we conclude that ASU increased the informativeness of reported earnings but also induced greater conservatism in auditors. The standard appears to have been effective in the first year of adoption at least in terms of providing useful information to the market. However, the change in auditor s reporting strategy seems to be an unexpected consequence of the new policy. At this point, it is unclear whether the increased conservatism of auditors suggests more accurate going concern assessments or overly conservative decisions. Our study makes the following contributions. First, we provide detailed descriptive analyses of managements disclosure pursuant to the new standard. Second, we provide some preliminary assessments of the impact of the standard. We contribute specifically to the debate on the potential usefulness of mandating management going concern assessments by documenting increase in earnings informativeness, and to the more general subject of the 5 This is based on the full sample of all firms. We also conduct a similar analysis for financially distressed firms. For the distressed sample, we find a 3.26% overall increase in the probability of issuance of a going concern opinion, 0.40% due to change in auditor reporting strategy and 2.86% due to change in client risk characteristics. 7

9 usefulness of FASB s standards (Khan, Li, Rajgopal, and Venkatachalam 2017). Third, we contribute to the literature on the auditor s going concern decision (Mutchler, Hopwood and McKeown 1997; Behn, Kaplan, and Krumwiede 2001; Myers, Schmidt, and Wilkins 2014) by suggesting that the management s disclosure may become an important input into the decision. We note one caveat to our study. In 2015, the PCAOB introduced a new regulation requiring auditors to disclose the names of engagement partners on audits in a new Form AP (PCAOB s Release No , PCAOB 2015). This rule became effective for audit reports issued on or after January 31, The PCAOB expects these disclosures to increase transparency and accountability for key participants in the audit. Because the timing of the Form AP rule coincides with that of ASU , we must consider its potential confounding effects for our analyses. A priori it is not clear that the first-time partner disclosures could affect overall investor perceptions about financial reporting. These disclosures are expected to become useful over time as a partner s past performance becomes available. 6 It seems unlikely therefore that our investor perception results are affected by the new partner disclosures. However, if the expectation of disclosure of their names causes engagement partners to become conservative in their going concern decisions, our conservatism results could reflect the joint effect of ASU and the Form AP disclosures. While we acknowledge this possibility, the differences in our results for different types of management disclosures suggests that ASU has some effects that are separate from those of Form AP. The next section discusses background and hypotheses development. Sections 3 and 4 6 At the time of the release of the final rule, PCAOB Board member Jay Hanson remarked over time, coupled with information about the partners' experience and history, making [partner identity information] available to investors may incrementally increase their ability to make judgments about audit quality, and, by extension, the credibility of financial statements. 8

10 present research design and results of our first and second research questions, respectively. Our conclusions appear in section 5. II. BACKGROUND AND HYPOTHESES DEVELOPMENT The Auditor s Going Concern Opinion An auditor s going concern opinion decision follows several steps. Auditing Standard 2415 (SAS No. 59 prior to reorganization by the PCAOB) requires auditors to first evaluate, based on the audit procedures conducted during the audit, whether there is substantial doubt about the client s ability to continue as a going concern within a reasonable period not to exceed one year from the date of the financial statements. The concern about substantial doubt arises if the auditor identifies negative conditions and events (for example, work stoppages and legal proceedings) which, when viewed in aggregate, raise the possibility that the client may not survive beyond 12 months from the year end. 7 If the evaluation suggests that there could be substantial doubt, the auditor obtains management plans to address the doubts about survival. If, again in the auditor s assessment, the management s stated plans can be effectively implemented and might alleviate the concerns about substantial doubt, the auditor would not issue a going concern modified opinion, but should consider the need for disclosure of the principal conditions and events that initially caused him to believe there was substantial doubt (AS 2415). In practice, auditors have not typically included disclosures when not issuing an AGC. In the absence of such perceived alleviation, the auditor issues a going concern modified opinion, which is an audit report 7 See Carson, Fargher, Geiger, Lennox, Raghunandan, and Willekens (2013) for a research synthesis of going concern literature. 9

