Auditor resignations: clientele effects and legal liability

https://doi.org/10.1016/S0165-4101(00)00019-7Get rights and content

Abstract

I examine two hypotheses of auditor resignation: litigation risk and clientele adjustment. I find resignation is positively related to increased client legal exposure, and to occurrence of clientele mismatch. The summary-measure approach allows me to distinguish clientele mismatch caused by changes in auditor (supply-side) characteristics vs. changes in client (demand-side) characteristics. Evidence suggests resignation is likely driven by supply-side changes. I also find investors react negatively to resignations, and the price drop varies cross-sectionally with litigation risk. Further, the tendency of dropped firms to engage small auditors is positively related to increased litigation risk, and to mismatch with large auditors.

Introduction

This paper examines potential reasons for auditor resignation and the associated effects on client firms. There are many papers that view auditor changes as client-initiated realignments driven by changes in client demand (e.g. Johnson and Lys, 1990). However, we know relatively little about auditors’ incentives to terminate auditor–client relationships. This paper attempts to fill this gap by providing a comprehensive and rigorous analysis of the economic factors affecting auditor resignation.

Auditor resignations have attracted widespread media interest and have been highlighted by accounting firms. One of the big-six auditors, KPMG Peat Marwick, reports it is dropping 50–100 audit clients annually, as compared with 0–20 ten years ago. Other big-six auditors provide similar statistics (Berton, 1995).1 Why have auditor resignations become more prevalent in recent periods?2 The rise of auditor resignations coincides with several recent trends that potentially affect auditors’ cost functions and opportunity sets: (1) increases in auditor legal liabilities, (2) changes in audit technology, and (3) expansion of non-audit services offered by audit firms. Although the media has focused exclusively on the first, the latter trends could also contribute to auditor resignation. I explore two distinct, although non-mutually exclusive, explanations for auditor resignation based on these trends.

The first explanation posits that auditors drop certain clients to reduce their legal exposure. I refer to this explanation as the litigation-risk hypothesis. Anecdotal evidence suggests that increased litigation cost is the primary factor driving the trend of auditor resignations (Berton, 1995; MacDonald, 1997a). Krishnan and Krishnan (1997) find that auditor resignations and dismissals differ in dimensions that could reflect litigation risk. However, whether litigation risk is important in the presence of other plausible hypotheses is an open question. In addition, several related questions need to be addressed, such as why other auditors willingly pick up dropped clients, and whether these auditors have a comparative advantage in auditing risky clients. I examine the litigation-risk hypothesis in more depth based on various institutional facts, and attempt to address the incremental effect of litigation risk in the context of a plausible alternative.

The second explanation states that recent changes in the audit industry (other than changes in the legal environment) have altered the relative benefit and cost of each client, motivating auditors to change their client mix. This is broadly referred to as the clientele-adjustment hypothesis. First, auditors’ production functions have changed as a result of technological advances. Second, auditors’ opportunity sets have expanded as a result of the growing demand for non-audit work. Assuming each auditing firm specializes in an optimal portfolio of clients, changes in auditors’ production and opportunity sets could alter the composition of the set of clients. In the process of adjusting their client mix, auditors will shift away from poorly matched clients, leading to auditor resignation. Note the clientele mismatch here is distinct from the mismatch described in Johnson and Lys. While mismatch in their work is motivated by client-specific changes in audit demand (demand-side effects), the clientele mismatch here focuses on changes in auditors’ cost structures and opportunity sets (supply-side effects). The research design used to isolate the supply-side effects and identify the source of clientele mismatch represents a major methodological innovation of this paper.

The primary purpose of this study is to understand the extent to which the litigation-risk and the clientele-adjustment hypotheses describe auditor resignations. To do this, I examine the economic circumstances surrounding auditor resignations, including changes in auditor or client characteristics prior to auditor resignation, the associated stock price reactions, and the characteristics of successor auditors.

I find evidence that both litigation risk and clientele adjustment contribute to auditor resignation. The second finding is particularly interesting since litigation risk is the only explanation that has received attention from the popular press. Further, I use a methodology that enables me to distinguish clientele mismatch caused by changes in auditor characteristics from mismatch caused by client-specific changes. The evidence identifies the former as the more likely cause of auditor resignation.

