Auditor Fees and Going-Concern Reporting Decisions on Bankrupt Companies: Additional Evidence Free

Between the years 2009 and 2013, while audit revenues at the Big 4 audit firms grew by only $3 billion, revenues from non-audit services (NAS) increased $14 billion (27 percent) to $65 billion (Harris 2014). The spike in NAS provided by Big 4 auditors has not gone unnoticed by U.S. auditing regulators. Several board members of the Public Company Accounting Oversight Board (PCAOB) have recently expressed concerns that the expansion of NAS within the major audit firms threatens auditor independence.¹ Reduced independence can lead to audit failure including where bankrupt companies did not receive going-concern (GC) audit opinions prior to their collapse.

The recent concerns of PCAOB board members regarding the possible effect on auditor independence from the growth in NAS echo those of legislators following several high-profile accounting scandals (e.g., Enron) at the start of the millennium (U.S. Senate 2002; U.S. House of Representatives 2002a). In the wake of Enron and the ensuing congressional hearings, the Sarbanes-Oxley Act of 2002 (SOX) (U.S. House of Representatives 2002b) was enacted making it unlawful for auditors to provide several types of NAS to their attest clients. However, SOX (U.S. House of Representatives 2002b) placed no restrictions on supplying other types of potentially lucrative NAS, including tax compliance and tax planning, merger and acquisition consulting, and employee benefit plan audits.

While much of the debate concerning independence has focused on NAS fees, several researchers (DeFond, Raghunandan, and Subramanyam 2002; Basioudis, Evangelos, and Geiger 2008) argue that higher total fees, irrespective of their classification, also may threaten auditor independence. Because total fees comprise NAS fees as well as fees for audit services, I examine in this paper the impact of NAS fees and audit service fees on the propensity of auditors to issue GC audit opinions to U.S. companies prior to filing bankruptcy.

My examination focuses on the auditor fees paid by 203 publicly held U.S. companies that filed bankruptcy in the years 2002–2013. While Callaghan, Parkash, and Singhal (2009, with U.S. data, find no relation between auditor fees and GC decision making earlier in the millennium, examining more recent data using a large sample of bankrupt companies is worthwhile given the concerns of regulators about the threat to auditor independence from the growth in NAS at the major audit firms.² In analysis, I find no significant association between the propensity of auditors to issue a GC opinion to a soon-to-be bankrupt firm and NAS fees and audit fees. However, I find evidence that firms that retain their incumbent auditor for a financial statement audit for the year subsequent to filing bankruptcy are significantly less likely to be issued a GC audit opinion prior to filing bankruptcy. While this result suggests that auditor expectations of future benefits reduce independence and impair current GC decision making, additional analysis in this paper suggests an alternative explanation.

The remainder of the paper is organized as follows. The next section discusses the background and presents the study's hypotheses. I then describe the method and related analysis. Following the results, the paper ends with a summary and conclusions.

The concept of auditor independence includes auditor objectivity and the ability to withstand client pressure to engage in substandard reporting (DeFond et al. 2002). Client pressure may come from managers who face the loss of lucrative compensation packages, as well as possible tainting of their professional reputation for being associated with a failed business. Auditors have economic incentives that threaten their independence as well as market-based institutional incentives to act independently. Market-based incentives that relate to reputation and litigation costs are well documented in the literature (DeFond et al. 2002). Economic incentives to issue an audit opinion unmodified for GC uncertainties relate to the monetary benefits from client services provided. A crucial assumption is that auditors are inclined to sacrifice their independence and be less objective in their audit reporting when the magnitude of their service fees creates economic bonding with the client (Simunic 1984).

Several U.S. studies examine the association between NAS fees and GC audit opinions. DeFond et al. (2002) and Geiger and Rama (2003) find no evidence of an association between auditors' issuance of GC opinions to financially distressed clients and NAS fees. More recently, Callaghan et al. (2009) find no evidence of a relation between auditor fees and GC reporting decisions for bankrupt companies.³

While auditors may now have fewer opportunities to generate NAS fees, what is unclear is whether auditors in audit engagements that include the supply of allowable NAS are more inclined to place a higher value on retaining such clients and not jeopardize the loss of revenue by issuing a GC opinion. Such behavior seems possible given the reemergence of NAS at the major audit firms and the possible pressure on partners to grow revenues and profits (Wyatt 2004; Hermanson 2009; Harris 2014).⁴ Alternatively, it could be argued that increased regulatory oversight of auditors following the high-profile accounting scandals at the beginning of the millennium make it more likely that audit reporting on subsequently bankrupt clients would not be compromised in audit engagements where NAS are provided by auditors. Hence, I present my first hypothesis in its null form:

• H1: 

  There is no significant relationship between GC audit opinions issued prior to bankruptcy and NAS fees for the opinion year.

