Audit Committee Accounting Expertise and Audit Quality – the Case of Going-Concern Opinions

This study examines whether audit committee accounting expertise and other audit committee characteristics promote or deter the likelihood of receiving going-concern reports from the auditors and whether such characteristics shield auditors from dismissals after the issuance of a going-concern report. The study finds no significant association between the likelihood of a going-concern report and audit committee accounting expertise or other audit committee characteristics. No significant association is also found for auditor dismissals following going-concern reports and audit committee accounting expertise. These results contrast with prior literature that examined data preceding the passage of the Sarbanes-Oxley Act of 2002 (hereafter SOX) or the period immediately thereafter. Additional analysis shows that audit committee accounting expertise is found to improve the information in going-concern audit opinions by reducing Type I errors, however. Overall, these findings shed light on the evolving role of audit committees in overseeing the auditors and have implications for regulators interested in improving audit quality and investors interested in improving the effectiveness of audit committees.


Introduction
The role of audit committees in protecting the credibility of financial reporting and auditor's reporting behavior has been studied extensively in the accounting literature. Carcello and Neal (2000) find that the greater the percentage of affiliated directors on the audit committee, the lower the probability of auditor going-concern opinions. Carcello and Neal (2003) report that audit committees with more independent directors and higher governance expertise, shield auditors from dismissals after the issuance of new going-concern opinions. In a review of the research on auditor going-concern opinions, Carson, Fargher, Geiger, Lennox, Raghunandan, and Willekens (2013) note that the findings on audit committees and auditor reporting behavior pertain to the period before SOX or the period immediately following the passage of SOX. They call for research that examines the role of audit committee characteristics such as financial expertise and tenure, on the auditor going-concern opinions and for a more recent period. This study answers their call by examining whether audit committee characteristics such as the accounting financial expertise of the audit committee promote or deter auditor going-concern opinions and the role of audit committee characteristics in dismissals of auditors following a going-concern opinion for the years 2016-18.
Prior literature has shown that some audit committee characteristics improve the oversight of financial reporting. In particular, several studies have documented that the accounting financial expertise of the committee members is associated with higher financial reporting quality (Dhaliwal, Naiker, & Navissi, 2010, Krishnan & Visvanathan, 2009). Other studies also document a positive association between audit committee characteristics and higher audit quality (Lisic, Myers, Seidel, & Zhou 2019). To assess the role of audit committee characteristics in improving audit quality in a more recent period, this study begins by examining the relationship between audit committee characteristics and auditor going-concern opinions. A similar analysis is also conducted for auditor dismissals following a going-concern opinion. The sample is collected over the period 2016-18 to address the robustness of findings in prior literature pertaining primarily to the period before or surrounding the passage of SOX. The sample consists of financially distressed firms (Carcello & Neal, 2000) as defined in Blay and Geiger (2013). The analyses are conducted by expanding on the models of Carcello and Neal (2000) and Carcello and Neal (2003). While several audit committee characteristics are used, the key variable of interest is the accounting financial expertise of the committee, consistent with its prominent role documented in audit committee literature. The findings indicate that Several additional analyses are conducted to address the robustness of the findings. Procedures examining the potential effects of endogeneity produce similar results. Tests examining CEO influence on the audit committee (Lisic et al., 2019) do not indicate any notable differences in the analysis. The results are also unaffected when alternative measures of audit committee expertise, such as the presence or proportion of experts, and the number of experts, are used in the model.
The study makes the following contributions to the literature: first, by answering the call for research on auditor going-concern opinions and audit committee characteristics (Carson et al., 2013), the study finds evidence that differs from prior research underscoring the evolving role of the audit committee characteristics and auditor reporting behavior. Second, the study finds that while accounting financial expertise of the audit committee does not play a role in going-concern opinions and subsequent dismissals of auditors, it improves the information in audit opinions by reducing Type I errors. This finding has not been documented previously and thus extends the literature on the role of audit committees in improving audit quality. Third, the evidence in the study contrasts with evidence on internal control weaknesses and audit committee accounting financial expertise (Lisic et al., 2019) indicating the relative shift in the relationship between audit committees and auditors regarding going-concern opinions. Finally, regulators interested in the oversight role of audit committees should find these results helpful in assessing the effectiveness of requiring a narrow or broader definition of financial expertise from board members.