I document a significant and negative stock reaction to auditor resignations. Since auditors have private information about their clients, auditor resignations potentially signal the clients’ high litigation risk or lack of growth, causing their stock prices to drop. The negative stock reaction to auditor resignations is consistent with DeFond et al. (1997) and Wells and Loudder (1997), and contrasts with earlier studies that fail to detect significant stock reactions to general auditor changes (e.g. Johnson and Lys). Further, I find the stock-market reaction varies in cross-section with a client's change in litigation risk: the greater the increase in litigation risk, the larger the drop in stock price, provided the resignation occurs during an annual audit. This finding is new to the literature.

Finally, I examine the characteristics of successor auditors of the dropped firms. Compared to firms with client-initiated auditor changes (henceforth, auditor changes), dropped firms tend to switch to small auditors. Further, the tendency of a dropped firm to engage a small auditor varies cross-sectionally with litigation risk: the greater the increase in litigation risk, the more likely is a client to engage a small auditor. This supports the litigation-risk hypothesis if small auditors’ lack of ‘deep pockets’ and reputation capital reduces their litigation costs. The tendency to align with a small successor auditor is also consistent with the clientele-adjustment hypothesis: a client that becomes mismatched with a large auditor is more likely to engage a small successor auditor.

This paper contributes to the literature in several ways. First, it addresses auditor resignations from a broader perspective than previous papers, and attempts to discern the relative importance of two plausible explanations of auditor resignations. Second, the research design represents a significant methodological improvement over earlier attempts. Importantly, the methodology enables me to distinguish clientele mismatch caused by supply-side effects from those by demand-side effects. By identifying the latter as the more likely cause of auditor resignations, this study provides new insights into the impact and source of clientele mismatch.3

Section 2 develops the main hypotheses. Section 3 outlines the sample selection procedure. Section 4 describes the methodology to construct summary measures for the two hypotheses. Section 5 discusses changes in auditor or client characteristics prior to auditor resignation. Section 6 presents stock market reactions to auditor resignation. Section 7 examines the characteristics of successor auditors. Sensitivity analysis is in Section 8. Section 9 concludes.

Section snippets

Auditor resignation: potential explanations

This section discusses two explanations of auditor resignations and formalizes the testable implications.

Auditor-resignation sample

Public firms are required to file Form 8-K with the SEC in the case of an auditor change. Financial Reporting Release (FRR) No. 31 of 1988 specifically requires firms to report any auditor resignations. I use keyword searches in LEXIS/NEXIS for a comprehensive list of phrases that may indicate auditor resignations. I identify 629 auditor resignations from 1987 to 1996, including six added from Dow Jones News Retrieval.

Of the 629 firms, 240 are not listed in the 1997 Compustat Annual Industrial

Proxies for the two hypotheses

The difficulty in testing litigation-risk and clientele-adjustment hypotheses is that litigation risk and clientele mismatch are not directly observable. In this section I describe the procedures to construct summary (index) measures for litigation risk and clientele mismatch.

Auditor resignation and changes in auditor or client characteristics

After constructing the proxies, I run logit models to test the two hypotheses of auditor resignations. I use two different control samples, the random sample and the auditor-change sample. The dependent variable is one for an auditor-resignation client, and zero otherwise. I include several variables in the regression to control for other factors that might affect auditor resignations. Auditor qualification might directly increase the likelihood of auditor resignation (over and above its effect

Auditor resignation and market reaction

I examine next client stock returns and investigate the information content of auditor resignations. As discussed in Section 2, a negative market reaction is predicted under the litigation-risk hypothesis. It is also possible under the clientele-adjustment hypothesis if the sample is dominated by cases where auditor resignation conveys lack of client growth.

