Discussion of Auditor Litigation and Modified Reporting on Bankrupt Clients

(1)

Accounting Research Center, Booth School of Business, University of Chicago

Discussion of Auditor Litigation and Modified Reporting on Bankrupt Clients Author(s): Timothy B. Bell

Source: Journal of Accounting Research, Vol. 32, Studies on Accounting, Financial Disclosures, and the Law (1994), pp. 31-38

Published by: Wiley on behalf of Accounting Research Center, Booth School of Business, University of Chicago

Stable URL: http://www.jstor.org/stable/2491437 Accessed: 27-06-2016 07:58 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley, Accounting Research Center, Booth School of Business, University of Chicago

are collaborating with JSTOR to digitize, preserve and extend access to Journal of Accounting Research

(2)

Journal of Accounting Research Vol. 32 Supplement 1994

Printed in US.A.

Discussion of

Auditor Litigation and Modified Reporting on Bankrupt Clients

TIMOTHY B. BELL*

1. Introduction

The Palmrose and Carcello (henceforth P&C) paper uses a sample of 655 public companies that declared bankruptcy between 1972 and 1992 to examine whether modified audit reports issued prior to bankruptcy serve to protect auditors from the effects of litigation against bankrupt

clients.1 The authors find that auditor litigation was absent for a major-

ity (82%) of the sampled companies; unmodified audit reports were issued for 64% of the sampled companies with auditor litigation.

Modification of audit reports and incidence of auditor litigation are not independent-58% of the no litigation subsample had modified audit reports, compared to 36% of the auditor litigation subsample. After controlling for the existence of apparent irregularities, whether the client reported a profit just before bankruptcy, client size, and other factors, modified audit reports are not incrementally significant in a multivariate analysis of the incidence of auditor litigation.

The authors have documented several empirical regularities that demonstrate the complexity of auditor litigation, and they provide in- sights about why modifications are not significant factors in explaining the incidence of auditor litigation after controlling for other factors. My comments are intended to supplement the authors' observations and results in the hope of shedding additional light on the empirical findings.

*Director, Assurance Services, KPMG Peat Marwick.

Henceforth, I refer to this as the "protection hypothesis."

31

Copyright ?, Institute of Professional Accounting, 1995

(3)

32 JOURNAL OF ACCOUNTING RESEARCH, SUPPLEMENT 1994

2. Bankruptcy, Fraud, and the Protection Hypothesis

2.1 STRATEGIC BANKRUPTCY

Following passage of the Bankruptcy Reform Act of 1978, bankruptcy and its attendant litigation increasingly have been used by corporate management and commercial creditors to accomplish wealth and risk transfers (see, for example, Delaney [1992]). Marsh and Cheng [1985]

suggest that approximately 19% of business bankruptcy filings between 1978 and 1983 were due to the 1978 broadening of the bankruptcy law.

Much of the bankruptcy research recently reported in the finance and accounting literature, including this paper, appears to subscribe to what Delaney calls the economic fundamentalist stance-the assumption that managers have no choice when they enter Chapter 11. If this assumption is false, i.e., if bankruptcy is often not a passive reaction to economic im- peratives, inferences drawn from empirical tests of hypotheses related to the incidence of bankruptcy should be cautiously interpreted.

Strategic bankruptcy may help explain why litigation does not occur for most bankrupt public companies. If bankruptcy is a strategic choice, both management and creditors may prefer to report negative results that will cast doubt on the company's ability to continue as a going concern. This conservative reporting may also reduce the risk of litigation to auditors. Delaney [1992] presents in-depth analyses of three widely publicized cases (Johns-Manville, Continental Airlines, and Texaco) to illustrate how organizations invoke bankruptcy to avoid current financial burdens and shift future financial risk to other parties. A summary of the facts and circumstances surrounding one of these cases, Johns- Manville, is presented below.

Johns-Manville filed for bankruptcy in 1982 to protect the company and its commercial creditors from financial risks related to asbestos lawsuits. According to Delaney [1992, p. 169]:

Morgan Guaranty, Manville's lead lender for decades, gradually became convinced that it stood a better chance of getting its money back with a Chapter 11 filing than by allowing Manville to fight individual claims in court. Morgan had a well-placed representative on Manville's board of directors. He spearheaded a drive to hire a new auditor, which in turn immediately approved an SEC filing estimating future asbestos liabilities for the very first time. This paved the way for the designation "bankrupt" and a Chapter 11 filing only weeks later.