The rest of the study is organized as follows. Section 2 reviews the prior literature and develops the hypotheses, section 3 describes the methodology, and section 4 presents the empirical results. The final section provides a summary and notes limitations to the study.

Hypotheses Development
The relationship between audit committee composition and auditor reporting behavior has been of interest to regulators and researchers. The Blue Ribbon Committee made several recommendations to improve the effectiveness of audit committees including the composition and independence of committee members (BRC, 1999). In 2002, SOX introduced several important changes in audit committee composition and responsibilities. Following SOX, audit committee members need to be independent and the audit committee is in charge of the appointment, compensation, and oversight of auditors. SOX underscored the importance of financial expertise on the audit committee by requiring disclosures on whether the committee includes at least one member who is a "financial expert." The role of financial expertise on the audit committee in improving financial reporting quality and auditor reporting behavior has been extensively studied in the prior literature (Krishnan, 2005, Zhang, Zhou & Zhou, 2007. Audit committee financial expertise has been found to be associated with lower restatements (Abbott, Parker & Peters, 2004, Agrawal & Chadha, 2005, greater accounting conservatism (Krishnan & Visvanathan, 2008), internal control weaknesses (Krishnan & Visvanathan, 2007), audit fees (Abbott, Parker, Peters & Raghunandan, 2003) and accrual quality (Dhaliwal et al., 2010). The SEC definition of a financial expert includes those who have either direct accounting experience or supervisory experience over accounting personnel (SEC, 2003). Dhaliwal et al. (2010) and Krishnan and Visvanathan (2007) find that supervisory experience is not associated with higher financial reporting quality, unlike specialized accounting experience. Lisic et al. (2019) argue that audit committee members with accounting expertise are likely more ethically aware because members who are accounting professionals must heed the AICPA's Code of Professional Conduct. This professional impetus likely incentivizes the members to ensure the quality of financial reporting, encourage the auditors to be diligent, and finally safeguard the auditor in instances of qualified audit opinions.
Prior literature examining the role of audit committees and auditor opinions finds differing results based on sample compositions and periods that vary. Carcello and Neal (2000) study financially distressed firms in 1994 and report that auditors of firms with more affiliated directors on the audit committee are less likely to issue a going-concern opinion. Carcello and Neal (2003) examine auditor dismissals following new going-concern opinions for the period 1988-99 and find that audit committees that protect auditors from dismissals are characterized by more independence, more governance expertise, and lower stock ownership. They use a sample of firms that received a new going-concern opinion from a Big-N auditor and subsequently dismissed the auditor. Hoitash and Hoitash (2009) examine data for the year 2004 and find that level of assurance increases and the probability of dismissal decreases with stronger audit committees. Lisic et al. (2019) find that the probability of adverse internal control audit opinions is higher and the likelihood of subsequent auditor dismissals is lower for firms with audit committees having greater accounting expertise. Their data is restricted to accelerated filers over the period 2004-13. In a review of the audit literature, Carson et al. (2013) note that prior research on going-concern opinions and audit committee characteristics pertain to periods before or immediately following the passage of SOX that requires independence as a requirement for all audit committee members. Thus, they call for research examining other audit committee characteristics such as accounting financial expertise, tenure, and background, in the context of auditor going-concern opinions. Following Carson et al.'s (2013) call for research, this study examines the relationship between issuance of going-concern reports and audit committee accounting financial expertise along with other audit committee characteristics. Carcello and Neal (2000) note that a going-concern audit opinion is often a result of a contentious negotiation among auditors, management, and the audit committee, and given that issuing a going-concern report is a difficult and ambiguous audit task, the auditor is susceptible to management pressure through the potential for dismissal or fee reductions. A strong audit committee may protect the auditor in such instances by mitigating management pressure. This implies that strong audit committees are likely associated with a higher frequency of auditor going-concern reports in instances where such opinions are warranted. In line with this prediction, Carcello and Neal (2000) find evidence showing an inverse relationship between the probability of receiving a going-concern report and the proportion of affiliated i.e., non-independent, directors on the audit committee in the pre-SOX period. Lisic et al. (2019) argue that in the post-SOX period, internal control weaknesses rather than going-concern opinions are a better setting to examine the relationship between audit committee accounting expertise and audit quality as subsequent restatements could be used to verify whether the auditors failed to report an internal control weakness. They note that in the case of going-concern reports, subsequent bankruptcies tend to be few. However, note that Lisic et al. (2019) examine only accelerated filers i.e., firms with greater than $75 million in public float, in contrast to Carcello and Neal (2000) who examine financially distressed firms, as the latter group is more likely to receive going-concern reports. Thus, it is unclear if Carcello and Neal's (2000) findings continue to apply to the post-SOX setting. Moreover, as discussed before, their findings pertain to audit committee independence which is mandatory after SOX, and they did not study the role of audit committee accounting expertise nor other characteristics of audit committees that have been shown to play a significant role in improving accounting and audit quality (Carson et al., 2013). Thus, the first research question is stated in the null form: H1: For financially distressed firms, the issuance of an auditor going-concern report is not associated with audit committee accounting expertise.
Prior studies have found that firms that receive a going-concern audit report are likely to change auditors (Geiger, Raghunandan & Rama, 1998). Carcello and Neal (2003) using pre-SOX data examine dismissals of auditors after the issuance of a new going-concern audit report, as it is the responsibility of the audit committee to protect the independence of the auditor to ensure audit quality. They find that audit committees that protect auditors from dismissals are characterized by more independence, more governance expertise, and lower stock ownership. They do not find any significance for the role of the committee's financial expertise. Hoitash and Hoitash (2009) use data for one year after the passage of SOX and find an inverse relation between dismissals and audit committee financial expertise. In the context of dismissals after an audit report on internal control weaknesses, Lisic et al. (2019) find that dismissals after such reports are inversely related to audit committee accounting expertise. Considering the differing evidence reported by prior studies and the sample differences in those studies, it is unclear whether auditor dismissals following the issuance of a going-concern opinion are related to audit committee accounting expertise in periods after SOX. Thus, the second research question is stated in the null form: H2: Auditor dismissal following the issuance of an auditor going-concern report is not associated with audit committee accounting expertise.

Research Design
To test hypothesis 1, a going concern model augmented by the audit committee and other governance variables is used. For the going concern model, the study follows DeFond, Lim, and Zang (2016) and for the audit committee and governance variables, the study follows Carcello and Neal (2000) and related literature (Lisic et al., 2019). The model is specified as follows: where: GCO = 1 if the company receives a going-concern audit opinion, zero otherwise; FIRSTGC= 1 for clients receiving a going concern opinion for the first time, and 0 otherwise; AFED = 1 if the audit committee has at least one accounting expert, zero otherwise. An accounting expert is a director with experience as a certified public accountant, auditor, principal or chief financial officer, controller, or principal or chief accounting officer; AFE_% = the proportion of audit committee members who are accounting experts, where an accounting expert is defined as above. The definitions for all other variables are provided in the appendix. For the model with GCO as the dependent variable, prior period going-concern opinion (LGCO) is also included in the model. Consistent with Carcello and Neal (2000), GCO is used as the dependent variable and in line with the recent literature (DeFond et al., 2016), FIRSTGC is also used as the dependent variable.
The primary variables of interest are AFED (or AFE_%) and other audit committee variables measuring the average board tenure of audit committee members (ACBRDTEN), the average number of public company boards on which audit committee members sit (GOVEXP), and the average stock ownership of audit committee members (ACOWN).
Whether the CEO is also the chairman of the board (CEOCHAIR), and two auditor characteristics, auditor industry specialization (AUSPL) and auditor tenure (TENURE) are also included based on prior literature (Lisic et al., 2019). All other variables in model 1 are included as controls as described in DeFond et al. (2016).
