Investigating fraud losses, can companies reduce losses associated with fraud by voluntary disclosing

... Major Professor: Dr John Rigsby Title of Study: INVESTIGATING FRAUD LOSSES: CAN COMPANIES REDUCE LOSSES ASSOCIATED WITH FRAUD BY VOLUNTARY DISCLOSING? Pages in Study: 101 Candidate for Degree o f... to reduce future litigation losses, then an inference can be made that market participants might perceive voluntary disclosure of fraud as a strategic maneuver to mitigate losses associated with. .. smaller companies, with the dollar amount o f loss associated with fraud increasing exponentially with company size In addition, they found a positive relationship between fraud duration and fraud

INVESTIGATING FRAUD LOSSES: CAN COMPANIES REDUCE LOSSES ASSOCIATED WITH FRAUD BY VOLUNTARY DISCLOSING? By Thomas John Olach A Dissertation Submitted to the Faculty of Mississippi State University In Partial Fulfillment o f the Requirements for the Degree o f Doctor o f Philosophy in Accountancy in the School o f Accountancy Mississippi State, Mississippi December 2005 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. UMI Number: 3201854 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3201854 Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission. INVESTIGATING FR AU D L O SS E S : CAN C O M PA N IES R E D U C E L O S S E S A SSO C IA TE D WITH FR AU D BY VOLUNTARY D ISC L O SIN G ? By Thomas J. Olach Approved: JohnfT. Rigsby f // J Associate Professor bf Accountancy 'irector of Dissertation Professor of Quantitative Analysis (Committee Member) Mark Lehman Associate Professor of Accountancy (Committee Member) L. Dwight Sneathen, Jr. Assistant Professor of Accountancy (Committee Member) 7 Ronald D. Taylor Professor of Marketing (Committee Member) R a rh a rfl A. A Spencer fin o n re r » Barbara v Director of Graduate Studies in Business The College of Business and Industry Sara Freedman Dean of the College of Business and Industry R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. Name: Thomas John Olach Date of Degree: December 10, 2005 Institution: Mississippi State University Major Field: Accountancy Major Professor: Dr. John Rigsby Title of Study: INVESTIGATING FRAUD LOSSES: CAN COMPANIES REDUCE LOSSES ASSOCIATED WITH FRAUD BY VOLUNTARY DISCLOSING? Pages in Study: 101 Candidate for Degree o f Doctor of Philosophy As the enforcement division of the Securities and Exchange Commission (SEC) adjusts its strategy by looking into ways to motivate publicly-held companies to report fraud immediately upon detection, there is currently a demand to prove that incentives exist for managers of companies to voluntarily disclose fraud. Identifying such motives could greatly expand both the capital markets literature stream as well as the voluntary disclosure literature stream. The voluntary disclosure literature stream has shown that management has certain incentives to make good versus bad disclosures. These incentives normally include opportunism or political motives. There has been little research in the voluntary disclosure literature that investigates whether or not managers choose to disclose bad news, such as fraud, because they want to appear forthcoming to the market. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. The focus of this study is to determine if companies experience lower losses if they voluntarily disclose fraud rather than waiting until the media or the SEC releases the news. The losses investigated are the (1) decline in company market value, as measured in abnormal returns, (2) the length o f the companies’ and fraud perpetrator(s)’ SEC investigation, and (3) civil penalties and disgorgement imposed upon fraud perpetrators. The results of this study indicate that voluntary disclosing companies do not mitigate market value decline or costs and resources associated with a SEC investigation. There was a similar decline in market value whether the fraud is voluntarily disclosed or not and there was no significant difference in the duration o f either the companies’ or fraud perpetrators’ investigation periods. The findings indicate, that for the period investigated, neither the market nor the SEC distinguishes between voluntary disclosing and involuntary disclosing companies. Also, the finding related to civil penalty and disgorgement amounts indicate that a company’s choice to voluntarily disclose fraud does not significantly impact civil penalty and disgorgement amounts imposed on fraud perpetrators. Key words: fraud, voluntary disclosure, involuntary disclosure. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. ACKNOWLEDGEMENTS I would like to thank all of my committee members for taking the time to help me learn from this dissertation. I want to thank Dr. Rigsby, my chairperson, for motivating me to stay on top of my work and regularly meeting with me to make sure I was always on the right track. I also owe thanks to Dr. Lehman for his knowledge o f fraud and for his help with structuring my paper. I would like to thank Dr. Sneathen for always being available to answer my questions. Also, thanks to Dr. Taylor and Dr. Tahai, my two statistical committee members, for encouraging and teaching me to work independently while testing my data. It was in Dr. Tahia’s time series class where I developed the idea to perform this research. I would also like to thank the accounting and business law faculty at Minnesota State University-Mankato for encouraging me to complete my dissertation in a timely manner and for providing the necessary funding to do so. I will do my best to make sure your generosity pays off in the years to come. ii R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. TABLE OF CONTENTS Page ACKNOWLEDGEMENTS............................................................................................. ii LIST OF TABLES............................................................................................................ v LIST OF FIGU RES.......................................................................................................... vii CHAPTER I. INTRODUCTION............................................................................................... 1 Background.......................................................................................................... Voluntary Disclosure o f Fraud............................................................................ Scope and Objectives of the Study.................................................................... Usefulness of the Study...................................................................................... 1 2 3 5 II. LITERATURE REVIEW ................................................................................... 8 Capital Markets Literature.................................................................................. Confounding Factors....................................................................................... Lawsuit Filings or Threats of Future Litigation........................................... Auditor Resignations and Terminations....................................................... Accounting Restatements................................................................................ Qualified or Withdrawn Audit O pinions...................................................... Bankruptcy Filings.......................................................................................... Changes in Senior M anagement.................................................................... Audit Violations.............................................................................................. Trading Halts, Suspensions, and Delistments.............................................. Terminology Used to Describe Fraud........................................................... Summary.......................................................................................................... Voluntary Disclosure Literature........................................................................ Disclosure Q uality.......................................................................................... Corporate Securities Regulation Literature...................................................... The SEC Investigation Process...................................................................... Other Relevant Fraud Research......................................................................... 8 11 12 15 16 17 18 18 19 20 22 23 24 28 29 30 36 iii R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. CHAPTER III. Page HYPOTHESES AND RESEARCH DESIGN................................................. 38 Investigating the Market’s Reaction to Fraud................................................... Investigating the Duration o f the Media Exposure Period and Lingering Litigation Period.............................................................................................. Investigating Civil Penalty and Disgorgement Amounts................................. Research Design................................................................................................... Testing the Company Decline in Market V alu e.......................................... Testing for Collinearity................................................................................... Prediction of V ariables................................................................................... Investigating the Duration o f Investigation Periods........................................ Prediction of V ariables................... Investigating Civil Penalty and Disgorgement Amounts................................. Sample and Data Collection................................................................................ 38 41 42 43 44 47 48 51 53 56 60 RESULTS............................................................................................................ 66 Introduction........................................................................................................... Company Descriptive Inform ation.................................................................... Results for the Market’s Reaction Testing........................................................ Results for the Duration of the Companies’ Investigation Period................. Results for the Duration of the Fraud Perpetrators’ Investigation P erio d Results for the Civil Penalty and Disgorgement A m ounts............................. 66 66 69 74 78 81 CONCLUSIONS................................................................................................. 86 Summary and Implications................................................................................. Limitations............................................................................................................ Future Research................................................................................................... 86 91 91 REFERENCES................................................................................................................. 93 IV. V. APPENDIX A. Supplementary Testing o f Market’s Reaction T e st........................................ 96 B. Classification of Companies Included in Final Sample.................................. 99 iv R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. LIST OF TABLES TABLE Page 2.1 Sources o f First Announcements.......................................................................... 9 2.2 The Frequency o f SEC Suspensions Since 1995................................................. 21 2.3 Civil Penalties for Civil A ctions........................................................................... 35 3.1 Summary o f Predictions and Variable Type for H 1 .......................................... 51 3.2 Summary of Predictions and Variable Type for H2 & H 3 ................................ 55 3.3 Summary o f Predictions and Variable Type for H4 & H 5 ................................ 60 3.4 Derivation of Final Sample.................................................................................... 62 3.5 Source o f Initial Fraud Disclosure........................................................................ 63 3.6 Companies With or Without Media Exposure Periods and/or Lingering Litigation Periods........................................................................... 66 4.1 Descriptive Characteristics of Sampled Companies........................................... 68 4.2 Average Abnormal Returns Around the Initial Disclosure of Fraud................ 70 4.3 Cross-Sectional Association Between Cumulative Abnormal Returns and Independent Variables.............................................................................. 