11 including an explanatory paragraph, containing the phrases that includes the terms "substantial doubt and going concern. Management s Responsibility to Assess Going Concern Status In 2014, the FASB instituted ASU This new standard was a culmination of repeated attempts by the FASB to include guidance on the preparation of financial statements as a going concern and on management s responsibility to evaluate and disclose uncertainties about an entity s ability to continue as a going concern in US GAAP (FASB 2014). The ASU requirements, while generally similar to the requirements in the auditing standard for the AGC, nevertheless have some new features. ASU requires management to consider whether there is substantial doubt about survival and, if the initial assessment suggests the possibility of substantial doubt, to assess if management plans could be implemented effectively and could alleviate the substantial doubt. While this is similar to the auditing standards, there are two important differences. First, the period for the assessment is 12 months beyond the date of issuance, or date available for issuance, of the financial statements. This is longer than the auditor s look-forward period, 12 months beyond the balance sheet date. Second, the auditing standard does not have a clear definition of substantial doubt. ASU states that substantial doubt exists when it is probable that an entity will be unable to meet its obligations as they become due within one year after the date that the financial statements are issued (FASB 2014; Booker and Booker 2016). The definition of probable (i.e., a future event is likely to occur) is intended to be consistent with ASC 450, Contingencies. If the plans are determined to not alleviate the substantial doubt, management must disclose that there is substantial doubt about survival, its evaluation about conditions and plans, and whether financial statements are prepared on a going concern basis (FASB 2014). However, 10

12 even if the plans are determined to alleviate substantial doubt, ASU requires extensive disclosures that are not required for the auditor: the principal conditions or events that raised substantial doubt, management s evaluation of the significance of those conditions or events in relation to the entity s ability to meet its obligations, and the plans that are expected to alleviate the doubt. Investors Response to the New Standard Because ASU is a FASB standard, its direct target is financial reporting quality and not audit quality. FASB standards have the general goal of improving financial accounting and reporting standards to provide useful information to investors and other users of financial reports. Specifically, for ASU , the goal is to improve financial reporting quality by ensuring that the entity s adoption of the going concern presumption is appropriate, and that appropriate disclosures about going concern issues are provided. Although a majority of the attention on the standard has focused on the disclosures by entities with substantial doubt concerns, the standard applies to all entities. In its discussion of the standard, the FASB describes two benefits for all entities. First, the standard describes management s responsibility to evaluate and disclose uncertainties about an entity s ability to continue as a going concern, which is a critical presumption for preparing financial statements under GAAP for all entities. Second, by providing a definition for substantial doubt, the standard is expected to reduce diversity in the timing and content of existing footnote disclosures for all entities. Although viewed as a cost of implementing the standard, the Board also expects that, because of the significant judgments involved in the evaluation, entities may 11

13 need to implement and document underlying processes and controls. 8 This too can add to the effectiveness of management assessments of going concern. Thus, financial reporting quality can improve for all entities because their management has systematically evaluated the going concern presumption, possibly instituting new controls, procedures and monitoring. 9 Although the FASB does not discuss implications for auditing, comments on the ASU exposure draft suggest that an additional layer of rigor can also occur because the auditor must in turn audit these changes to the reporting process and footnote disclosures. Further, although auditors are not required to audit the MD&A, they must read the new information arising from the standard, to ensure consistency with the information in the financial statements (Cohen, Gaynor, Holder-Webb, Montague 2008). We examine whether the market perceives an improvement in financial reporting quality, thus responding more positively to earnings surprises in the POST period compared with the PRE period. Earnings reports are valued by investors when the reported numbers more accurately reflect true economic value (Teoh and Wong 1993). Other things equal, the earnings response coefficient, is positively associated with the precision in the earnings number (Holthausen and Verrecchia 1988). Thus, financial reporting changes that can be expected to signal greater precision in the earnings number can increase the earnings response coefficient 8 The accounting firm Ernst & Young notes that, management will need to evaluate whether it has adequate processes and internal controls in place to comply with the going concern requirements (EY 2017). 9 Some commenters provided similar views. For example, we support the proposed amendments and believe that they will improve the quality of financial reporting about going concern matters. As a result, we believe users of financial statements will have access to more timely and decision-useful information (BDO). we believe the users of financial statements will be provided a clearer understanding of the events or conditions that may impact an entity s ability to continue as a going concern as well as management s plans to mitigate those conditions and events at an earlier stage than under the current auditor driven model (McGladrey). 12