To compute daily abnormal return, I follow Brown and Warner (1985), and assume the one-factor market model as the return-generating

Characteristics of successor auditors

In this section I study the successor auditors of the dropped firms to further differentiate the two potential explanations. As argued in Section 2, it is less costly for small auditors to bear litigation risks. Hence, the litigation-risk hypothesis predicts that dropped firms will engage small auditors, and the tendency to do so is positively related to the increase in their risks. The clientele-adjustment hypothesis also has predictions for the successor auditors: clients that become

Sensitivity analysis

The inferences so far depend on how well the proxies capture the ‘true’ variables. I perform various sensitivity checks to mitigate concerns for measurement error. In the main tests, an ex ante measure, the probability of auditor litigation, is used to proxy for litigation risk. Because this variable is continuous, I can measure the change in legal exposure from one year to another, and therefore perform my tests using the ‘change’ specification. However, this measure is estimated from other

Conclusions

This paper investigates two potential explanations of auditor resignations: litigation risk and clientele adjustment. Using a summary measure of ex ante litigation risk, I find a positive and significant relation between the increase in a client's auditor litigation risk and the likelihood it will be dropped by its auditor. This holds after controlling for alternative explanations.

Another apparent determinant of auditor resignation is the occurrence of clientele mismatch as a result of changes

References (32)

  • L DeAngelo

    Auditor independence, “Low Balling”, and disclosure regulation

    Journal of Accounting and Economics

    (1981)
  • L DeAngelo

    Auditor size and auditor quality

    Journal of Accounting and Economics

    (1981)
  • K Palepu

    Predicting takeover targetsa methodological and empirical analysis

    Journal of Accounting and Economics

    (1986)
  • Alexander, J., 1991. Do the merits matter? A study of settlements in securities class actions. Stanford Law Review...
  • Berton, L., 1989. SEC is studying independence of firms doing consulting and accounting work. The Wall Street...
  • Berton, L., 1995. Big accounting firms weed out risky clients. The Wall Street...
  • Brown, S., Warner, J., 1985. Using daily stock returns. Journal of Financial Economics...
  • Bockus, K., Gigler, F., 1998. A theory of auditor resignation. Journal of Accounting Research...
  • DeBerg, C., Kaplan, S., Pany, K., 1991. An examination of some relationships between non-audit services and auditor...
  • DeFond, M., Ettredge, M., Smith, D., 1997. An investigation of auditor resignations. Research in Accounting Regulation...
  • Dunbar, F., Juneja, V.M., Martin, D., 1995. Shareholder litigation: theory and evidence on deterrent value and merits....
  • M Firth

    The provision of nonaudit services by accounting firms to their audit clients

    Contemporary Accounting Research

    (1997)
  • Francis, J., Philbrick, D., Schipper, K., 1993. Determinants and outcomes in class action securities litigation....
  • Johnson, B., Lys, T., 1990. The market for audit services: evidence from voluntary auditor changes. Journal of...
  • Jones, C., Weingram, S., 1996. Why 10b-5 risk is higher for technology and financial service firms. Working paper,...
  • Kelly, J., 1994. Accountancy column: opening the books and limiting liabilities — Jim Kelly examines the ramifications...
  • Cited by (0)

    This paper is based on my dissertation completed at the University of Rochester. I gratefully acknowledge the guidance from my dissertation committee, Ray Ball, S.P. Kothari, Jerry Zimmerman, and especially Ross Watts (chairman). Helpful comments and suggestions were also received from Wayne Guay, Christoph Hinkelmann, Bob Holthausen (the editor), W. Bruce Johnson (the referee), Ken Kotz, Andy Leone, Jon Lewellen, James Livingston, Michelle Lowry, Ken Schwartz, Cliff Smith, Billy Soo, Denis Suvorov, and Narinder Walia. I also benefited from seminar participants at Boston College, Emory University, University of Minnesota, Northwestern University, University of Notre Dame, Penn State University, University of Pennsylvania (Wharton), University of Rochester, University of Washington at Seattle, Washington University at St. Louis, and the 1997 American Accounting Association Meeting. I thank the following individuals for providing many useful institutional details: Ray D’Agostino, Robert Herz from Coopers & Lybrand, Stanley Konopko from Arthur Anderson, Edward Short, and Ingrid Stanlis. Finally, I acknowledge financial support from the Olin Foundation and Deloitte & Touche Foundation. Any errors are my own.

    View full text