73 Cross-Sectional Association Between the Duration of the Media Exposure Period and Fraud Disclosure C hoice........................................... 77 Cross-Sectional Association Between the Duration o f the Lingering Litigation Period and Fraud Disclosure Choice........................................... 80 Cross-Sectional Association Between Civil Penalties Imposed Upon Fraud Perpetrators and Fraud Disclosure C hoice........................................ 83 4.4 4.5 4.6 v R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. TABLE 4.7 A .l Page Cross-Sectional Association Between Disgorgement Imposed Upon Fraud Perpetrators and Fraud Disclosure C hoice 85 . Univariate Regression Analysis Between Cumulative Abnormal Returns and Independent Variables.............................................................................. 98 vi R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. LIST OF FIGURES FIGURE 1 Page The Corporate Fraud Process................................................................................... vii R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 5 CHAPTER 1 INTRODUCTION According to the Association o f Certified Fraud Examiner (ACFE) Report to the Nation (2004), occupational fraud and abuse costs U.S. companies more than $660 billion a year.1 When corporate fraud is committed, significant losses are bound to be incurred. These losses are normally amplified the longer the losses are concealed. The adverse market effects associated with corporate fraud and concealment include reputational damages, civil penalties, information asymmetry, increased agency costs, costs o f investigation(s), and lost profits. Background There has been recent legislative action to help reduce corporate fraud and concealment. The Sarbanes-Oxley Act o f 2002 is designed to help mitigate corporate fraud by establishing the Public Company Oversight Board (PCAOB) to help monitor management activities, investigate fraud, impose sanctions upon fraud perpetrators, and make Chief Financial Officers (CFOs) and Chief Executive Officers (CEOs) more costs from companies and their auditors, it will be several years before it can be determined if the Act has helped reduce corporate fraud. Therefore, offering a low-cost remedy for the problem o f corporate fraud and concealment where the results can be 1 The ACFE defines occupational fraud as “the use o f one’s occupation for personal enrichment through the deliberate misuse or misapplication o f the employing organization’s resources or assets.” 1 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 2 immediately observed would be the ideal way to mitigate corporate fraud and concealment. If it can be determined that companies that voluntarily disclose fraud experience less of a market value decline than involuntary disclosers, an incentive may be created to motivate managers to disclose fraud immediately upon detection. Unfortunately, while the SEC has been informing publicly-held companies during the past several years that they can avoid civil penalties and charges by being forthcoming with fraud and cooperating with the SEC during its investigations, only three companies are known to have received benefit after voluntarily disclosing fraud. These three companies are Seaboard Corp., Homestore, Inc., and Electro Scientific Industries (ESI). Rather than positive, the SEC appears most likely to use negative reinforcement. Reason (2005) suggests that civil penalties and charges are most likely to be imposed by the SEC for frauds with long duration periods and those perpetrated at the highest levels. Voluntary Disclosure of Fraud Fraud is defined by Albrecht and Albrecht (2004) as: ...a generic term that embraces all the multifarious means which human ingenuity can devise, which are resorted to by one individual, to get an advantage over another by false representations. No definite and invariable rule can be laid down as a general proposition in defining fraud, as it includes surprise, trickery, cunning and unfair ways by which another is cheated. The only boundaries defining it are those which limit human knavery, (p. 5) Fraud exists when a material misrepresentation is made, intentionally or recklessly, which is believed and acted upon by the victim(s) and causes them damage. Voluntary disclosure o f fraud by management is defined, for purposes o f this study, as R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 3 management’s choice o f disclosing the company’s fraud to the public before the Securities and Exchange Commission (SEC) informs the target company it is opening a formal investigation pertaining to the company’s alleged fraud or an announcement is made by someone external to the company.2 While the SEC has no specific regulation regarding the disclosure o f fraud, the Securities Exchange Act o f 1934 requires companies to disclose fraud if it materially impacts the financial statements. However, because o f the subjective nature o f fraud disclosures, disclosures are frequently not made when they should be. The SEC does have Regulation FD, which requires companies to disclose material information to all market participants in a timely manner to mitigate insider trading opportunities. Regulation FD is intended to protect small investors. On June 17, 2002, the SEC proposed amendments to Regulation FD requiring that companies file a Form 8-K to disclose the occurrence o f 11 new extraordinary corporate events, including material impairments, write-offs, and restructuring charges. The amendments would also accelerate the filing deadline for Form 8-K to require filing within two business days after the occurrence o f an event requiring disclosure. Scope and Objectives of the Study When fraud is disclosed, the most significant loss experienced by a company is normally the decline in the company’s market value, which is the primary focus of the study. The market’s reaction is investigated during an 11-day period around the fraud’s 2 The SEC investigates not only fraud, but also nonfraudulent unacceptable accounting practices. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 4 initial disclosure. More specifically, the period examined will be {-5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5}, with zero being the initial disclosure date. In addition to examining market value, two tests are performed to determine the relationship between voluntary fraud disclosures and the duration o f the SEC’s investigation period. The first test examines whether a company can reduce the media exposure period by voluntarily disclosing fraud. For purposes o f this study, the media exposure period is defined as the period beginning the date the fraud is initially disclosed and ends the date the company settles its case with the SEC. During this period o f time, most o f the SEC’s investigation is normally focused upon the company. The second test examines whether the lingering litigation period is shorter for voluntary disclosing companies. The lingering litigation period is defined as the period beginning the date the company settles its case with the SEC and ends the date the last fraud perpetrator settles their case with the SEC. During this period o f time most o f the SEC’s investigation is normally focused upon the fraud perpetrators. Figure 1 depicts the typical timeline o f the entire corporate fraud process. Separating the companies’ investigation period (media exposure period) from the fraud perpetrators’ investigation period (lingering litigation period) may help explain whether there is a difference in the degree o f association between the duration of a specific investigation period and the companies’ forthcomingness with their fraud. While the investigation related to the company may be shorter for companies that voluntarily disclosed fraud, the investigation related to fraud perpetrators should not be affected, R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 5 since fraud perpetrators should not benefit from management’s and the board of directors’ actions. Lingering litigation period (LLP) Media exposure period (MEP) 1 1 I------------------------1------------------------1 Initial fraud disclosure SEC informal investigation SEC formal investigation I Company settles with SEC J I Final settlement between SEC and perpetrators Figure 1: The Corporate Fraud Process In addition, this study tests the association between companies’ forthcomingness with fraud and the amount of civil penalties and disgorgement imposed upon fraud perpetrators by the SEC. If the SEC considers whether management and the board of directors voluntarily disclosed their fraud as a factor of how much to seek in civil penalties and disgorgement, the SEC may inappropriately impose less civil penalties and disgorgement amounts on perpetrators simply because o f management’s and the board of directors’ actions. Usefulness of the Study The findings from this study may benefit several parties. If it can be shown that the decline in company market value is less for companies that voluntarily disclose fraud, companies will have an incentive to immediately disclose fraud. Consequently, shareholders will benefit because losses associated with their stock holdings will likely be reduced due to management choosing to disclose fraud immediately upon detection. In addition, if companies are given an incentive to voluntarily disclose fraud, a broad R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 6 spectrum of market participants will benefit because public information is released symmetrically due to immediate disclosure o f fraud. Receiving more timely information about their companies’ fraud will better market participants’ ability to make decisions. Furthermore, a reduction in the disincentive to disclose fraud likely means that managers will be less reluctant to detect fraud. This study also has public policy implications since evidence is gathered to evaluate whether the SEC allows fraud perpetrators to benefit by seeking less in monetary penalties and disgorgement when the choice to disclose the fraud voluntarily was likely not made by the perpetrators. Consequently, the SEC may choose to modify current sentencing guidelines when calculating civil penalty and disgorgement amounts to better punish fraud perpetrators and to ensure injured shareholders are adequately compensated for their losses. While the timeline depicted in Figure 1 is the most frequent scenario, there are several situations other than the typical fraud situation discussed above that need to be considered. First, the fraud perpetrators may settle with the SEC before the company does. This situation is rare because companies normally like to accelerate their settlement with the SEC to mitigate reputational damages and they normally have sufficient resources to do so, unlike the fraud perpetrators. If the fraud perpetrators settle before the company does, this situation gives rise to a media exposure period, but no lingering litigation period. This is because the SEC’s investigation is focused upon the company throughout the entire period. Furthermore, there will be no unsettled cases lingering after the company settles its case with the SEC. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. Another situation occurs when the SEC takes action only against the fraud perpetrators and not the company. As a result, there is no media exposure period. The lingering litigation period begins the date the fraud is initially disclosed and ends the date o f the last fraud perpetrator settlement. No media exposure exists because the primary focus o f the SEC’s investigation is only upon the fraud perpetrators, and not the company. A third possible situation occurs when the SEC takes action only against the company, and not fraud perpetrators. This scenario is rare because companies cannot perpetrate their own fraud, and normally after the SEC conducts its investigation at least one fraud perpetrator is identified. This situation would give rise to a media exposure period, but no lingering litigation period. Finally, another situation may arise when both the company and the fraud perpetrators settle their case on the same date, which occurs fairly frequently. For purposes o f the study, this situation gives rise to a media exposure period, but no lingering litigation period. The media exposure period will still begin the date the fraud is initially disclosed and end the date the company settles its case with the SEC. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. CHAPTER II LITERATURE REVIEW Four literature streams are examined in this chapter and how they relate to the study. They are: (1) capital markets, (2) voluntary disclosure, (3) corporate securities regulation, and (4) other fraud literature. Capital Markets Literature An important purpose of the study is to investigate the market’s reaction to fraud to determine if the decline in market value a company incurs can be reduced by voluntarily disclosing its fraud. It is challenging to measure the market’s reaction to fraud because o f the many other adverse events that frequently accompany fraud disclosure. If not properly controlled for, confounding can be a prevalent problem when investigating the market’s reaction to fraud. This is because a ‘domino effect’ typically occurs where the initial announcement of fraud leads to a stream o f other adverse events. Dechow et al. (1996) provides evidence o f a number o f potential adverse events that can be triggered by a fraud disclosure when they examined abnormal returns o f 92 companies subject to enforcement actions for earnings management practices. They found that when other adverse announcements accompanied an initial earnings management announcement, both the news of the earnings management practices and the other announcement(s) frequently resulted in significant market reactions. Table 2.1 lists 8 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 9 sources of first announcements for earnings management practices used by Dechow et al. (1996), along with the market’s reaction to each announcement as measured by abnormal returns. The frequency of each announcement source is also given. Several o f these items are incorporated in this study and will be discussed later. TABLE 2.1 SOURCES OF FIRST ANNOUNCEMENTS Abnormal return upon initial disclosure -.057 Number of companies 26 SEC investigation announcement -.043 25 Auditor raises concerns or is fired -.250 13 Class-action lawsuits -.031 8 Press statement criticizing company’s accounting practices -.084 7 Unable to meet earnings expectations announcement -.030 5 Termination of executives -.029 2 Suspension of trading1 -.601 2 Date of first AAER .000 4 TOTAL -.088 92 Source of first announcement Accounting restatement 1 Among these sources o f initial disclosure, the most adverse market reaction was when earnings management practices were reported simultaneously with a trading suspension announcement. This result, however, should be viewed with skepticism. Subsequently, Shumway (1997) investigated the market’s reaction by using delisted returns, and he found delisted returns contained bias when returns data is obtained from Center for Research in Security Prices (CRSP). R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 10 Among the sources listed in Table 2.1, the item that had the most adverse market reaction is the trading suspension announcement. The item most frequently simultaneously disclosed with earnings management practices are accounting restatements. As with Dechow et al. (1996), most existing capital markets literature that measured the market’s reaction to fraud and other adverse company announcements did so by examining abnormal returns. Abnormal returns are defined as stock returns in excess of market-adjusted stock returns. Cumulative abnormal returns are also used when investigating the market’s reaction to fraud and other adverse announcements. Feroz et al. (1991) tested the market’s reaction to enforcement actions at the initial disclosure date, the SEC investigation announcement date, and the settlement date using a sample of 224 companies subject to enforcement actions for unacceptable accounting practices during the period between 1982 and 1989. The study reported two day abnormal returns of -12.9 percent upon the initial disclosure o f an unacceptable accounting practice. The study also found that abnormal returns, upon an SEC investigation announcement, were -6.0 percent. There was no effect on the companies’ market value on the settlement date. Dechow et al. (1996) found similar results, i.e. abnormal returns o f -9.0 percent when earnings management practices were initially disclosed. The fact that a company’s market value significantly declines when a SEC investigation is announced indicates the SEC serves as a viable sanctioning agent because managers have an incentive to avoid further losses associated with investigations. This R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 11 study investigates whether or not managers can reduce some o f their company’s market value decline by being forthcoming with their fraud. Confounding Factors Defond and Smith (1991) suggested that the results from the Feroz et al. (1991) market’s reaction test related to the dates that the unacceptable accounting practice was initially disclosed and the date the SEC investigation was announced may be distorted due to confounding. This distortion is attributable to market participants anticipating from other events either the unacceptable accounting practice announcement or the SEC investigation. For example, if a company announces a material restatement, market participants may expect this news to trigger a SEC investigation, if one is not already being performed. Therefore, when an unacceptable accounting practice announcement or a SEC investigation announcement is made, market participants may not be obtaining new news, but rather a confirmation of an existing expectation. If market participants can effectively predict a SEC investigation, they will likely expect other consequences will occur associated with the SEC investigations. Some o f these adverse consequences may include increased litigation risk, investigations by other regulatory agencies, the detection o f more fraud, and disassociations o f suppliers and customers. Because the timing of the market’s reaction is uncertain regarding announcements disclosing unacceptable accounting practices, this study will control for several possible adverse events that frequently accompany fraud disclosure. While prior research dealing with the market’s reaction to fraud is limited, there has been a significant amount of research dealing with the other adverse events that R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 12 frequently occur when companies disclose fraud. When a company experiences significant events and these events are not disclosed in the company’s 10-Q or 10-K, they need to be disclosed in a Form 8-K within 15 calendar days pursuant the Securities Act o f 1934. The following are adverse events or factors that frequently accompany fraud disclosures and, therefore, are considered in the research design o f this study: • • • • • • • • • Lawsuit filings or threats of future litigation Auditor resignations and terminations Accounting restatements Qualified or withdrawn audit opinions Bankruptcy filings Changes in senior management Audit violations Trading halts, suspensions, and delistments Terminology used to describe fraud Lawsuit Filings or Threats of Future Litigation When fraud is initially disclosed, the risk o f litigation increases significantly. Cox and Thomas (2003) found that approximately 49 percent of the SEC’s enforcement actions during the period 1997 to 2002 may lead to class action lawsuits. About 21 percent of these enforcement actions related to issuer financial statement and reporting acts, 20 percent related to securities offerings, and 8 percent related to insider trading acts. Feroz et al. (1991) found that 81 percent o f those companies with enforcement actions for unacceptable accounting practices were sued by shareholders. Because existing evidence indicates lawsuits frequently result when a company discloses fraud, the expectation o f future litigation besides the fraud itself can lead to a significant decline in company value. Some owners who sense that an initial disclosure R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 13 of fraud will result in litigation will sell, or attempt to sell, their holdings immediately after the disclosure is made. Other risk-taking shareholders may wait to see if the company will incur subsequent litigation after the fraud is initially disclosed. If so, they will likely sell their holdings on or very close to the date of the first class-action lawsuit filing announcem ent. Thus, signals can be sent to market participants on two dates - the initial fraud disclosure date and the date of the first class-action lawsuit filing. Two studies that have examined the market’s reaction to fraud are Hedge et al. (2003) and Elayan et al. (2002). Hedge et al. (2003) tested not only the market’s reaction of initial fraud disclosures, but also the disclosure of the first class-action lawsuit filing. They found that companies’ market values declined 35.0 percent upon the initial disclosure of the fraud and that there was an additional 5.3 percent decline when the first class-action lawsuit filings were announced. Elayan et al. (2002) investigated market’s reaction to accounting irregularity announcements and found that companies’ market values decline 25 percent when irregularities are publicly disclosed. Furthermore, abnormal returns were significantly negative four days prior and two days prior to the initial fraud disclosure, indicating that news o f accounting irregularities are ‘leaked’ out prior to disclosure.2 This study controls for the leakage of information by investigating returns up to four days prior to the disclosure of fraud when investigating the market’s reaction to the initial fraud disclosure. Elayan et al. (2002) also found that market participants reacted more 2 The decline in abnormal returns four days prior to the initial fraud announcement was significant at the .01 significance level and the decline two days prior to the initial fraud announcement was significant at the .10 significance level. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 14 adversely to initial fraud disclosures when the company trades on the NASDAQ, versus the New York Stock Exchange (NYSE). This is likely due to market participants’ concerns about future litigation losses. Another reason market participants may react more adversely to companies that trade on the NASDAQ is that the delisting process on the NASDAQ is informal, while other exchanges normally have a formal delisting process requiring court cases and possibly appeals before delistment o f a company. Therefore, market participants may be concerned that if a company disclosing fraud is listed on the NASDAQ, it may be at greater risk o f being delisted. Existing fraud literature also provides evidence that indicates which fraud types are most vulnerable to litigation. Bonner and Palmrose (1998) found that approximately 58 percent o f their sample of 261 companies with enforcement actions resulted in either company or auditor litigation. This study also found that frauds involving fictitious transactions and frequently occurring frauds are the two fraud types most vulnerable to auditor litigation. Fictitious transactions frauds are those that involve the making o f false documents to conceal the fraud. Fictitious transactions are likely frequently litigated because it is more difficult for the fraud perpetrators to argue against physical evidence, such as false sales invoices, that prosecuting attorneys can present during trial. The most frequently occurring fraud are those involving fictitious revenues. Both Feroz et al. (1991) and Bonner and Palmrose (1998) found that over 50 percent of SEC enforcement actions consist o f fictitious revenues. This finding indicates prosecuting attorneys have become skilled in litigating these types o f cases which increases their probability of prevailing against the defendants. For frequently occurring R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 15 frauds, there are higher expectations for auditors to detect them because market participants likely perceive auditors as having significant experience with them.3 Because of the high probability o f prevailing in either type o f fraud, prosecuting attorneys may have incentive to fight them in court. Taking into account the high frequency o f enforcement actions that lead to litigation and the fact that some fraud types are more vulnerable to auditor litigation, an extension of this evidence is to test to see if market participants can effectively predict litigation based upon specific fraud type. In other words, can market participants predict if frequently occurring frauds will lead to future litigation? While Bonner and Palmrose (1998) found both fictitious transactions and frequently occurring frauds are significantly associated with auditor litigation, in this study I will only test frequently occurring frauds because it is too difficult to determine if fraud was concealed by fictitious transactions in the initial fraud disclosure. Fictitious revenues will serve as a proxy for frequently occurring frauds because it is the type o f fraud that has been found to be most vulnerable to enforcement actions (Feroz et al., 1991; Bonner and Palmrose, 1998). Auditor Resignations and Terminations When fraud is detected, a company’s independent auditor will frequently resign from an engagement or, in some cases, be terminated by the client. Because the SEC requires the company to promptly disclose news o f an auditor resignation or termination, this alone suggests that the announcement provides material new information to the 3 Approximately 38 percent o f the sample used by Bonner and Palmrose (1998) consisted o f enforcement actions that resulted in auditor litigation. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 16 market. There is also evidence that indicates the market reacts negatively to auditor resignation announcements. Wells and Loudder (1997) document a significant market reaction on the day o f the 8-K filing and the day after when the media reports the resignation o f the company’s auditors. In contrast to Wells and Loudder (1997), Defond and Smith (1992) found that most o f a company’s market value decline associated with an auditor resignation announcement occurs prior to the 8-K filing date. Defond and Smith (1992) found negative returns up to 17 days prior to the 8-K filing date, indicating that the market can foresee the resignation o f the auditor before it is announced. Because of the evidence found by these studies, it is expected that simultaneous disclosure o f fraud and an auditor resignation announcement will affect adversely affect the market value o f a company, possibly beginning many days prior to the initial disclosure date. Accounting Restatements An accounting restatement is normally necessary after fraud is detected. Palmrose et al. (2001) found that 69 percent o f their sample o f 403 restatement announcements provided information on who required the restatement. Normally, those responsible were either management of the company, the company’s independent auditor, or the SEC. When a material accounting restatement announcement is made, it is likely that this news will cause a significant market reaction. They found abnormal returns of -9.2 percent when companies announce accounting restatements, with 1.25 years being the average number o f years restated. In addition, the study also found that most o f the market value decline was attributable to the revenue account and that more severe R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 17 reactions occur when the perception of management’s integrity and competence is low and when the amount of the accounting restatement is not disclosed.4 Thus, it is reasonable to expect that the market’s reaction to news of an accounting restatement made after material fraud is detected will be significant. Qualified or Withdrawn Audit Opinions Initial fraud disclosures may be announced simultaneously with qualified audit opinions or opinions that have been withdrawn by the auditor. Ameen and Guffey (1993/1994) investigated the market’s reaction to first time audit qualifications for changes in trading volume and bid-ask spreads5. The investigation period consisted of the week prior to the disclosure of a qualified opinion, the week during the disclosure, and the week after the disclosure. They found that trading volume decreases during the week the qualified opinion was disclosed, but not the week before or after disclosure. In all three weeks, the bid-ask spread increased, but the increase was insignificant. Based on evidence found by the Ameen and Guffey’s (1993/1994) study, it is likely that qualified audit opinions simultaneously disclosed with initial fraud disclosures will contribute to a decline in market values. It is possible that the qualified audit opinion arose from the company’s fraud. However, the fraud also may be a symptom of some other event(s) that led to the qualified opinion, such as the company’s ability to continue as a going concern. Again, a situation exists that indicates other factors reported in the initial fraud disclosure may impact the market’s reaction besides the fraud itself. 4 Skinner (1994) reports that bad news, as measured by adverse earnings announcements, is normally disclosed in qualitative terms, rather than quantitative terms. 5 A limitation o f this study is that only 46 firms that traded over-the-counter were tested to obtain these results. This reduces the generalizability o f these results. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 18 Bankruptcy Filings Another adverse event that may be simultaneously disclosed with fraud disclosure is bankruptcy announcements. Calderon and Green (Winter 1998/1999) found that approximately 38 percent o f their sample of banking companies subject to enforcement actions eventually went bankrupt.6 Companies tend to experience a significant decline in market value when bankruptcy announcements are made (Eberhart et al., 1990). may be associated with bankruptcy for various reasons. Fraud For example, fraud may be committed to conceal and/or enhance failing financial performance, while on other occasions the fraud may be so material and cause such significant losses to the company that the fraud itself forces the company into bankruptcy. Changes in Senior Management The ACFE’s Report to the Nation on Occupational Fraud and Abuse (2004) found that approximately 88 percent o f organizational fraud cases result in termination o f the fraud perpetrator(s).7 When a company’s fraud is simultaneously disclosed with the termination or resignation o f at least one member of senior management, the market may react in several ways. A negative signal may be sent to market participants regarding management’s integrity, and the adverse perception of one management member’s integrity may have an adverse effect on the perception o f other members o f management. Market participants may also react more adversely because the announcement o f the firing or forced retirement confirms the existence of the company’s fraud. Companies 6 Generalizability o f these results may be questionable considering that only banking companies, which are subject to high regulation, were included in the sample. 7 This result is based on 428 responses o f the 508 occupational fraud cases the ACFE sampled. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 19 generally would not want to unnecessarily accept the risk o f being liable for falsely accusing a senior manager of fraud and also bear the announcement cost o f firing a senior executive without strong evidence of the existence o f fraud. Therefore, the announcement provides evidence that fraud exists and that the accusations reported likely are accurate. On occasion, however, the market may also afford the company credit for quickly and effectively identifying and punishing the perpetrator(s).8 Market participants may perceive that there is a reduced risk o f future frauds occurring because the perpetrator is no longer employed by the company. Realistically, though, this factor is not as likely to impact the market’s reaction as much as the first one. Calderon and Green (1998/1999) found that the most common perpetrators of fraud consists of senior vice presidents, board members, owners, the president, the CEO, and the CFO. Thus, when fraud occurs it often does so at the highest level of management and the governance system. Feroz et al. (1991) found that approximately 72 percent o f the SEC enforcement action targets either fired senior management or forces them to resign. Audit Violations Feroz et al. (1991) found that 42 percent o f the target companies in their sample had auditors that were sanctioned by the SEC. With respect to audit firm size, smaller audit firms tend to be penalized more frequently than larger audit firms. Large audit firms may have less penalties imposed upon them because they have less audit failures or 8 Sometimes the identification and the firing or forced retirement may be pursuant the SEC’s actions and not the company’s. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 20 simply because large audit firms can afford better legal representation. Therefore, whether the auditor was involved in the fraud, whether the auditor should have detected the fraud, or whether the audit firm is a large or small firm are all factors that can have a significant impact on how the market reacts to the initial fraud disclosure. Trading Halts, Suspensions, and Delistments Securities exchanges have the authority to halt or suspend a company’s securities trading. A typical halt normally lasts a short period o f time, from a couple o f hours to a day or two. Frequently, when management detects a material misrepresentation in their company’s financial statements, they will be the ones to request that the exchange halt trading of its stock. Halting trading o f the company’s stock will allow management to make an announcement regarding the misstatement and, therefore, reduce losses associated with future litigation. During fraud situations, the SEC can suspend a company off a major stock exchange for up to ten trading days. Calderon and Green (1998/1999) report that 8 percent o f the banking companies in their sample had their trading suspended. Table 2.2 indicates the frequency of SEC suspensions since 1995. Because the target company’s existing shareholders may incur significant losses, the SEC is reluctant to request a suspension o f a company’s trading activities unless it absolutely feels it is necessary. The SEC will not suspend a company unless it believes the public will be making investment decisions based on false or misleading information. Trading does not automatically resume after the SEC suspension ends. If a company trades in the over-the-counter (OTC) market, SEC regulations require a broker-dealer to R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 21 review information about a company before it publishes a quote. However, companies that trade on an exchange, such as NYSE or NASDAQ, will resume trading once the SEC suspension ends. Normally, when a company’s trading resumes, its stock price may be significantly lower because trading suspensions raise significant uncertainties about the company. TABLE 2.2 THE FREQUENCY OF SEC SUSPENSIONS SINCE 1995 Year 2003 2002 Number of Suspensions 8 15 2001 2000 2 11 1999 1998 1997 1996 1995 21 11 16 11 2 Major stock exchanges frequently delist companies when they do not meet the exchange’s minimum requirements.9 An announcement stating a company has been delisted is likely to have more o f an adverse effect on market participants than an announcement stating a company has been suspended because a delistment implies that it is uncertain whether the company will be listed on the exchange again or not. However, management may be the ones to initiate a delistment because it is a way for the company 9 Each exchange is different with respect to the delisting process. For example, the NYSE’s delisting process is formal and consists o f court hearings and sometimes appeals. Conversely, the NASDAQ can delist a company simply because market participants are no longer interested in trading its stock. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 22 to avoid scrutiny by the market. The number o f self-induced delistments in 2003 nearly tripled because of increased regulatory costs associated with the Sarbanes-Oxley Act. Shumway (1997) reports cumulative abnormal returns o f -14 percent the day after a delistment announcement is made and that trading volume is unusually high around the announcement date.10,11 Comparative Systems Corporation provides a good example regarding the differences between a trading halt, SEC suspension, and a delisting off a major exchange. On May 9, 1996, Comparative Systems asked the National Association o f Securities Dealers (NASD) to halt trading o f its shares to give the company time to review and address alleged misstatements in its financial statements. On May 10, 1996, the company disclosed that a material amount of its assets were improperly valued on its balance sheet and that it had not arranged the funding needed to market its latest product, as previously announced. On May 14, the SEC suspended the trading o f Comparative System’s stock for two weeks. On June 12, 1996, Comparative Systems was delisted by NASDAQ because of the improper valuation of the company’s assets. Terminology Used to Describe Fraud Sometimes when fraud is initially disclosed, the term ‘error’ will be used rather than fraud or irregularity. The difference between the term ‘error’ and the terms “irregularity” and “fraud” is that error is an unintentional misstatement or omission of amounts or disclosures in the financial statements, whereas irregularities and fraud are 10 Shumway (1997) investigated this by examining delisted returns. This study further found that there is a delisting bias when using data obtained from Center for Research in Security Prices (CRSP). 