Although much of the debate regarding independence and audit quality has focused on NAS fees, prior research (e.g., DeFond et al. 2002; Whisenant, Sankaraguruswamy, and Raghunandan 2003; Basioudis et al. 2008) notes that any examination of the influence of NAS fees must also include concurrent testing of the impact of audit service fees. Some earlier studies (Geiger and Rama 2003; Basioudis et al. 2008) find a positive association between audit fees and GC decisions on financially distressed firms. More recently, Blay and Geiger (2013) find audit fees are significantly positively related to GC decisions after controlling for potential simultaneity bias induced by endogeneity. However, Robinson (2008) and Callaghan et al. (2009) find no evidence that audit fees are significantly related to the likelihood of a client receiving a GC opinion prior to filing bankruptcy. In view of the inconclusive results, I present my second hypothesis in its null form:

• H2: 

  There is a no significant relationship between GC audit opinions issued prior to bankruptcy and audit service fees for the opinion year.

My GC empirical model to examine the association between auditor fees and the issuance of GC audit opinions to financially distressed client companies shortly before filing bankruptcy is based on the model employed by Callaghan et al. (2009). However, there are a couple of differences. Since my period of examination includes the global financial crisis (GFC) and Geiger, Raghunandan, and Riccardi (2014) find an increase in auditor GC reporting after the onset of the GFC, I include an indicator variable, CRISIS, for audit opinions issued during the GFC. I also include an indicator variable, FUTURE AUDIT, for classifying companies known to have retained their auditor for a financial statement audit for the year subsequent to filing bankruptcy. Hence, I use the following logistic regression model:

The variables are defined as follows:

• OPINION = 1 if a going-concern audit opinion issued within a year prior to filing bankruptcy, else 0;

• PROBANKZ = the probability of bankruptcy indicated by the Altman (1968) Z-score;

• LN(ASSETS) = natural logarithm of total assets at fiscal year-end in millions of dollars;

• LN(AGE) = natural logarithm of the number of years since firm was listed on a stock exchange;

• RETURN = the firm's stock return over the opinion year;

• VOLATILITY = the variance of the residual from the market model in the opinion year;

• ROA = the firm's return on assets over the opinion year;

• LEV = total liabilities over total assets at fiscal year-end;

• CLEV = the change in LEV during the opinion year;

• LLOSS = 1 if the firm has a net loss in the year prior to the opinion year, else 0;

• INVESTMENTS = short- and long-term investments (including cash and cash equivalents) deflated by total assets at fiscal year-end;

• FUTURE FINANCE = 1 when firm issues debt or equity in year after opinion year; else 0;

• BIG4 = 1 if Big 4 auditor, else 0;

• OP CASH FLOW = operating cash flows divided by total assets at fiscal year-end;

• BKRPTCY_LAG = the number of days between the audit report date and bankruptcy filing date;

• DEFAULT = 1 if firm in technical or payment default, else 0;

• AFTER = 1 if audit opinion dated January 1, 2002 through December 31, 2003, else 0;

• CRISIS = 1 if audit opinion dated between September 1, 2008 through June 30, 2009, else 0;

• LIQUIDATE = 1 for firms known to have liquidated after filing bankruptcy, else 0;

• FUTURE AUDIT = 1 if client retained incumbent auditor for an audit of the financial statements for the year subsequent to filing bankruptcy, else 0;

• LN(TOTAL FEE) = natural log of audit and non-audit fees paid to the auditor;

• FEERATIO = the ratio of non-audit fees to total fees paid to the auditor;

• LN(AUDIT FEE) = natural log of audit fees paid to the auditor; and

• LN(NAS FEE) = natural log of non-audit service (NAS) fees paid to the auditor.