14 (Teoh and Wong 1993; Hackenbrack and Hogan 2002). Hackenbrack and Hogan (2002) argue, for example, that certain types of auditor changes can signal greater precision, engendering a higher ERC after the auditor change. Similarly, Chen, Krishnan, Sami, and Zhou (2013) find that there was an increase in ERC in the first year of implementation of internal control audits for firms that did not have internal control material weaknesses. The inference is that the market perceived the internal control audits as having improved the quality of financial information presented by the firm. The critical question is whether investors perceive ASU as generating a higher quality of information. We distinguish between firms that receive AGCs and clean opinions from their auditors. AGC Firms For AGC firms, the auditor follows auditing standards to conclude that there is substantial doubt about survivability that is not alleviated by management plans. Management is required by Securities and Exchange Commission (SEC) rules, but not (prior to ASU ) by the FASB, to provide disclosures when the auditor issues an AGC. 10 The filings including the AGC should also include appropriate and prominent disclosure of the financial difficulties giving rise to that uncertainty and a discussion of a viable plan that has the capability of removing the threat to the continuation of the business and can enable the issuer to remain viable for at least the 12 months following the date of the financial statements being reported on must be included (SEC 2017). The FASB believes that the lack of guidance in GAAP led to 10 The SEC s Financial Reporting Manual (SEC 2017) states the following: Filings that include reports having going concern modifications must also include appropriate and prominent disclosure of the financial difficulties giving rise to that uncertainty. Discussion of a viable plan that has the capability of removing the threat to the continuation of the business must be included. The plan may include a best efforts offering so long as the amount of minimum proceeds necessary to remove the threat is disclosed. The plan should enable the issuer to remain viable for at least the 12 months following the date of the financial statements being reported on. 13

15 considerable diversity in whether, when, and how an entity discloses the relevant conditions and events in its footnotes. As discussed, ASU provides more detailed guidance than the SEC about firm disclosures. Thus any change in investors perceptions due to the new standard depends on whether it provides new information in the first year compared with the previous year. The FASB notes that the new standard may not result in new information in many audited financial statements because the amendments are similar to current U.S. auditing standards (emphasis added). This may be particularly true for AGC firms because of the SEC-mandated disclosures discussed above which were in place in the pre-asu period. Effectively, the ERC for the PRE period is based on the AGC and SEC-mandated disclosures, while the ERC in the POST period is based on the AGC, the SEC-mandated disclosures, and any new disclosures arising from ASU Because of the existence of some mandated disclosures in the PRE period, we state our first hypothesis in null form: H1: ERCs for firms with going concern opinions will not be different in the first post-asu year from that in the previous year. Non-AGC Firms For the clean opinion firms, the absence of an AGC in the PRE period provided assurance of the absence of going concern problems. Under ASU , investors have, in addition, management s proactive assessment of going concern, possibly after institution of processes/controls to facilitate this assessment. Such assessment provides an increase in information which may be viewed positively because it reduces information asymmetry. If so, the ERC can increase in the post-asu year. However, the content of the information 14

16 can also influence investor response. Consequently, we develop two hypotheses below for firms with clean audit opinions. ASU requires detailed disclosures when the initial assessment raises concern about substantial doubt that are subsequently found to be alleviated by management plans. As we discuss in the next section, the non-agc companies fall into two broad categories, those that (1) have no going concern disclosures or state explicitly that they have no going concern issues and (2) describe the nature of the substantial doubt, management plans, and the basis for judging that the substantial doubt is alleviated. In contrast, clean companies in the PRE period were not required to make any disclosures. Consequently, there were relatively few voluntary disclosures by clean companies regarding going concern issues in the PRE period. For group (1), investors are essentially provided a double assurance in the POST period that the firm is clean because the auditor did not issue a going concern opinion and the management confirms the absence of going concern issues in the near future. In contrast, only the auditor provided the assurance in the PRE period. Therefore, we posit that investors are more confident that clean firms are in fact free of going concern uncertainties. The double assurance can increase the precision of the earnings number (Teoh and Wong 1993). As a result, we expect the earnings of these clean firms to be more informative in the POST period. Our second hypothesis is as follows: H2: ERCs for firms with no auditor going concern opinions and with no negative management disclosures regarding going concern will be higher in the first post-asu year than in the previous year. We acknowledge however that there are counter arguments. First, the evaluation of whether there is substantial doubt involves significant judgement because it is based on qualitative and quantitative factors. Thus managers may not disclose going concern issues 15