11 A limitation o f Shumway (1997) is that the study only investigated over-the-counter (OTC) companies. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 23 intentional acts. Intended misrepresentations, versus unintended misrepresentations, will likely impair market participants’ perception o f management’s integrity. Therefore, it is likely that when initial fraud disclosures contain the terms ‘fraud’ or ‘irregularity’, or n n suggest scienter was present, the market’s reaction may be more adverse. ’ This notion is consistent with Palmrose and Scholtz (2000) which found market participants react more adversely to accounting restatement announcements when the announcement conveys information that reduces the perception o f management’s integrity. Allowing disclosures stating ‘error’ versus ‘fraud’ or ‘irregularity’ will increase the number of testable observations for this study. Summary As discussed above, there are many other reasons why the market may react to initial fraud disclosures other than the fraud itself. Because o f all the other negative events that occur when fraud is initially disclosed, it is very difficult to capture what exactly the market is reacting to. Thus, the model that will be used in this study to interpret the reasons why the market reacts the way it does when fraud is initially disclosed will include several o f the variables that expand upon findings from prior studies. These variables will be used to help us understand what the market is specifically reacting to when fraud is initially disclosed and will allow us to determine if companies that voluntarily disclose fraud experience less market value decline than companies that involuntarily disclose fraud. 12 Scienter is required for a violation o f the antiffaud provisions. Aaron v. SEC, 446 U.S. 680, 691 (1980). 13 The Supreme Court defines scienter as “a mental state embracing intent to deceive, manipulate, or defraud.” Ernst & Ernst vs. Hochfelder, 425 U.S. 185, 193 n.12 (1976). R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 24 Voluntary Disclosure Literature This study seeks to identify a loss-reducing benefit associated with voluntarily disclosing fraud. According to positive accounting theory, managers may disclose bad news for several reasons, including contractual efficiency, political reasons, and opportunism (Watts and Zimmerman, 1986). While there is no specific law or SEC rule that requires companies to disclose fraud, the Securities Act of 1934 requires companies to disclose material events that may affect the decision-making ability o f external users. Hence, if a company’s fraud results in litigation, as it frequently will, the court will consider a company’s forthcomingness when determining punitive and civil penalty amounts. Thus, the decision to disclose fraud is a strategic one. A big concern for companies that do not voluntarily disclose fraud is that this information will eventually leak to other third parties such as the companies’ suppliers, customers, investors, and creditors. Consequently, management will be perceived to have been concealing the fraud and, thus, lack integrity. Companies must decide whether the benefits of early disclosure of fraud exceed the costs. The only benefit significant enough to outweigh the cost o f disclosing fraud early is the reduction in company market value incurred because o f uncertainties involving the companies’ future prospects. There are several factors associated with management’s choice o f voluntarily disclosing fraud versus involuntarily disclosing fraud. One motivation management may have is to try to reduce the risk of future litigation or SEC investigation. Skinner (1994) performed a study to determine when and why managers would voluntarily disclose both good and bad earnings announcements. He found that managers are more likely to R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 25 disclose bad news earnings announcements earlier than they would good news announcements. He found that managers’ incentive for disclosing the bad news early was likely because they wanted to reduce losses associated with future litigation. Future litigation losses will be reduced because the concealment period is shorter and, therefore, the suing parties will have a shorter time frame to argue that the bad news was concealed from them.14 A shorter concealment period would also reduce the amount o f damages a suing party can claim. If market participants’ thinking is consistent with evidence found by Skinner (1994) that managers have an incentive to disclose bad news voluntarily to reduce future litigation losses, then an inference can be made that market participants might perceive voluntary disclosure of fraud as a strategic maneuver to mitigate losses associated with future litigation. In addition, another motivating factor that market participants may perceive for disclosing fraud early is to reduce the possibility and the cost o f a SEC investigation. The SEC may be less likely to investigate a company that voluntarily discloses fraud because the company’s managers have proven themselves to be effective self-monitors o f fraud, thus defeating the purpose of a SEC investigation. Normally, initial fraud disclosures that are voluntarily made will include a statement that indicates the fraud was detected after performing an internal investigation. However, some internal investigations are motivated by management o f the company and some may be requested by the SEC. Therefore, it cannot be assumed with certainty that a company’s fraud was detected through an effective internal investigation unless 14 Skinner (1994) also found that the market reaction to bad news, versus good news, is more significant. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 26 information is disclosed in the initial fraud disclosure that indicates the internal investigation was instigated by management or if something in the disclosure specifically said the SEC was not involved yet up to this point. For example, some initial fraud disclosures report that the fraud was internally detected, and that the company subsequently notified the SEC. Market participants may also adopt a common sense notion that management would not disclose fraud early if they thought it would lead to increased losses associated with lawsuits and a SEC investigation. Hence, who is likely to be better predictors than management of future litigation or a SEC investigation? However, this line o f reasoning will be less likely to exist if alternative incentives exist other than attempting to avoid losses, such as when management’s compensation is heavily incentive-based. Another possible factor motivating management to voluntarily disclose fraud is reduction o f the media exposure period. As the period of media exposure increases, the company will likely see a larger decline in market value due to reputational effects, such as having trouble reestablishing and maintaining relationships with suppliers and customers. Furthermore, the longer the media exposure period, the longer the time attorneys o f shareholders filing class-action lawsuits have to prepare their case against the defendant-company, which may increase the likelihood of the company incurring losses from lawsuits. Management may also have opportunistic reasons to voluntarily disclose fraud. Aboody and Kasznik (2000) performed a study to see if CEOs will opportunistically manipulate the timing of bad news and good news earnings announcements when R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 27 compensated with fixed stock options. There sample consisted of 572 companies that compensated their CEO with fixed stock option plans. They found that CEOs will accelerate bad news earnings announcements and delay good news earnings announcements to maximize their own wealth from exercising their stock options. While it can be argued that shareholders may have obtained an indirect benefit when bad news is accelerated, these findings suggest that shareholders will likely realize that management was the intended direct beneficiary o f the fraud when it is disclosed soon before an award date. Consequently, shareholders’ perception o f management’s integrity may be more adverse when fraud is disclosed soon before an award date. Yermack (1997) performed a study similar to Aboody and Kasznik (2000), but investigated CEOs who were awarded with variable stock option plans. Thus, the objective o f this study was to determine not if managers manipulated disclosures to maximize their personal wealth, but rather to determine if managers would manipulate the award date to maximize personal wealth. He found that managers’ would opportunistically set exercise dates prior to good news announcements and after bad news announcements to maximize their own profits. Both Aboody and Kasznik (2000) and Yermack (1997) found evidence that suggests managers opportunistically use corporate disclosures to maximize their own wealth. However, this does not mean that others do not benefit as well. If bad news is disclosed immediately upon detection, shareholders have the opportunity to act knowing the true financial position of their company to minimize their losses. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 28 Another incentive managers may have to voluntary disclosed fraud is to increase its disclosure quality by making their company appear forthcoming. In other words, disclosing fraud early makes a company appear forthcoming which, in effect, increases market participants’ perceptions o f management’s integrity and the reliability o f the company’s financial statements. Disclosure Quality When a company increases the quality o f its financial reporting, it will likely result in higher analyst ratings. Unfortunately, the metrics developed by researchers after the Association of Investment Management and Research (AIMR) discontinued its disclosure ratings in 1997 do not consider the timing of disclosures as a factor of disclosure quality. Prior research has found that companies can benefit from enhancing the quality o f their disclosures. Thus, market participants may perceive management’s intentions for voluntarily disclosing fraud related to the benefits associated with increased disclosure quality. Botosan (1997) investigated how managers try to reduce external financing costs by increasing disclosure quality. She sampled 122 companies in the machinery industry during 1990 and she found an inverse relationship between disclosure quality and companies’ external financing costs for companies with low analyst following, but no relationship for companies with high analyst following. She suggests that the reason she found no association between companies with high analyst following and cost of capital is because her disclosure index was based on the annual report, and companies with high analyst following are likely to communicate to the public through the analysts R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 29 themselves. A limitation of her study is that she did not consider companies’ timeliness of disclosure, which could have been captured by identifying whether companies are forthcoming or nonforthcoming with bad news announcements such as fraud.15 Prior research studies found that more information is made public about larger companies relative to smaller companies (Atiase, 1985). Other studies have found that the more analysts following a company, the more information there is available for market participants (Dempsey, 1989; Lobo and Mahmoud, 1989; Shores, 1990). These findings suggest that the market may react differently to fraud disclosures made by large, high analyst companies following than fraud disclosures made by small, low analyst following companies. Corporate Securities Regulation Literature This section reviews the existing documentation o f the SEC’s investigation process and the literature dealing with the legal aspects o f corporate fraud and fraud disclosure as it relates to this study. Since its creation in 1934, the SEC’s mission has been to administer and enforce the federal securities laws in order to protect investors, and to maintain fair, honest, and efficient markets. The SEC works very closely with Congress, other federal departments and agencies, self-regulatory regulators such as stock exchanges, state securities regulators, and various private sector organizations to provide assurance that the market has relevant and timely information regarding the financial 15 Other limitations o f Botosan (1997) include the use o f only one industry in her sample during only one year. Also, her disclosure index was self-created, and therefore, might be subject to some bias. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 30 status o f publicly-held corporations and to prevent fraudulent uses o f internal information. Pursuant the 1934 Act, all publicly-held corporations must disclose to the SEC any material events that affect the financial status o f the corporation, which includes fraudulent acts. When a target company is notified by the SEC that it is being formally investigated, this is considered a material event. While the SEC has no specific rule that a company disclose to the public that the SEC is formally investigating, it has been customary to do so. Because current accounting literature does not provide an accurate depiction o f the SEC investigation process, a summary o f the investigation process as discussed by Mahoney et al. (2002) is presented next. The SEC Investigation Process The SEC investigation period begins after the SEC obtains leads which may come from public complaints, market surveillance programs o f the major exchanges, communications with state and federal agencies, and reviews o f 1933 and 1934 SEC filings. The first phase of the investigation period is called the informal (or preliminary) investigation. During this time, the Commission sends some o f its staff investigators to investigate the alleged fraud.16 The staff members will seek cooperation from informants who have relevant information regarding the fraud. Management has incentive to voluntarily produce the documents and testimony requested by the staff investigators 16 The terms Commission and staff investigators are used for the remainder o f this paper. Staff investigators are the individuals that work for the enforcement division responsible for investigating companies suspected of unacceptable accounting practices. The term Commission is used, as opposed to SEC, because o f the need to discuss staff investigators apart from the SEC. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 31 because lack of cooperation is the primary reason informal investigations lead into formal investigations. Once staff investigators feel they have obtained sufficient evidence to warrant a formal investigation, they will present this evidence to the Commission. The Commission will consider certain factors such as the cost involved in formally investigating this company, the likelihood the formal investigation will provide the SEC with sufficient evidence to warrant filing a lawsuit against the company, and whether or not the SEC thinks it will prevail. Because the SEC has limited resources, it cannot investigate all companies that need to be investigated.17 If the Commission feels sufficient evidence has been obtained during the informal investigation, it will send a Wells notice to the target which gives them one last chance to state reasons why a formal investigation is not necessary and why the Commission io should take no further action. If the SEC is not convinced by management’s reasons, then it will proceed into a formal investigation. The primary reason the Commission would turn an informal investigation into a formal investigation is to obtain subpoena power because management did not, or could not, furnish needed documentation or testimony to draw a conclusion about the issue being investigated.19 Because companies may incur significantly more costs during a formal investigation, an incentive exists for 17 After the Sarbanes-Oxley Act (2002) was passed, the enforcement division o f the SEC had a significant increase in its budget. 18 The staff investigators are not required to issue a Wells notice in certain circumstances such as there is concern regarding the liquidation o f assets, destruction o f evidence, management poses flight risk, or if necessary to protect the public interest. If the staff investigators do not issue a Wells notice, the company may still submit reasons why a formal investigation is not necessary. 19 Testimonial evidence, during the formal investigation, is normally given under oath before a court reporter. This is why news media reports many senior managers pleading the Fifth Amendment privilege against self-incrimination in testifying. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 32 management to be forthcoming with documents and testimony during the informal investigation. Once the SEC begins a formal investigation, it becomes difficult to keep this information concealed from the market. While the SEC maintains a policy o f not discussing information related to their investigation, ‘leaks’ o f information will likely occur. This occurs because once the SEC transcends from a informal investigation to a formal investigation, the staff investigators begin to exercise subpoena power, which means it may contact external parties, such as suppliers, customers, prior employees, etc. Because the nature o f each investigation varies, the SEC has no set guidelines, but only approaches. Management’s cooperation, though, is o f great importance to the SEC when they perform their formal investigation. The primary reason managers will cooperate in the formal investigation is because they hope it will lead to a more beneficial outcome. Both the company and the SEC normally benefit from cooperation because the investigation period will be shorter, which saves both parties time and costs. Once staff investigators gather enough evidence during the formal investigation to convince themselves the company committed accounting fraud and will support legal proceedings, they will present this evidence to the Commission. The Commission will then decide whether or not to authorize the staff to file either administrative or civil charges.20,21 Administrative cases are held by a judge that is independent o f the SEC and no jury is used. The judge will give an initial decision which can be appealed by the SEC or 20 The SEC cannot file criminal charges. However, it does work very closely with some criminal investigations. 21 Some factors considered when the SEC decides how to proceed include the seriousness o f wrongdoing, the technical nature o f the matter, tactical considerations, and the type o f sanction or relief to obtain. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 33 the company. A final administrative sanction may include cease-and-desist orders, suspension or revocation of registrations, censures, bars from association with the securities industry, payment of civil penalties, and disgorgement. Civil cases are heard in a district court where juries are sometimes used to decide the case and the SEC will usually request an injunction that prohibits the company and the individuals involved from violating SEC regulations.22 The SEC may seek different types o f punitive and compensatory damages in a civil case consisting o f civil penalties, disgorgement (ill-gotten gains/profits), disbarment, termination from serving as a director or officer, cease and desist orders, suspension or revocation o f broker-dealer and investment adviser registrations, and censures. The SEC can seek either civil penalties and/or disgorgement against any individual or company in federal court for violations o f securities laws. When doing so, the SEC must show that it is in the public’s interest to impose civil penalties and/or disgorgement. The following is criteria used by the SEC when determining if it will seek civil penalties: • • • • • • Whether the act was fraudulent or not. The harm imposed upon others by the act. The extent o f unjust enrichment to anyone, taking into account the extent to which restitution has been made to persons injured. Prior securities violations. To deter future violations. Where permitted by other regulations. 22 The injunction also can require a target company or individual to perform various actions, such as audits, accounting for frauds, or special supervisory arrangements. 23 To prevent the spending or dissipation o f disgorgement during an on-going investigation, the SEC will likely seek a temporary restraining order. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 34 The recently enacted “Fair Fund” provision authorizes the SEC to designate civil penalties collected from fraud perpetrators to benefit injured shareholders. This enactment is advantageous to injured shareholders because obtaining full disgorgement amounts from fraud perpetrators is rare. Thus, the SEC may make civil penalty amounts available to the injured shareholders, as opposed to flowing directly to the U.S. Treasury. This act also allows the SEC to bar or suspend an officer or director who has violated the antifraud provisions. The SEC administers four statutes which provide for civil penalties: The Securities Exchange Act o f 1933, the Securities Exchange Act of 1934, the Investment Company Act o f 1940, and the Investment Advisers Act o f 1940. Table 2.3 illustrates the three-tiered system used by the SEC in determining civil penalties. Except for insider trading acts, the maximum civil penalty is currently $120,000 for individuals and $600,000 for the company. Not all fraudulent activities are encompassed in the three tier system. For insider trading, the civil penalty cannot exceed three times the profit gained or loss avoided related to the sale or trade o f stock involved. The maximum civil penalty for violations o f the Foreign Corrupt Practices Act (FCPA) for either individuals or corporations is $11,000. The SEC may also seek disgorgement from an individual or company. Disgorgement is the ill-obtained profits (gains) the company or perpetrators receive when performing a fraudulent act. While disgorgement is most frequently imposed for insider trading, it can also be imposed for other illegal acts. Disgorgement is not calculated based upon the harm of the injured parties, but rather, it represents the amount the fraud perpetrator(s) accumulated from their fraud. The Sarbanes-Oxley Act o f 2002 has R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 35 changed the handling o f disgorgement paid by target companies or their managers, which used to be paid directly to the U.S. Treasury. Now, the disgorgement funds are set aside in a trust to be distributed upon settlements of future litigation resulting from the fraud. TABLE 2.3 CIVIL PENALTIES FOR CIVIL ACTIONS (updated February 1, 2001) Nature o f offense (Based on degree o f guilt and severity o f offense) Tier Civil penalty amount (can’t exceed this amount) Individuals First Non-serious offenses Second Third Acts o f fraud, deceit, manipulation, or deliberate/reckless disregard o f regulatory requirement, such as failure to file Acts of fraud, deceit, manipulation, or deliberate/reckless disregard of regulatory requirement, and the violation directly or indirectly resulted in substantial losses or created a significant risk o f substantial losses to others Corporations $6,500 $60,000 $60,000 $300,000 $120,000 $600,000 Although the SEC normally seeks civil penalties upon actual perpetrators, the targeted company also may receive civil penalties. This is a peculiar act for the SEC, considering its primary objective is to watch out for the investors’ interests and these penalties will ultimately come out of the company’s shareholders’ pockets. However, it is likely that the SEC will seek civil penalties and disgorgement from companies when R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 36 they feel an example can and should be made. For example, WorldCom is a company whose fraudulent accounting practices received massive media exposure. WorldCom settled its case with the SEC by agreeing to pay civil penalties o f $500 million. It is not uncommon for companies to settle their case with the SEC before the court finds the company guilty o f fraud so that the companies can avoid excessive reputational damages. Before deciding whether to litigate or settle, the target company will consider other litigation and criminal investigations pending, the anticipated decline in the company’s market value, the expense of litigating the SEC, and the probability o f a favorable versus unfavorable outcome. Feroz et al. (1991) found that normally companies settle with the SEC by consenting to an injunction that prohibits future violations o f securities laws. There has been limited research