The predecessor auditor, Coopers & Lybrand, had issued a qualified

audit report in 1981 when, pursuant to FAS No. 5, Manville had dis-

closed in a footnote to its annual report that damages for the asbestos- related lawsuits could not be accurately estimated. In 1982, Manville hired an outside consulting firm to prepare an estimate of the total cost of the asbestos claims. Manville then replaced Coopers & Lybrand with Price Waterhouse, who approved the SEC filing by Manville that

stated the future liability could be accurately estimated and booked

under FAS No. 5. Having booked the approximate $2 billion liability,

(4)

AUDITOR LITIGATION AND REPORTING ON BANKRUPT CLIENTS 33

Manville could claim "equity insolvency" in a voluntary Chapter 11 filing. The end result of the Johns-Manville Chapter 11 reorganization was that commercial creditors were granted full payment on the money owed to them and uncertainty about the asbestos-related tort liability was removed.

Absent detailed information about the circumstances of the Johns- Manville bankruptcy and the related qualified audit report, we would likely categorize this case as one where opinion modification protected the auditor from litigation. However, it is not apparent that auditor protection was ever at issue; rather, Johns-Manville made a strategic choice to file bankruptcy and wanted a qualified report with the con- tingent liability booked to ensure the success of its strategy.

The Johns-Manville case highlights the importance of understanding the detailed facts and circumstances surrounding bankruptcy and litigation when designing empirical tests about bankruptcy.2 Empirical tests of the protection hypothesis should be performed on a sample con- taining only cases where auditor protection is at issue, i.e., cases where bankruptcy was not a strategic choice but rather an economic impera- tive. As a practical matter, obtaining a thorough understanding of the facts and circumstances for each sample company would be prohibi- tively costly and sampling in this manner could introduce researcher bias into the experimental design due to the subjective nature of the task. However, including strategic bankruptcies in the sample used to test the protection hypothesis represents a limitation of the current paper's research design. It is analogous to testing a vaccine's ability to protect against a virus using a sample that includes subjects that have not been exposed to the virus. Estimates of the relative frequency of "protection" are not accurate.

2.2 FINANCIAL IRREGULARITIES

P&C report that irregularities were involved in some of their sample cases. It is questionable whether a sample used to test the protection hypothesis should include cases where litigation implicates the auditors for failing to find material irregularities. In those cases, plaintiffs are seeking recovery from losses caused by fraud, not bankruptcy.

Opinion modification and auditor protection are not at issue since, by definition, auditors failed to discover and report the fraud.3

2 Strategic bankruptcies would also have implications for the design of research aimed at estimating failure prediction models and comparing model predictions to auditor judgments about a client's ability to continue in existence.

3 SAS No. 53 (paragraph 26) states that "if the auditor has concluded that the financial statements are materially affected by an irregularity, the auditor should insist that the financial statements be revised and, if they are not, express a qualified or an adverse opinion on the financial statements, disclosing all substantive reasons for his opinion."

Opinion modifications for going-concern and other uncertainties are not allowed under generally accepted auditing standards as a vehicle for auditor reporting of irregularities.

(5)

34 TIMOTHY B. BELL

Panel A: Relative Frequencies of Irregularities and Opinion Modifications 452 No Litigation 80 Auditor Litigation

(85%) (15%)

4 Irregularities 448 No Irregularities 28 Irregularities 52 No Irregularities (1%) (99%) (35%) (65%)

1 Modified 257 Modified 3 Modified 22 Modified (25%) (57%) (11%) (42%)

Panel B: Decomposition of the 500 Cases Not Involving Irregularities 448 No Litigation 52 Auditor Litigation

(90%) (10%)

84 Profit 364 Loss 22 Profit 30 Loss (19%) (81%) (42%) (58%)

21 Modified 236 Modified 2 Modified 20 Modified (25%) (65%) (9%) (67%) FIG. 1.-Decomposition of the sample of 532 cases used in the multivariate analysis of the protection hypothesis. The sample consists of the 532 auditor litigation/no litigation cases used by P&C for the multivariate analysis reported in their table 3. Panel A shows that no irregularities were involved for 500 of the 532 sample cases. The opinion modification rate of 57% for the no litigation/no irregularities subgroup was significantly higher than the opinion modification rate of 42% for the auditor litigation/no irregularities subgroup (chi-square = 4.28, p = .038). Panel B shows that 42% of the 52 cases involving auditor litigation reported profits before bankruptcy, compared to 19% of the 448 no litigation cases (chi-square = 15.48, p = .0001). Opinions were modified for 9% of the auditor litigation cases reporting prebankruptcy profits, compared to 25% of the no litigation cases reporting prebankruptcy profits (chi-square = 2.60, p = .11). Opinion modifications were issued for roughly two-thirds of both the no litigation and auditor litigation subsamples. Data were provided by the authors.