The continuation of an entity as a GC is assumed in financial reporting unless significant information to the “contrary” exists. Several control variables in the model used by Callaghan et al. (2009) include “contrary” factors identified in Statement on Auditing Standards (SAS) No. 59 (AICPA 1988). For example, a contrary factor that suggests a GC audit opinion is appropriate is financial distress, represented in the model by PROBANKZ. Other factors suggesting a GC audit opinion is appropriate include LN(AGE), RETURN, and VOLATILITY. Firms with lower Z-scores (PROBANKZ), have been recently listed (LN(AGE)), and that have lower stock returns and higher stock volatility are at greater risk for bankruptcy.

Other indicators of bankruptcy risk include higher leverage (LEV), an increase in leverage (CLEV) over the opinion year, reporting a loss in the prior year (LLOSS), violating debt covenants (DEFAULT), poor return on assets (ROA), and weak operating cash flow (OP CASH FLOW). Also, because Mutchler, Hopwood, and McKeown (1997) find that GC audit opinions are less likely as the number of days between the audit report date and the bankruptcy filing date increase, I include a bankruptcy lag (BKRPTCY_LAG) variable and expect its coefficient to be negative. I also include a control variable for auditor type (BIG4) since Big 4 auditors may have more to lose from class-action lawsuits arising from not issuing a GC opinion prior to a client's bankruptcy filing (Geiger and Rama 2003).

SAS No. 59 (AICPA 1988) also identifies several factors that “mitigate” the probability of bankruptcy thereby decreasing the chances of a GC opinion. Larger companies, represented by LN(ASSETS), have more leverage with their creditors in the event of a slowdown in business. Additionally, to the extent companies are more liquid (INVESTMENTS) or generate liquidity through new financing (FUTURE FINANCE), they are more likely to avoid bankruptcy. I winsorize continuous variables at the 1st and 99th percentiles to reduce the influence of extreme observations.

Following Callaghan et al. (2009), I include the variable AFTER because Geiger, Raghunandan, and Rama (2005) find that audit opinions issued in the years 2002–2003 to bankrupt companies were more likely to be GC modified due to media attention and congressional scrutiny from the accounting scandals (e.g., Enron) at the start of the millennium, compared to opinions issued in the preceding period (2000–2001).⁵ Since my sample includes audit opinions for which the auditor signature date is the year 2002 or 2003, I include AFTER for opinions dated in 2002 or 2003 and I expect its sign to be positive. Also, following Callaghan et al. (2009), I include an indicator variable (LIQUIDATE) for classifying companies known to have liquidated after filing bankruptcy, and I expect its coefficient to be positive.

Geiger et al. (2014) examine the period before and after the start of the global financial crisis (GFC) using data for distressed U.S. companies. While Geiger et al. (2014) do not examine the possible effect of auditor fees on GC audit decisions, they do document an increase in overall GC reporting for both Big 4 and non-Big 4 auditors after the start of the GFC. In addition, Audit Analytics (2011) in analysis of GC opinions issued annually over the years 2000–2010 reports more GC opinions in 2008 than in any other year for the decade. Also, Audit Analytics (2011) finds that GC opinions, as a proportion of total opinions issued, were at their highest level (21 percent) in 2008. Since my examination period also includes the GFC, I include the variable CRISIS for audit opinions dated between September 1, 2008 and June 30, 2009, and I expect its coefficient to be positive, given the findings of Geiger et al. (2014).⁶

While the relation between current auditor fees and GC audit decisions has been the focus of much prior research (e.g., DeFond et al. 2002; Geiger and Rama 2003; Callaghan et al. 2009), recent evidence from Blay and Geiger (2013) suggests that GC audit decisions also may be influenced by auditor expectations of future economic benefits. In results consistent with reduced auditor independence, Blay and Geiger (2013) find that auditors' current GC decisions are negatively associated with total fees paid to them by the audit client in the subsequent two years.⁷ Given the findings of Blay and Geiger (2013), I include an indicator variable, FUTURE AUDIT, for classifying bankrupt firms known to have retained their auditor for a financial statement audit for the year subsequent to filing for bankruptcy. Since auditors may be inclined to refrain from issuing a GC audit opinion to those distressed clients that they hope to retain their incumbent status, I expect the coefficient on FUTURE AUDIT to be negative.