17 because they regard a potentially significant problem as not significant. Second, managers have incentives to hide or delay bad news due to career concerns (Kothari, Leone, and Wasley 2005). For example, the empirical evidence from Canada and U.K., where management is required to disclose going concern issues, indicates that the proportion of firms that report going concern uncertainties is quite small, and the information disclosed by the management is not incrementally useful (Uang et al. 2006; Ontario Securities Commission 2010). If the market considers it likely that firms would not disclose all the relevant information even when required to do so, the new policy may not change the market s perception of earnings informativeness, and there would be no change in ERC in the POST period. Third, the threshold requiring the disclosure about going concern uncertainty only when it is probable (i.e., the future event or events is likely to occur) may be too high to reveal any new information about the uncertainty as investors are likely to have already known about it. 11 For group (2) above, disclosures in the POST period provide information relating to going concern problems. In the PRE period, investors knew only the outcome of the auditor s going concern evaluation process. If the auditor decided to not issue a going concern opinion after incorporating management s plans, investors simply saw the absence of a going concern opinion without knowing the underlying contextual details, unless management provided voluntary disclosures. Mayew et al. (2015) document that some firms discussed going concern issues in their MD&As, and this disclosure has incremental predictive ability for bankruptcy 11 Board Member Thomas Linsmeier voted against the ASU proposal, arguing that by requiring disclosure only when it is probable that an entity will be unable to meet its obligations as they become due within one year after the date the financial statements are issued (or available to be issued), the guidance in this Update will provide information about going concern uncertainties that is too late to be of significant benefit to users of financial statements (FASB 2014). 16

18 even after controlling for the auditor opinion. 12 Therefore, if firms have already disclosed sufficient information in their MD&As or elsewhere in the 10-K, ASU may not provide new information. However, the proportion of voluntary going concern disclosures in the MD&A for clean companies in Mayew et al. (2015) is small. Because the disclosures are now mandatory (rather than voluntary), and the new policy would likely have legal consequences if management does not comply with it, it is reasonable to expect that firms are more likely to disclose going concern information in the POST period than in the PRE period. Even if management s disclosures provide new information, the bad news nature of these disclosures might lower ERC. For example, Choi and Jeter (1992) and Dong, Robinson, and Robinson (2015) document lower ERCs for firms with qualified/going concern audit opinions. Therefore, although we expect the new policy to provide new information to the market, whether this information would change market s valuation of non-agc firms earnings is an empirical question. Accordingly, our third hypothesis is in null form: H3: For the firms whose managements discuss going concern issues when the auditor does not issue a going concern opinion, there is no change in ERCs in the first post-asu year from the previous year. Auditors Response to the New Standard The auditor s going concern opinion decision is complex, and characterized by grey areas requiring judgment (Carcello and Neal 2000; Goh, Krishnan, and Li 2013; Carson et al. 2013). Specifically, auditors can be viewed as determining a range of financial distress over which an AGC can be issued, and choosing a threshold above which to issue it. Researchers have argued 12 Specifically, the proportion of voluntary disclosures in Mayew et al. (2015) is only about 0.2 percent when the auditor does not issue a going concern opinion. In our sample, 2.6% of firms with no AGCs in the POST period disclosed going concern issues. 17

19 that, given the judgement needed in the process, auditors may move the threshold in response to different factors. Lowering (raising) the threshold would result in a higher (lower) probability of issuing a going concern opinion, other things equal. Francis and Krishnan (1999) argue that uncertainties relating to estimating accruals causes a lowering of the threshold engendering reporting conservatism. Similarly, Goh et al. (2013) argue that the existence of internal control material weaknesses increases the auditor s uncertainty surrounding the substantial doubt assessment, causing it to lower its threshold for the going concern opinion, other things equal. Blay (2010) argues that the fear of losing a client, other things equal, makes the auditor less conservative. In addition to responding to uncertainties surrounding the required assessments, auditors may also move the threshold in response to regulatory changes. For example, the passage of the Private Securities Litigation Reform Act of 1995 relieved litigation pressure on auditors, possibly making them less conservative than before (Francis and Krishnan 2002; Geiger and Raghunandan 2002). Prior to ASU , management plans were incorporated in the auditor s going concern decision, but the plans were proffered to the auditor in response to the latter s expressed concerns about substantial doubt. The important question is whether the active role forced on management by the new standard is likely to affect the auditor s going concern opinion decision. We posit that several forces may be at work. First, auditors are aware that management must assess its going concern status, and report it. This can result in greater faith in the information provided by management, thus moving the threshold to the right (i.e., less conservatism). Second, auditors will face more scrutiny from regulators (i.e., SEC, PCAOB) and more attention from the financial statement users in the first year of adoption of a new standard. It is likely that standard setters will pay more attention to going concern assessments to evaluate the efficacy of 18