To test hypothesis 2, an auditor dismissal model is used, augmented by the audit committee and other governance characteristics. Following Carcello and Neal (2003) and related literature on auditor dismissals (Hoitash & Hoitash, 2009), the following model is specified: where: DISMISSAL = 1 if the auditor is dismissed in the one-year window following the filing of the annual financial statements, zero otherwise (Lisic et al., 2019). The definitions for all other variables are provided in the appendix. The primary variables of interest are AFED (or AFE_%) and other audit committee variables ACBRDTEN, GOVEXP, and ACOWN.

Sample Description and Empirical Results
The sample is collected for the 3 years 2016-18. The sample period covers a recent period while the restriction of 3 years is because some of the audit committee variables such as accounting expertise need hand collection and verification. The sample collection procedure begins by considering all firms in Compustat for the years 2016-18. Firms in SIC codes 6000-6999 are excluded following prior studies (DeFond et al., 2016). As the focus is on the likelihood of receiving a going-concern report, the sample is restricted to financially distressed firms defined as firms with both negative income and cash flow from operations in the same year (Blay & Geiger, 2013). Auditor variables are collected from Audit Analytics. Audit committee and board variables are collected from BoardEx. While BoardEx has data on the educational background and professional experience of audit committee directors, this data is either missing or incomplete for several companies or directors. Moreover, several databases identify as the financial expert, the director designated as such by the company and as noted previously, firms can designate even those with supervisory experience as financial experts. The available data in BoardEx on education and experience was verified with the proxy filings and in numerous instances, BoardEx data was incomplete to assess the accounting experience of the directors. Thus, the accounting experience variable (AFED) was primarily collected directly from the proxy filings on the SEC EDGAR database. Table 1 details the sample selection procedure that yields 2,101 firm-years as the final sample. The sample is distributed across all major industry groups as classified in Frankel, Johnson, and Nelson (2002): Durable manufacturers, Transportation, Utilities, Retail, Services, and Computers (distribution not tabulated).  Table 2 provides descriptive statistics for the primary variables used in the study. The mean of GCO suggests that 22 percent of the sample received a going-concern opinion while the mean of FIRSTGC indicates that 7.95 percent of the sample firms received a first-time going-concern opinion i.e., they did not receive a going-concern opinion in the prior year. While the overall going-concern percentage is lower than the 37 percent reported for the year 1994 in Carcello and Neal (2000), the first time going-concern percent is similar to the 8.4 percent reported in DeFond et al.
(2016) for a longer period (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009). The mean for DISMISSAL shows that 5.57 percent of the sample firms dismissed their auditors in the subsequent year, which falls in the range of 3.4 percent reported in Lisic et al., (2019) and 7.3 percent reported in Hoitash and Hoitash (2009). The mean value of AFED and AFE_% are 77.6 and 29.3 percent respectively, indicating that less than one-fourth of the firms do not have an accounting expert on the audit committee and for firms with an accounting expert, accounting experts constitute slightly less than a third of the committee. The numbers for AFED are substantially higher than the 41.1 percent reported in Lisic et al. (2019) and the 21 percent reported for the proportion of accounting members by Hoitash and Hoitash (2009). At least two sample composition differences are to be noted in considering these numbers: the sample in the current study is more recent compared to these studies and only financially distressed firms are considered in the current study in contrast to all accelerated filers (Lisic et al., 2019) or all firms (Hoitash & Hoitash, 2009 Table 3 presents Pearson correlations. The univariate correlations indicate that going-concern opinions and subsequent auditor dismissals are positively correlated. Most of the audit committee variables are negatively related to going-concern opinions. Whether this relationship holds after controlling for other determinants of going-concern opinions or dismissals needs to be explored in the multivariate analysis. Other correlations among the control variables are consistent with prior literature.   Pearson correlations are reported for variables used in the analyses. Significant correlations at p-value < 0.05 are bolded. N=2101. Table 4 presents results from estimating model 1 with dependent variables defined as either going-concern opinion (GCO) or first-time going concern opinion (FIRSTGC) and audit committee accounting expertise defined as a dummy variable (AFED) or as a proportion (AFE_%). The results show that after controlling for variables that influence the likelihood of receiving a going-concern opinion, there is no significant association between the probability of receiving a going-concern opinion and the presence of an accounting expert on the audit committee. A similar lack of significant associations also obtains for the AFE_% variable and in the model that uses only first-time going-opinions. This evidence suggests that accounting experts on