I search BankruptcyData.com to identify public companies that filed bankruptcy from 2002 through 2013 for which the audit opinion date is after December 31, 2001. Prior research (Geiger et al. 2005; Feldmann and Read 2010) finds that auditors' reporting behavior changed in the wake of the tumultuous events of late 2001 (e.g., Enron scandal, congressional hearings). My search of BankruptcyData.com identified 1,435 publicly held bankruptcies during the years 2002–2013. This initial sample was reduced by 404 companies not listed in Compustat and by 102 companies in the financial services industries (SIC 60–69). Data requirements in CRSP and Audit Analytics result in a final sample of 203 financially distressed companies⁸ for which an audit opinion was issued in the year prior to filing bankruptcy.⁹ Panel A of Table 1 presents the distribution of the sample and audit opinion type by bankruptcy filing year.

Panel B of Table 1 presents descriptive statistics for the sample. In contrast to prior research (e.g., Geiger and Rama 2006; Callaghan et al. 2009), which finds less than half of failed companies are issued a GC opinion prior to bankruptcy, I find that 101 (50 percent) companies in my sample were issued a GC opinion in the year preceding bankruptcy. Panel B also reveals a smaller proportion (63 percent) of bankrupt companies audited by Big 4 auditors than reported in prior research (e.g., Robinson 2008; Callaghan et al. 2009). This finding is consistent with audit firm realignment in recent years, particularly switches from Big 4 to non-Big 4 auditors (e.g., Rama and Read 2006; Hogan and Martin 2009).

Table 2 presents the univariate analysis for the variables in Panel B of Table 1 classified by audit opinion type. As expected, the mean and median PROBANKZ scores are significantly lower for the GC group, and the mean and median DEFAULT values are significantly higher for companies that were issued a GC opinion. In addition, the mean and median values for RETURN and VOLATILITY suggest GC companies have significantly lower returns and greater volatility than the non-GC group. Additionally, companies receiving a GC opinion are significantly more leveraged (LEV) and have a significantly larger increase in leverage (CLEV). Further, companies not issued a GC modification prior to bankruptcy have a significantly higher mean score for the variable FUTURE AUDIT.¹⁰

Regarding auditor fee variables, Table 2 shows significant differences (p ≤ 0.05) in the mean and median values for AUDIT FEE and TOTAL FEE between the two audit opinion groups. However, the mean and median values of FEERATIO for non-GC companies are not significantly different from the GC group. While these results, with respect audit service fees and total fees, suggest reduced auditor independence, I rely on multivariate analysis to test my hypotheses.

In Table 3, Models 1–4 show the following combinations of fee variables: both LN(NAS FEE) and LN(AUDIT FEE); LN(TOTAL FEE) alone; FEERATIO alone; and both LN(TOTAL FEE) and FEERATIO, respectively.¹¹ The inclusion in Model 1 of both LN(AUDIT FEE) and LN(NAS FEE) concurrently tests both of my hypotheses. In Model 1, the finding that LN(NAS FEE) and LN(AUDIT FEE) are insignificant supports my first (second) hypothesis regarding no significant association between NAS fees (audit service fees) and the likelihood of a distressed company receiving a GC audit opinion prior to filing bankruptcy.

In Table 3, the finding that the FEERATIO is insignificant, either alone (Model 3) or with controlling for total fees (Model 4), is consistent with the result in Model 1 for the variable LN(NAS FEE). The results of Model 2, testing the relation between the OPINION and LN(TOTAL FEE), indicate there is no significant relation between fees paid to the auditor for total services and GC decisions. Of interest is the finding that in all combinations of fee variables, the negative coefficient on FUTURE AUDIT is highly significant (p-value < 0.01).¹²

I perform several additional analyses. First, I examine the effect of unexpected fees.¹³ In untabulated analysis using residuals from the fee-determinant models, I find no significant relation between GC decisions and the unexpected fee variables, although the negative coefficient on the indicator variable FUTURE AUDIT remains significant. Also, after controlling for potential bias induced by endogeneity, I find no evidence of a significant association between NAS fees and audit service fees and GC opinions, providing support for both H1 and H2.