20 the standard and to ensure that firms are applying the new standard appropriately. Such heightened scrutiny could cause auditors to become more conservative. Further, some commenters responding to 2013 exposure draft expressed the concern that the new management disclosures can increase auditors litigation exposure. Third, we conjecture that auditors now have less pressure to avoid issuing a going concern opinion for fear of losing the client because managements are also responsible for disclosing going concern issues. That is, auditors that may have delayed disclosing bad news (i.e., issuing a going concern opinion) to please the client would feel free to issue a going concern opinion without worrying about losing the client particularly if the client is forced to disclose problems on its own. This would increase reporting conservatism. Given the conflicting predictions, we state our fourth hypothesis in the null form. H4: After controlling for change in client characteristics, auditors reporting conservatism is not different in the first post-asu year compared with the previous year. III. INVESTOR PERCEPTIONS IN THE PRE- AND POST-PERIODS Management Disclosures pertaining to Going Concern ASU became effective for fiscal years ending after December 15, 2016, and for annual periods and interim periods thereafter. We assemble a sample of PRE and POST firmyear observations using December 15, 2016 as the cutoff. Because of the fairly intense data coding that is required, we confine our sample to observations that were available on the date that we started the process, July 1, We started with all observations that were available for year ends from November 1, 2015 to January 31, Fiscal years ending between November 1, 2015 and December 15, 2016 comprise the PRE period, and fiscal years ending between December 16, 2016 and January 31, 2017 comprise the POST period. 19

21 Table 1 presents our sample selection procedure. We start with 12,282 firm-year observations available in Audit Analytics for the PRE and POST periods. We then eliminate (1) 1,419 observations that relate to funds or trusts, use non-u.s. GAAP, or have missing SIC Codes and (2) 6,252 observations due to unavailability of data on Compustat, CRSP, and/or IBES. This results in 4,611 firm-year observations comprising 2,328 observations for the PRE period and 2,283 observations for the POST period. Next, we hand-collected data on management disclosures about going concern from 10-K filings for every firm-year in the sample. While the POST period had mandatory disclosure requirements, Mayew et al. (2015) document that managers provided voluntary disclosures relating to going concern (during ) in the MD&A sections of their Form 10-K filings. Because our empirical models are intended to capture changes in investor perceptions from the PRE to the POST periods, we examine 10-K filings in both periods to document the disclosures. Using Python, we conducted keyword searches for phrases such as substantial doubt and going concern. 13 We classified the observations into five groups, those with (1) explicit statements that they had no GC issues (MExpNoGC), (2) no explicit statement about the presence or absence of going concern issues, which we interpret as indicating the absence of GC issues (MSilent) 14, (3) mild suggestions of GC problems (MildMGC), (4) discussions about 13 Two authors were involved in coding the disclosures. We read each paragraph that contained the phrase going concern. Many of these paragraphs did not in fact pertain to going concern problems. For example, many companies use the phrase going concern when referring to the new standard. Consequently, a large number of going concern phrases were eventually not used in the coding of going concern assessments by management. Where there were differences between the authors in coding, they were reconciled through discussion. 14 A number of firms mention ASU and state that its adoption had no material effect on their financial statements. Since this is not an explicit statement about the absence or presence of going concern issues, we code these observations as silent. 20