audit committees do not play a significant role (by either encouraging or dissuading the auditors) in the auditors' decision to provide a going-concern opinion. The results show that other audit committee characteristics such as governance expertise (GOVEXP), the tenure of audit committee members (ACBRDTEN), and audit committee directors' share ownership (ACOWN) also do not play a significant role in the reporting of going-concern opinions. Among all governance variables, CEOCHAIR is the only significant variable though only in specifications where the dependent variable is the going-concern opinion. The results show that control variables that influence the likelihood of going-opinion are mostly in line with prior literature (LAT, ZSCOREZ, CUMRET, CHLEV, and INVEST are significant with expected signs as specified in DeFond et al., 2016). Overall, the results indicate none of the key audit committee characteristics play a role in the going-concern decision of auditors and the null is not rejected. Note that Carcello and Neal (2000) propose and find a significant role for the audit committee in supporting the auditor's going concern decision. Their results are based solely on the variable measuring the independence of the audit committee which is now a requirement for all firms in the post-SOX period.  (1) using the conditional logit model for all firms. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Logistic regression is run clustered by firm (DeFond et al., 2016). Intercept, year, and industry fixed effects are included. Table 5 presents the results for model 2. The dependent variable is auditor dismissals, and the model uses only new going-concern opinions following Carcello and Neal (2003). The focus of hypothesis 2 is whether dismissals following a new going-concern opinion are associated with audit committee accounting expertise. Thus, the variables of interest are the interactions between AFED*FIRSTGC or AFE_%*FIRSTGC. Other audit committee variables are also interacted with FIRSTGC to assess their role in auditor dismissals. The results in table 5 show that none of the audit committee variables, except for GOVEXP interacted with FIRSTGC, are significant. While GOVEXP and the interaction GOVEXP*FIRSTGC are significant, the combined coefficient is not significant in a joint test (p-value 0.25 not reported in the table). This evidence indicates that audit committee characteristics do not play a significant role in the dismissal of auditors after a new going-concern opinion in contrast to the results in Carcello and Neal (2003) in the pre-SOX period. The insignificance of the accounting expertise interaction is consistent with Hoitash and Hoitash (2009), however.  (2) using the conditional logit model for all firms. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Logistic regression is run clustered by firm (DeFond et al., 2016). Intercept, year, and industry fixed effects are included.
The lack of association between either going-concern opinions or dismissals with audit committee accounting expertise or other audit committee characteristics raises the issue of whether audit committees play any role in improving audit quality through the auditor's decision to report going-concern opinions. While prior literature Neal 2000, 2003) approached this issue by viewing the audit committee's role as an unbiased intermediary between the firm's management and the auditor in the case of going-concern opinions, the role may have shifted post-SOX. Newton et al. (2016) argue that in the post-SOX era, in contrast to opinions on internal control weaknesses, going-concern opinions are more predictable and thus no more a significant factor in a firm's decision to keep or dismiss an auditor. In line with this reasoning, they find evidence that auditor opinion shopping is significant for internal control weaknesses but not for going-concern opinions. Their evidence is potentially why the results in this study do not find an association between audit committee variables and going-concern opinions or auditor dismissals. While the role of audit committees may have shifted away from protecting the auditor with respect to going-concern reporting decisions, audit committee accounting experts can still play a role in ensuring that the going-concern opinion is appropriate and thus is informative to investors and other stakeholders. That is, by ensuring that the going-concern opinion is warranted by underlying business fundamentals, audit committees can help minimize errors in the opinions and thus improve audit quality. To assess whether audit committee accounting expertise improves the accuracy of the going-concern reporting, the following models are adapted from Geiger and Rama (2006) and Berglund, Eshelman, and Guo (2018): Pr (BANKRUPTi,t+1 = 1 | GCOi,t = 1) = β 0 + β 1 AFED it (or AFE_% it ) + β 2 ACBRDTEN it + β 3 GOVEXP it + β 4 ACOWN it + β 5 LAT it + β 6 ZSCORE it + β 7 OCF it + β 8 BIG4 it + β 9 REPORTLAG it + Year fixed effects + Industry fixed effects + e it (3a) Model 3a assesses the Type I error rate of going-concern opinions and is estimated for firms that received a going-concern opinion. REPORTLAG is the natural log of the number of days between the fiscal year-end and the earnings announcement date. The dependent variable BANKRUPT is an indicator variable that equals 1 if firms file for bankruptcy within 1 year following the fiscal year-end date and is 0 otherwise. A positive coefficient on AFED (or AFE_%) would indicate low Type I errors i.e., firms that receive a going-concern opinion tend to file for bankruptcy subsequently.