Panel A of figure 1 shows that 88% of the 32 sample cases involving irregularities also involved auditor litigation,4 while irregularities and opinion modifications jointly occur for only 4 of the 532 sample observations used to test the protection hypothesis. For some or all of the 32 irregularities cases, management and commercial creditors chose to

4 These frequencies were obtained from the authors and are not reported in their paper. The 32 irregularities were from the sample of 452 "no litigation" and 80 "auditor litigation" cases used in the multivariate analysis of the protection hypothesis that excluded the "other litigation" cases. Observations were deleted from the original sample due to missing data required for computation of the financial distress index.

(6)

have the companies voluntarily file for Chapter 11 protection of their interests, even though the client's ability to continue as a going concern was not at issue.

The recent fraud and bankruptcy at The Leslie Fay Companies, Inc.

(LF) is a case in point. BDO Seidman, LF's auditor, discovered financial irregularities in late January 1993. In early February 1993, LF disclosed that the irregularities could wipe out reported profits for 1992 and 1991. LF's stock price plunged from $12 to $4 per share, and at least a dozen shareholder suits were filed against LF and BDO Seidman. In March 1993, LF's CEO reported that estimates of the effects of the irregularities on the company's net worth indicated limited damage. LF's management and its commercial creditors entered negotiations about extension of lines of credit, and in April 1993 LF made a strategic choice to file a Chapter 11 petition, stating it had arranged a $100 million debtor-in-possession line of credit with Citibank to fund continu- ing operations. In May 1993, BDO Seidman stepped down after having been advised by the SEC that there was an independence problem, given that LF and BDO Seidman were joint defendants in shareholder lawsuits. Arthur Andersen was appointed the successor auditor.

Without giving consideration to the facts and circumstances related to the LF case, we would likely categorize the case as "auditor litigation-no modification-followed by bankruptcy." However, commercial creditors were willing to extend lines of credit for working capital after the irregularities were revealed, indicating they believed the company was still viable as a going concern. Apparently, bankruptcy was strategi- cally employed to protect creditors' claims from litigation. Auditor protection from litigation related to bankruptcy was not at issue at the time litigation was initiated, since LF had not yet filed bankruptcy.

Although P&C do not report plaintiffs' allegations for the sample cases involving irregularities, they do report that auditor litigation precedes bankruptcy for 55% of the sample cases exhibiting auditor litigation. My discussions with the authors indicate that a large number of these early litigation cases also involved irregularities. Therefore, it seems reason- able to assume that, for some of these cases, litigation implicated the auditor for failing to discover fraud, not for failing to signal impending bankruptcy. A question arises about whether, within the context of bankruptcy and the protection hypothesis, such cases should be classified as "no litigation," since the lawsuits occurred prior to bankruptcy.

3. Tests on the Subsample Exhibiting No Irregularities

In this section, I report results of tests of the incidence of auditor litigation using the subsample of companies exhibiting no irregularities. Data required to perform these tests were provided by the authors.

These test results shed some light on why opinion modification was not incrementally significant in the multivariate analysis reported by P&C.

(7)

36 TIMOTHY B. BELL

Panel A of figure 1 shows that 57% of the 448 sample cases involving no irregularities and no auditor litigation received modified opinions, compared to 42% of the no irregularity sample cases involving auditor litigation (chi-square = 4.28; p= .038). Panel B of figure 1 presents a breakdown of the no irregularities subsample of 500 observations between companies reporting profits prior to bankruptcy and those reporting losses. For the 52 cases involving auditor litigation, 42%

reported profits just before bankruptcy, compared to 19% of the 448 no litigation cases (chi-square = 15.48; p = .0001). These results indicate a statistically significant dependence between prebankruptcy reporting of profits and auditor litigation.

Panel B of figure 1 also shows that opinions were modified for only 2 of the 22 auditor litigation cases reporting prebankruptcy profit, compared to 21 of the 84 profit cases from the no litigation subsample.

Opinion modifications were issued for roughly two-thirds of both the no litigation and auditor litigation loss subsamples. Opinion modification was not incrementally significant in a multivariate analysis of the incidence of litigation after controlling for the existence of reported

profit. The observed t-statistic for the profit dummy variable was 3.157.

The opinion modification dummy variable took on a negative sign with an associated t-statistic of -.743. The negative sign resulted from the higher modified reporting rate for the no litigation-profit subsample,

compared to the auditor litigation-profit subsample.5 Finally, I ran a

chi-square test of the independence between incidence of litigation and opinion modification for the subsample of companies with no irregularities reporting prebankruptcy profits. Although the test results are suspect because one cell is sparse (only two modifications in the auditor litigation subsample), they suggest the possibility of statistical dependence (chi-square = 2.60; p = .11).