22 GC concerns and mitigating factors (MGCMit) that alleviate the substantial doubt, and (5) explicit statements about substantial doubt relating to going concern (MGCExp). 15 A mild suggestion of GC arises when a firm uses modal words (e.g., may or could ) or conditional sentences in stating its going concern issues. 16 Table 2, Panel A shows the frequency of the management disclosure groups, crossclassified by AGC, for the combined sample of PRE and POST period firm-years. Not surprisingly, managements explicit statements about substantial doubt overlaps almost completely with auditors issuance of AGCs. Of the 110 firms with auditors AGCs, the management of 103 firms also explicitly acknowledge going concern problems. Of the remaining seven firms, six acknowledge going concern issues somewhat mildly (MildMGC) and one (which belongs to the PRE period) had no comments about going concern (MSilent). Among the 4,501 clean firm-year observations that did not receive an auditor s going concern opinion, managements disclosures vary. A majority (4,299) of them fall in the MSilent group, and 101 firms belong to the MExpNoGC group, stating explicitly that the firm has no going concern issues. We find 86 firms with mild acknowledgment of going concern 15 MGCExp firms: These raise substantial doubt regarding our ability to continue as a going concern ; There is substantial doubt about its ability to continue as a going concern. The structure and content of going concern statement made by the management is almost identical to the going concern statements included in the auditor s report. 16 Some examples of MildGC: If the condition were to persist for any appreciable period of time, our viability as a going concern could be threatened. ; Our ability to continue as a going concern is dependent upon our ability to obtain additional equity or debt financing. 21

23 problems, and 12 firms disclosing going concern problems but with mitigating plans to alleviate the problems. 17 Three explicitly mention going concern problems. 18 In Table 2, Panel B, we focus on clean firms, (AGC=0) and present the distribution of management going concern disclosures for the PRE and POST periods separately. Although the individual numbers are small, the changes from the PRE to POST years reflect the effect of the new standard. First, the MSilent category decreased from 97.5% to 93.5%. 19 Although this category dominates in both years, the introduction of ASU provides a different interpretation for MSilent in the PRE and POST periods. In the PRE period, investors were not able to draw any inferences from the lack of disclosure about the going concern issues in the 10- Ks. In the POST period, however, management is required to assess going concern and disclose where needed. Thus, no disclosure in the POST period actively implies the absence, in management s judgment, of going concern issues. Second, although the number of observations with any kind of management going concern disclosure is not large, the standard elicited specific statements, for example about the 17 ASU says: If, after considering management s plans, substantial doubt is alleviated as a result of management s plans, management must disclose conditions that raised substantial doubt as well as management s plans that alleviated substantial doubt in the footnotes ( ); If, after considering management s plans, substantial doubt is not alleviated, management must include a statement in the footnotes indicating that there is substantial doubt about the entity s ability to continue as a going concern ( ). Therefore, some firms may have mitigating plans which are not sufficient enough to alleviate adverse conditions and thus still state going concern problems. These firms will fall within the MGCExp group rather than within MGC with Mitigating Factors. 18 The three firms include the following sentences in their 10-Ks: (1) Our recurring losses from operations have raised substantial doubt regarding our ability to continue as a going concern ; (2) our planned operations raise doubt about our ability to continue as a going concern ; (3) the company s planned operations raise doubt about its ability to continue as a going concern. 19 We considered whether the observations including some form of going concern disclosures were the result of early adoption of the standard. Five firms mention that they adopt the standard early. Others made no mention of early adoption of the standard, making it likely that they are voluntary disclosures. 22

24 absence of going concern issues, or about management plans and mitigating factors, which were not available in the PRE period. Overall, these specific disclosures increased from 2.5% to 6.5%. The proportion of firms explicitly stating the absence of going concern increased from 0.57% to 3.96%, and the proportion of firms explicitly disclosing some going concern issues (mild, with mitigation, or unalleviated substantial doubt) increased from 1.93% in the PRE period to 2.57% in the POST period. The proportion of disclosures for MGC with mitigating factors (MGCMit) increased significantly after ASU became effective (from 0 to 0.54%), and we conjecture that this is a major change introduced by the new policy. 20 Unfortunately, the small number of observations restricts our ability to conduct meaningful statistical tests for each category. Consequently, in the regressions reported later, we combine these three groups into one group (MGCALL). Location of Disclosure ASU requires disclosure about going concern problems in the footnotes. However, information about going concern issues also appear in other sections of Form 10-K, especially in the Risk Factors and MD&A sections. 21 Table 2, Panel C provides information about the location of the disclosures. For the companies with auditor going concern opinion 20 The Sears example discussed earlier falls in the MGC with Mitigating Factors (MGCMit) group. We believe that the actions discussed above are probable of occurring and mitigating the substantial doubt raised by our historical operating results and satisfying our estimated liquidity needs 12 months from the issuance of the financial statements. Sears auditor, Deloitte, did not issue an AGC, possibly because the substantial doubt was alleviated. Thus, management disclosure provides additional information to investors. 21 The FASB included question 7 in its 2013 exposure draft for ASU , specifically seeking comments about the location of the disclosures in the 10-K: For SEC registrants, would the proposed footnote disclosure requirements about going concern uncertainties have an effect on the timing, content, or communicative value of related disclosures about matters affecting an entity s going concern assessment in other parts of its public filings with the SEC (such as risk factors and MD&A)? Please explain. 23