Panel A of Table 6 presents the results for estimating model 3a. The coefficients on the audit committee accounting expertise variables are positive and significant (AFED at 10 percent level and AFE_% at 5 percent level) indicating that Type I errors are lower for firms with accounting experts on the audit committee. Coefficients on other audit committee variables are not significant. Panel B of Table 6 presents results for estimating model 3b. Neither of the coefficients on audit committee expertise variables is significant nor are the coefficients on other audit committee variables. This evidence indicates that audit committee characteristics do not play a significant role in minimizing Type II errors. This finding should be considered in the context of the small sample size in panel B, as it considers only firms that file for bankruptcy following the issue of going-concern opinions. Overall, the evidence in Table 6 suggests that when auditors provide a going-concern opinion, audit committees with accounting experts ensure the accuracy of such opinions while in cases where no going-concern opinions are given by the auditor, the audit committees do not appear to question the auditor as to whether a going-concern opinion is warranted. In this narrow sense, accounting experts on audit committees promote audit quality in the post-SOX period.

Supplemental Analyses
Several additional analyses are performed to assess the robustness of the results.
The number of accounting experts: The variables used to represent accounting experts on the audit committee (AFED and AFE_%) do not consider the actual number of experts on the committee. To address the effect of having multiple experts on the audit committee, the analyses are conducted by using indicator variables that measure having only one accounting expert or more than one accounting expert. The results (not reported) show that having a single or multiple accounting experts does not alter the main results reported.
Supervisory financial expertise: Prior literature on audit committee accounting expertise (Krishnan & Visvanathan, 2008) also considers the role of supervisory financial expertise, defined as experience as the chief executive officer or as president of a for-profit corporation. When AFED is replaced by the presence of a supervisory financial expert, the results are similar. Note that in general, prior literature has not found significant associations with supervisory financial expertise in contexts where accounting expertise was found to be significant.
CEO influence: Lisic et al. (2019) examine whether CEO influence over the audit committee compromises its oversight role of auditors. They build upon the evidence in Cassell, Myers, Schmardabeck, and Zhou (2018) who argue that less co-opted audit committees are more effective at monitoring, and Coles, Daniel, and Naveen (2014) who argue that directors who join the board after CEO assumed office are less likely to be good monitors. Based on these studies, CEO influence is measured as the proportion of audit committee members who joined the board after the CEO taking office. The sample is partitioned into above and below the median values of the CEO influence variable and the analyses are conducted on the partitioned sample. The results (not reported) show that there is no significant difference in the coefficients on audit committee variables between the two partitions. One potential explanation for the lack of significance for CEO influence is monitoring by other executives. For example, Li (2014) finds beneficial effects of mutual monitoring by No. 2 executive for firms with weak corporate governance.