The results reported in this section indicate that opinion modifications occur equally frequently in cases with and without auditor litigation, if the client reported a prebankruptcy loss. If the client reported a prebankruptcy profit, the proportion of opinion modifications was higher for the no litigation subsample compared to the auditor litigation subsample, indicating that opinion modification may protect auditors from litigation when the client reports a profit and later goes bankrupt.

4. Other Empirical Issues

In this section I discuss two additional empirical issues-the combi- nation of two different financial distress indexes into one control variable and the use of resolution data to investigate degree of protection.

5A negative sign was predicted by the authors since the protection hypothesis implies an inverse relation between opinion modifications and incidence of litigation.

(8)

Distributions of distress model scores are highly sensitive to changes in sample composition, choice of estimation technique, and the choice of independent variables. For example, a recent internal research project at KPMG involved the estimation of two different failure prediction models for commercial banks. One model was estimated using unweighted sample data on 145 failed banks and 991 nonfailed banks.

The second model was estimated using the same sample and same independent variables, and the WESML technique to weight the likeli- hood function for approximate population proportions-150 failed banks and 11,000 nonfailed banks.

The unweighted model score cutoff associated with a 91 % hit rate on the failed banks and a 97.5% hit rate on the nonfailed banks was .30. Us- ing the weighted model, roughly analogous hit rates were achieved at a cutoff of .02. At a cutoff of .02, the unweighted model hit rate on the nonfailed sample dropped to 87% and the hit rate on the failed sample rose to 98%. P&C used the unweighted model from Zmijewski's [1984]

paper and combined its scores with those from the Sinkey, Terza, and Dince [1987] model for financial institutions.6 Based on the commercial bank illustration presented above, it is questionable whether the com- bination of distress scores from these two models yields a meaningful control variable. If the two distributions are significantly different, one model's high-risk scores could more closely resemble the second model's low- or moderate-risk scores. P&C could compare the two distributions to see if medians, minimums, and maximums are roughly equivalent.

P&C investigate auditor resolutions to learn more about whether opinion modifications protect auditors from the costs of litigation. Using 49 cases for which resolution data are available, the authors observe that 16 cases were dismissed with no payments made by the defendants. How- ever, for 9 of these 16 cases, no opinion modifications were rendered.

The lowest auditor payments are associated with litigation in which all audit reports at issue have been modified, and the highest auditor payments are for observations where none of the reports was modified.

However, it is not apparent that a $1 million resolution represents greater auditor protection than a $100 million resolution. Suppose the original plaintiff 's claim for the $1 million outcome was $1.2 million and

the original claim for the $100 million outcome was $200 million; for

which case was the degree of protection greater?

5. Participants' Comments

Participants at the conference offered the following additional comments about the paper. Questions were asked about the validity of an SEC Enforcement Action as a surrogate for the existence of a material 6 I do not know whether Sinkey, Terza, and Dince [1987] used a weighting technique

such as WESAIL or whether their sample proportions were representative of the population.

(9)

38 TIMOTHY B. BELL

irregularity. One participant commented that about 25 % of the SEC En- forcement Actions are administrative actions that do not indicate the existence of a material irregularity. Participants also questioned why the sample was limited to bankrupt public companies audited by the Big Six (Big Eight). One participant commented that there is a significantly higher proportion of qualified audit reports for non-Big Six audits compared to Big Six audits. Another participant commented that smaller firms who are likely audited by non-Big Six firms may exhibit more aggressive reporting practices. Exclusion of private companies (e.g., Phar-Mor) and companies audited by non-Big Six firms (e.g., LF) may limit the generalizability of the results. Finally, some participants argued that distressed nonbankrupt companies should have been in- cluded in the sample, since it is typically distress, and not bankruptcy, that leads to shareholder/creditor losses and related litigation.

REFERENCES

DELANEY, K. J. Strategic Bankruptcy: How Corporations Use Chapter 11 to Their Advantage.

Berkeley: University of California Press, 1992.

MARSH, G. A., AND D. C. CHENG. "The Impact of the Bankruptcy Reform Act on Business Bankruptcy Filings." Alabama Law Review (Winter 1985): 515- 47.

SINKEY, J.; J. TERZA; AND R. DINCE. "A Zeta Analysis of Failed Commercial Banks." Quarterly Journal of Business and Economics (Autumn 1987): 35-49.

ZMIJEWSKI, M. E. "Methodological Issues Related to the Estimation of Financial Distress Prediction Models."Journal of Accounting Research (Supplement 1984): 59-82.