25 (AGC=1), footnote disclosures dominate in both periods (91.5% and 100% in the PRE and POST periods respectively) possibly because of the SEC requirement. However, among AGC=0 companies, there is a distinct increase in footnote disclosures in the POST period. Where management explicitly states that there are no going concern problems, 84.6% of the disclosures in the PRE period and 99% in the POST period appear in the footnotes. Among the clean (AGC=0) companies with some going concern problems, 9.1% provide footnote disclosure in the PRE period. The corresponding percentage for the POST period is 35.1%. Empirical Model We examine if the adoption of ASU enhanced investors perceptions about financial reporting quality, leading them to re-evaluate the quality of reported earnings. We estimate the earnings response coefficient (ERC), the market s responsiveness to earnings announcements by the slope coefficient in a regression of unexpected returns on unexpected earnings (see for example, Balsam, Krishnan, and Yang 2003; Francis and Ke 2006; Baber, Krishnan, and Zhang 2014). The basic ERC model is as follows: CCCCCC = ββ 0 + ββ 1 UUUU + 7 kk=2 ββ kk cccccccccccccccc 13 + ββ kk UUUU cccccccccccccccc kk=8 + εε (1) where CAR is the three-day cumulative abnormal return around the earnings announcement date, and UE is a measure of earnings surprise. 22 The earnings announcement date is for the first quarter following the 10-K filing in which the management disclosures (voluntary in the PRE period and mandatory in the POST period) are included. The parameter ββ 1 is the coefficient of 22 We use an event study design with three-day cumulative returns rather than a long-term association study design because our goal is to investigate the information content of earnings announcement rather than value-relevance (Collins and Kothari 1989; Hackenbrack and Hogan 2002). 24

26 interest, namely the earnings response coefficient. CAR is computed as the common stock return less the value-weighted market return summed over the three-day period centered on the earnings announcement date UE is earnings per share less the most recent analyst forecast deflated by the beginning stock price for the first quarter. Earnings Responsiveness to Auditors Going Concern opinion in the PRE and POST Periods In order to compare the ERCs of going concern firms and clean firms (hypotheses H1), we extend the model to estimate ERC for each group: CCCCCC = αα 0 + αα 1 UUUU AAAAAA + αα 2 UUUU AAAAAAAAAAAA + 9 kk=3 αα kk cccccccccccccccc 15 + αα kk UUUU cccccccccccccccc + υυ (2) kk=10 where AGC indicates firm-year observations with auditor going concern opinions. AClean indicates firm-year observations that did not receive going concern opinions from the auditors. Control variables are discussed below. The coefficients αα 1 and αα 2 indicate the mean ERC for the AGC and AClean groups, respectively. Since our goal is to compare ERCs for the PRE period and POST periods, we estimate model (2) separately for the PRE and POST periods and examine differences in coefficients between the two periods. 23 Specifically, the difference in αα 1 between the PRE and POST periods indicates whether the ERC for AGC firms change. Hypothesis H1 predicts no sign for the difference in αα 1. Next, to test hypotheses H2 and H3, we examine whether the ERC changes for firms with different management going concern disclosures. We restrict this analysis to clean firms because, as discussed, there is an almost-complete overlap between the auditor s going concern opinion and management s substantial-doubt going concern disclosures. We use two types of 23 Thus the model allows all the coefficients to vary across the two periods. 25