As noted, I restrict analyses to a financially distressed sample of firms that exhibit at least one of the following: (1) negative cash flow from operations, (2) negative working capital, (3) a retained earnings deficit, or (4) a bottom-line loss. As part of the additional analysis, I identify from my sample a subsample of 154 highly financially distressed firms that exhibit at least three of these four stress signals. Results of multivariate (untabulated) analyses based on the highly distressed subsample are substantially similar to those reported using the full sample. Specifically, the various fee variables continue to be insignificant at conventional levels and I continue to find a highly significant negative coefficient on the variable FUTURE AUDIT.¹⁴

Finally, with respect to the significant negative relationship between FUTURE AUDIT and the OPINION, one interpretation is that this result is driven by reduced auditor independence. That is, the auditor's GC decision shortly before the client's bankruptcy filing is adversely influenced by the auditor's interest in generating economic benefits subsequent to the client's bankruptcy. However, an alternative explanation emerges following a partitioning of the full sample of 203 bankrupt firms into those for which FUTURE AUDIT equals 1 (n = 67) versus those where it is 0 (n = 136). Table 4, which presents a univariate analysis for several of the study's control variables classified by the FUTURE AUDIT designation, shows that the mean and median scores for PROBANKZ, CLEV, OP CASH FLOW, ROA, and LLOSS are relatively more favorable for the FUTURE AUDIT subsample of companies compared to those of the non-FUTURE AUDIT group.¹⁵ Table 4 also shows the FUTURE AUDIT subsample is larger (total assets) and has a significantly greater likelihood of obtaining future financing (FUTURE FINANCE) in the year subsequent to filing bankruptcy, based on a comparison of means. Since the firms comprising the FUTURE AUDIT group are relatively larger, less financially distressed, and more likely to be able to generate liquidity by issuing new debt or equity securities than those in the non-FUTURE AUDIT group, their auditors may have judged them as having a better chance of avoiding bankruptcy prior to the Chapter 11 filing and, thus, chose to not issue a GC audit opinion.

The recent rise in non-audit services (NAS) provided by Big 4 auditors has the attention of auditing regulators. Several board members of the Public Company Accounting Oversight Board (PCAOB) have recently expressed concerns that the growth in NAS within the major firms threatens auditor independence (Harris 2014; Tysiac 2014). Reduced independence can result in audit failure including situations where companies were not issued going-concern (GC) audit opinions from their auditors prior to filing bankruptcy.

In this paper, I examine the relationship between NAS fees (and audit fees) and the propensity of auditors to issue GC opinions to public companies prior to filing bankruptcy in the period 2002–2013. In main analysis, I find no evidence of a significant relationship between the propensity of auditors to issue a GC audit opinion to a subsequently bankrupt client and fees for audit and NAS. These findings may be particularly timely given the concerns expressed by regulators regarding the expansion of NAS at major audit firms and the threats such expansion may pose to auditor independence. In addition, I find that firms that retain their auditor for a financial statement audit for the year subsequent to filing bankruptcy are significantly less likely to be issued a GC opinion prior to filing Chapter 11. While on one hand this result is consistent with reduced auditor independence, on the other hand results from additional analysis suggest firms that are relatively less distressed may be less likely to receive extensive auditor consideration for a GC audit opinion. Recent findings from Blay and Geiger (2013) suggest that highly financially distressed firms are more likely to receive extensive consideration of a GC audit opinion from their auditor than are “grey area” client firms that are less financially distressed.

The findings from this study suggest fruitful areas of future research. For example, research could be pursued to enhance understanding of how “grey area” distressed companies are different from highly financially distressed firms across various financial factors. Evidence in this regard might be useful for improving the quality of auditor GC decision making. Also, while this study and Callaghan et al. (2009) have examined whether “Type II misclassifications” are associated with fees for NAS and for audit services, future research might examine whether NAS fees and audit service fees are related to the incidence of “Type I misclassifications” (GC audit opinions issued to companies that remain viable).

Finally, some limitations should be noted. Given the modest number of bankrupt firms that were identified as having engaged their incumbent auditor for a post-bankruptcy financial statement audit, care should be used when evaluating my findings with respect to the variable FUTURE AUDIT. A further limitation of this study, particularly since my period of examination includes the recent GFC, is the exclusion from the sample of banks and other firms in financial services. Prior studies (e.g., Geiger et al. 2014) in this area generally have not included such firms in analysis because of their unique financial characteristics. Hence, another opportunity for future research would be an investigation into the effect of auditor fees on GC reporting decisions for clients in financial services.