Endogeneity: Factors that influence the decision to have an accounting expert on the audit committee are likely correlated with factors influencing the auditors' decision to issue a going-concern opinion or the audit committee's decision to protect the auditor from dismissal. Thus, there is a potential endogeneity issue in studying audit committee accounting expertise and auditor going-concern opinions. To address the endogeneity concerns on audit committee accounting expertise, Krishnan and Visvanathan (2008) propose a procedure to first estimate a model using determinants of audit committee accounting expertise and then using the predicted value from this first-stage model instead of AFED or AFE_%. Their model however includes several variables that are likely correlated with the decision to issue a going-concern opinion such as profitability, earnings volatility, age, and sales growth. Instead, the following governance variables are identified as instrumental variables, namely board size, audit committee size, and the number of public company boards on which the audit committee members sit, as these have no theoretical relationship with the decision to issue a going-concern opinion. With the substitution of these variables, the first-stage model is estimated by using AFED as the dependent variable. In this model, all the instrumental variables are significant. When the predicted value from the first-stage model is used instead of AFED in the analyses in tables 4-6, the results (not reported) are similar. Note that the procedure described here addresses the endogeneity issue specific to the context of accounting expertise in the audit committee. More broadly, unobservable firm and executive characteristics may drive the endogeneity problem in empirical models that use corporate governance variables (Coles and Li, 2020).

Conclusion
This study revisits the relationship between going-concern opinions and audit committee characteristics including accounting expertise. The study finds no significant association either between the likelihood of a going-concern report or dismissals following such a report and audit committee accounting expertise and other audit committee characteristics. The methodology uses financially distressed firms for a recent period (2016-18) and considers board, audit committee, and auditor characteristics in addition to audit committee accounting expertise. There are several explanations for why the results of the study differ from the findings in prior studies. First, prior studies (Carcello & Neal 2000, 2003 examined periods before the passage of SOX or the period immediately following SOX (Hoitash & Hoitash, 2009). Second, the sample composition of studies that find a significant role for accounting expertise of audit committees (Lisic et al., 2019, Hoitash & Hoitash, 2009 is not restricted to financially distressed firms only as is the case for studies examining going-concern opinions. Third, Carcello & Neal (2000) do not include several board and audit committee characteristics that have been found to enhance accounting and audit quality by subsequent literature. Finally, as noted in Newton et al. (2016), in recent years firms are less likely to engage in opinion shopping for going-concern opinions, implying that audit committees do not need to intervene to "protect" the auditors in such cases as was argued in prior literature.
In contrast to prior studies, the results also show that while audit committees do not appear to play a significant role in the auditor's decision in not issuing a going-concern report or subsequent auditor dismissals, the accounting expertise of the committee improves the accuracy of the going-concern opinion by reducing Type I errors. That is, audit committees do play a role in improving audit quality but through a different channel than what was found in prior studies. In sum, the results in the study further our understanding of how audit committee accounting expertise promotes audit quality by showing how this relationship differs based on sample composition and period.
The study is subject to the following limitations. First, while the sample considers a recent period, it encompasses only 3 years. Though the size of the sample is reasonable in comparison to prior studies (Carcello & Neal, 2000), a longer period and a larger sample may yield additional insights. Second, the endogeneity model considered in the study could not rule out all sources of endogeneity endemic to models that use corporate governance variables. Third, the study does not consider other external monitoring mechanisms that may substitute for or accentuate the role of corporate governance, such as product market competition (Giroud and Mueller, 2011) and stock compensation (Core and Guay, 1999). Future research can examine whether the significance of accounting experts in audit committees varies based on the competitive nature of the industry. Finally, Accounting Standards Update (ASU) 2014-15 issued by the Financial Accounting Standards Board (FASB) in 2014 requires management of companies to evaluate whether there is substantial doubt about the entity's ability to operate as a going concern and to make required disclosures. While early evidence on disclosures under this rule (Visvanathan, 2021) indicates that management disclosures generally accord with auditor going-concern opinions, with more years and data, future research can explore if audit committees play a pivotal role in cases where the disclosures by management and auditor going-concern opinions are not aligned. ISSN 1927-5986 E-ISSN 1927

Copyrights
Copyright for this article is retained by the author(s), with first publication rights granted to the journal.
This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).