dealing with the SEC’s intentions when determining penalties associated with unacceptable accounting practices. Beneish (1999) found that the SEC is not likely to sanction managers of companies with overstated earnings unless they participated in insider trading practices. Other Relevant Fraud Research Some fraud studies investigate what organizational characteristics are associated with fraud. Barnes and Lambell (2003) performed a study to determine if fraud affects organizations randomly or if certain characteristics make some organizations more vulnerable to fraud, such as industry classification, size, and organizational type (private versus public). They found that a company’s industry and its type are significant factors that make a company more vulnerable to fraud. They also found that larger companies R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 37 are more vulnerable to fraud than smaller companies, with the dollar amount o f loss associated with fraud increasing exponentially with company size. In addition, they found a positive relationship between fraud duration and fraud size and a positive relationship between fraud duration and collusion and seniority o f the perpetrator. Fraud duration and involvement has also been examined by other studies. Palmrose et al. (2001) found evidence that fraud duration does not result in a significant market reaction. This study found that in an accounting restatement situation, the number o f years restated does not have an increasingly adverse reaction on the market when the years restated increases. Adverse market reaction to accounting restatement announcements is increased, though, when the announcement suggests that management’s integrity is impaired. Calderon and Green (Winter 1998/1999) report that management frauds normally involve three or four officers per case.24 Most fraud literature reports fraud duration to average approximately two years, or a little longer (ACFE, 2004; Bonner and Palmrose, 1998; Calderon and Green, 1998/1999). 24 The actual finding was 3.26 officers per case. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. CHAPTER III HYPOTHESES AND RESEARCH DESIGN This study seeks to determine if benefits are obtainable by companies that voluntarily disclose fraud. The benefits will be measured in terms o f a reduction in three loss areas: (1) the decline in company market value, (2) the time and resources utilized in a SEC investigation, and (3) civil penalties and disgorgement. Investigating the Market’s Reaction to Fraud I theorize that voluntarily disclosing companies will experience a less significant decline in market value than involuntarily disclosing companies for several reasons. First, there are normally less adverse signals sent to the market when fraud is initially disclosed, as depicted below: Signal sent to the market 1. At least one employee of this company committed fraud 2. This company’s internal controls did not prevent the fraud Voluntarily Involuntarily disclosing disclosing companies companies a/ a/ a/ a/ 3. Management was not competent enough to prevent, detect, or keep the fraud concealed. V 4. This company attempted to conceal the fraud from shareholders, regulators, etc. V 38 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 39 As indicated above, there are likely two more adverse signals the market receives when fraud is involuntarily disclosed. Management will appear incompetent not only because their internal controls failed to prevent the fraud, but also because they could not effectively conceal it. Furthermore, the implication for involuntary disclosing firms is that management attempted to conceal the fraud or, at least, delay the disclosure o f the fraud. This is perhaps the most crucial o f the four adverse signals sent to the market upon an initial fraud disclosure because this suggests management’s integrity is impaired. Furthermore, when management attempts to conceal fraud from shareholders, market participants will likely perceive management as trying to maximize their own wealth at the expense of the shareholders. This evidence is consistent with findings o f Palmrose et al. (2001), where it was found that the market reacts more adversely to accounting restatements when the perception o f management’s integrity is impaired. In addition, the longer the fraud concealment period is, the greater concern there may be associated with future litigation and a SEC investigation. When a company voluntarily discloses its fraud, it is sending a signal to the market that the likelihood o f subsequent litigation or a SEC investigation is low. Otherwise, why would a company voluntarily disclose its fraud if management anticipates the disclosure will lead to either a shareholder lawsuit or SEC investigation and greater losses than otherwise? Another reason the decline in market value may be less for voluntary disclosing companies is that companies disclosing the fraud early are generally less likely to disclose the amount of fraud. One reason management may not disclose the dollar amount o f the fraud when voluntarily disclosing is because the dollar amount may not yet be available as it might R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 40 be for companies that wait until later to disclose fraud. Palmrose et al. (2001) found that the decline in returns associated with accounting restatement announcements is less when the amount of the restatement is not disclosed, as opposed to when the amount is disclosed. Feroz et al. (1991) found that the income effect of fraud, when disclosed in an initial announcement regarding an unacceptable accounting practice, has a significant impact on the market. The average income effect o f their sampled companies was greater than 50 percent.1 This study investigates the market’s reaction to fraud announcements that are voluntarily disclosed versus involuntarily disclosed. I theorize that voluntary disclosure of fraud is perceived by market participants less adversely than involuntary disclosed fraud. In addition to being less likely to contain fraud amounts and other details about the fraud than involuntary fraud disclosures, when fraud disclosures are voluntary made, management is more likely to appear in control and is more likely to report that the fraud was detected pursuant an effective internal investigation. The market’s perception o f management’s competence also increases, with the appearance o f the company having an effective internal control system or fraud program in place to mitigate the risk o f future fraud occurrence. Therefore, I will investigate the following research question: H I: Companies that voluntarily disclose fraud will experience a smaller decline in company market value upon initial disclosure of fraud than companies that involuntarily disclose fraud when controlling for the following factors: company size, degree of regulation, audit firm size, fraud size, fraud type, fraud duration, 1 Feroz et al. (1991) can be criticized for using income effect as a proxy for fraud size because not all fraud types affect income. Some frauds, such as improper asset valuations and understatement o f liabilities, affect the balance sheet. Furthermore, some frauds affect both balance sheet and income statement items. Hence, it is more practical to relate the fraud amount to equity. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 41 auditor violations, terminology used in disclosure, source o f disclosure, announcements regarding a SEC investigation, litigation, change of auditor, qualified (withdrawn) audit opinion, bankruptcy filing, and change in management. Investigating the Duration of the Media Exposure Period and Lingering Litigation Period Managerial cooperation is o f great importance to the SEC when they perform their investigations. Normally, the only incentive present for management to cooperate is the hope that cooperation will lead to a more beneficial outcome. This explanation is consistent with the SEC’s primary reason to move into a formal investigation because management did not provide them with necessary documentation or testimonial evidence. Therefore, both the SEC and target company are expected to benefit from cooperation because the entire investigation period will be shorter, which saves time and costs for both parties. I theorize that when a company voluntarily discloses fraud, effective cooperation is enhanced between management and the SEC. Consequently, the SEC will likely afford more credibility to information, both in documentation and testimonial form, brought forth by managers during the investigation than when the company involuntarily discloses fraud. In addition, if a company voluntarily discloses fraud, it normally is the result o f an effective internal investigation. The individuals who participated in the internal investigation are more likely to be perceived as having the necessary skills and to be trustworthy enough to assist in the investigation process. Greater cooperation and R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 42 reliability of data will likely reduce the length of the SEC investigation period. Hence, I posit the following research question: H2: Companies that voluntarily disclose fraud will have shorter media exposure periods than companies that involuntary disclose fraud when controlling for the following factors: company size, fraud size, the degree o f regulation, fraud duration, change o f auditor, audit violations, and audit firm size. Normally, there is still litigation between the SEC and fraud perpetrators o f the company after the company and SEC settle. While voluntarily disclosing companies will likely experience shorter media exposure periods than involuntary disclosing companies, the lingering litigation period should not be any shorter between the two groups. The following hypothesis, in alternative form, will be investigated. H3: Fraud perpetrators of companies that voluntarily disclose fraud will have a shorter lingering litigation period than fraud perpetrators of companies that involuntarily disclose fraud when controlling for the following factors: company size, fraud size, degree of regulation, fraud duration, change of auditor, audit violations, size o f audit firm, and the duration of the media exposure period. Investigating Civil Penalty and Disgorgement Amounts Civil penalties and disgorgement are frequently sought from and/or imposed upon individuals as opposed to the companies. A question exists as to whether or not the amount of civil penalties and/or disgorgement imposed upon fraud perpetrators o f companies that voluntarily disclosed their fraud is less than for fraud perpetrators o f involuntary disclosing companies. There are two alternative arguments that can be taken. On one hand, the same amount of civil penalty and/or disgorgement may be sought from/imposed upon individual fraud perpetrators’ whose company voluntary disclosed its fraud or R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 43 involuntarily disclosed its fraud because the fraud perpetrators likely had little or nothing to do with the choice to be forthcoming with the fraud. Conversely, less civil penalties and/or disgorgement amounts may be sought from/imposed upon fraud perpetrators o f voluntary disclosing companies because o f the shorter media exposure period. Similarly, the reduction in disgorgement amounts the SEC seeks (and the courts impose) from fraud perpetrators o f voluntary disclosing companies may be less because a shorter fraud concealment period reduces the amount o f injuries incurred by the plaintiffs (shareholders). The following research questions will be investigated: H4: Civil penalties imposed upon fraud perpetrators of voluntary disclosing companies are less than civil penalties imposed upon fraud perpetrators o f involuntary disclosing companies when controlling for the following factors: fraud size, frequently occurring frauds, the duration o f the total investigation period, audit violations, and disgorgement amounts. H5: Disgorgement imposed upon fraud perpetrators of voluntary disclosing companies is less than disgorgement imposed upon fraud perpetrators o f involuntary disclosing companies when controlling for the following factors: fraud size, frequently occurring frauds, the duration o f the total investigation period, audit violations, and civil penalty amounts. Research Design Two methods are used to test the market’s reaction. First, a market model is used to estimate abnormal returns. Second, a multivariate regression models is developed to test the market’s reaction to fraud during the 11-day window. The following marketadjusted model is used to calculate market adjusted abnormal returns: R jt ~ Uj b j R mt ~F d jt Rjt is the companies stock returns and Rmt is the market index stock returns obtained from CRSP. The companies’ stock returns are regressed against the market index stock R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 44 returns. The objective of using the market model is to generate expected returns so that unexpected returns specifically related to the voluntary disclosure o f fraud during the 11day investigation period can be inferred. Testing the Company Decline in Market Value HI tests the market’s reaction to fraud that is voluntarily disclosed versus fraud that is involuntarily disclosed during the eleven day window. The following is the regression model designed to test HI prior to any extraction of variables due to collinearity: CARj = CARi = b0 + biVD + b2CO_SIZE+ b3FS + b4FD + b5TERM + b6SOURCE + b7SECINV + bgSUIT + b9AAUD + bi0AMGMT + b n AUDVIO + b 12FOF + bi3REG + bi4BIG_X + bi5OP!N + bi6BKRT + bnC O LL + bigPERP + bi9(COLL*PERP) + a Cumulative abnormal returns for days {-5, -4, -3, -2,-1, 0, 1, 2, 3, 4, 5}, where 0 equals the date fraud is initially disclosed. VD = 1 if a company disclosed fraud before it was notified that it was being investigated by the SEC and the fraud was not announced by an external party prior to the company’s announcement, 0 if otherwise. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 45 CO SIZE = The log o f total assets at the end o f the last year o f the fraud. FS = 1 if the fraud amount is disclosed during the initial fraud disclosure window, 0 if otherwise.3 FD = 1 if the fraud duration is disclosed during the initial fraud disclosure window, 0 if otherwise. TERM = 1 if the term ‘fraud’ or ‘irregularity’ was used to describe the act reported during the initial fraud disclosure, 0 if otherwise.4 SOURCE = 1 if the fraud was initially reported in an AAER, 0 if otherwise. SECINV = 1 if a SEC investigation was pending during the initial fraud disclosure window, 0 if otherwise. SUIT = 1 if a fraud-related lawsuit was pending during the initial fraud disclosure window, 0 if otherwise. 2 If the total asset amount was not available at the end of the last year o f the fraud, thelast reported total asset amount was used as long as it was reported sometime during the fraud. 3 The fraud amount scaled by total stockholders’ equity would have been used here, but a majority o f the initial fraud disclosures did not disclose the amount of the fraud. 4 To be included in the sample, the act must ultimately be concluded as fraud.However, the initial disclosures do not have to state the terms ‘fraud’ or ‘irregularity’. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 46 AAUD = 1 if an independent auditor resignation or termination was reported during the initial fraud disclosure window, 0 if otherwise. AM GM T = 1 if the initial fraud disclosure reports that the company president, chairman of the board, CFO, or CEO is fired, retires, or takes a leave of absence, 0 if otherwise. AUDVIO = 1 if an audit violation of audit standards was reported during the initial fraud disclosure window, 0 if otherwise. FOF = 1 if the fraud involved the recognition o f fictitious revenues, 0 if otherwise. REG = 1 if a company is in the financial services, insurance, or utilities industry, 0 if otherwise. BIG_ X = 1 if the company’s independent auditors were a large audit firm when the fraud is detected, 0 if otherwise. OPIN = 1 if, during the initial fraud disclosure window, there is a qualified or withdrawn audit opinion, a statement made that the company’s financial statements remained unaudited after the requiring filing R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. date, or a statement was made indicating a company’s financial statements should not be relied upon, 0 if otherwise. BKRT 1 if a bankruptcy filing announcement was made during the initial fraud disclosure window, 0 if otherwise. COLL = The number o f alleged perpetrators reported during the initial fraud disclosure window. PERP 1 if the initial fraud disclosure reported at least one fraud perpetrator as having a title o f president, chairman of the board, CFO, or CEO, 0 if otherwise. COLL*PERP = Equals the number o f fraud perpetrators when at least one perpetrator is president, chairman o f the board, CFO, or CEO, 0 if otherwise. Testing for Collinearity Because there are so many factors to consider when investigating the market’s reaction to initial fraud disclosures, the risk of collinearity is high. Factor analysis was performed to determine if any factors explained 70 percent or more o f the total variance. The variables COLL, PERP, and COLL*PERP exceeded this threshold. Furrther analysis revealed these three variables had variance inflation factors greater than 5. Consequently, R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 48 these three variables were removed from the original model above. Furthermore, no companies in the final sample had qualified opinions or a bankruptcy announcement the day their fraud was initially disclosed. Therefore, variables OPIN and BKRT were removed from the model. The final model for HI is as follows: CARi = b0 + biVD + b2CO_SIZE+ b3FS + b4FD + b5TERM + b6SOURCE + b7SECINV + bgSUIT + b9AAUD + bi0AMGMT + bnAUDVIO + b i2FOF + bi3REG + bi4BIG_ X + e Prediction of V ariables The main variable of interest is VD, which represents whether or not companies voluntarily disclosed their fraud. The coefficient for VD is expected to be positive because market participants are likely to afford companies more credit for being forthcoming with fraud relative to involuntarily disclosing companies. The coefficients for the variables CO SIZE and FS should be negative because fraud size has been found to exponentially increase relative to company size and income effect has been found to be significantly associated with a decline in company market value (Calderon and Green, 1998/1999); Feroz et al. 1991). Although actual fraud size was not used to test HI because the fraud amount is seldom reported in the initial fraud disclosure, I predict the market will react more adverse when fraud is discussed in quantitative terms. This line of reasoning is consistent with Skinner (1994) which found that companies are more likely to disclose bad news in qualitative terms and good news in quantitative terms. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 49 The coefficient for FD should be negative because out o f all the factors Barnes and Lambell (2003) tested, fraud duration was the factor most associated with fraud size. Furthermore, the longer the fraud lasted, the longer management concealed the fraud from market participants, which likely increases the uncertainties regarding future litigation losses and impairs perceptions of management’s integrity. Existing evidence indicates market participants will react more adverse when management’s integrity is perceived low (Palmrose et al., 2001). Therefore, the coefficient for TERM should be negative because the perception o f management’s integrity will likely be less when the terms ‘fraud’ or ‘irregularity’ are reported, as opposed to ‘error’, because frauds and irregularities are intentional acts and errors are unintentional acts. The coefficient for SOURCE should be negative because there is likely more o f a market value decline when fraud is initially disclosed in an AAER than when fraud is disclosed in the other sources of information because AAERs contain more details regarding the fraud and fraud perpetrators. The coefficients on the bad news variables, which include SUIT, SECINV, AAUD, and AUDVIO should all be negative because these items have been shown by prior research to cause adverse reactions upon initial disclosure. The coefficient for FOF should be negative because fictitious revenues have been found by prior research to be significantly associated with litigation. The coefficient for AMGMT should be positive because the firing or retirement o f the fraud perpetrators is likely perceived by market participants as good news because the likelihood o f future fraud is reduced. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 50 The coefficients for REG should be negative because highly regulated companies tend to have higher litigation risk and other losses because there are more regulatory bodies to bring forth lawsuits, fines and penalties. Furthermore, other regulatory agencies, such as the FDA, may have their own disclosure requirements regarding the disclosure o f fraud. The coefficient for BIG_ X should be positive because if a company’s independent auditors are a large audit firm, the company is less likely to face litigation. Given the opportunity, shareholders are more likely to sue the company’s large audit firm, as opposed to the company because of two reasons. First, the large audit firm has “deeper pockets” than the company does, meaning they are likely in a better financial position to compensate shareholders for their losses after a fraud situation than the company is. Second, if shareholders sue the company, the litigation losses ultimately flow back to them. Conversely, if shareholders sue the company’s large audit firm, the litigation losses remains with the audit firm. Table 3.1 provides predictions and variable types for HI. APPENDIX A contains testing of the market’s reaction during the 11 day initial fraud disclosure window while regressing only VD on cumulative abnormal returns. The objective o f this second test of the market’s reaction is to gather evidence regarding whether or not companies that voluntarily disclose fraud experience less o f a market value decline than involuntary disclosers. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 51 TABLE 3.1 SUMMARY OF PREDICTIONS AND VARIABLE TYPE FOR HI Variable Coefficient Sign Dichotonomous Variable Coefficient Sign Dichotonomous VD + Yes SUIT - Yes C O S IZ E - No AAUD - Yes FS - Yes AMGMT + Yes FD - Yes AUDVIO - Yes TERM - Yes FOF - Yes SOURCE - Yes REG - Yes SECINV - Yes B IG X + Yes Investigating the Duration of Investigation Periods H2 and H3 test the association between a company’s forthcomingness with its fraud and the duration of two investigation periods. First, H2 is tested to determine if an association exists between a company’s choice to voluntarily disclose fraud and the duration o f the media exposure period. Next, H3 is tested to determine if an association exists between a company’s choice to voluntarily disclose fraud and the duration of the lingering litigation period. The MEP variable was included as an independent variable in H3 to determine if a significant inverse relationship exists between the time staff investigators spend performing the companies’ investigations and the time they spend performing the fraud perpetrators’ investigations. Its result will indicate if the duration of R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 52 the fraud perpetrators’ investigation is modified based upon the duration o f a company’s investigation. MEPj = bo + bi VD + b2CO_SIZE + b3FS + b4REG + b5FD + b6AAUD + b7AUDVIO + b8BIG_X + e LLPi = bo + biVD + b2CO_SIZE + b3FS + b4REG+ b5FD + b6AAUD + b7AUDVIO + bgBIG_X + b9MEP + e The variables used in models H2 and H3 are defined below: MEPi = The number of days in the period beginning the date fraud is initially disclosed and ending the date the company settles its case with the SEC. LLPj - The number of days in the period beginning the date the company settles its case with the SEC and ending the date the last perpetrator’s case is settled with the SEC. VD = 1 if a company disclosed fraud before it was notified that it was being formally investigated by the SEC and it was not announced by an external party prior to the company’s announcement, 0 if otherwise. CO_SIZE = The log of total assets at the end of the last year of the fraud. FS = Total dollar amount of the fraud scaled by the company’s total stockholders’ equity at the end of the last year o f the fraud. REG = 1 if a company is in the financial services, insurance, or utilities industry, 0 if otherwise. FD = The total number o f days o f the fraud duration. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 53 AAUD = 1 if an auditor resigned or was terminated because o f the fraud, 0 if otherwise. AUDVIO = 1 if it is announced that the company’s independent auditors were involved in the fraud or violated audit standards, 0 if otherwise. BIG X = 1 if the company’s independent auditors are a large audit firm, 0 if otherwise. Prediction of Variables The coefficient for VD is expected to be negative and significant when testing the media exposure period because the staff investigators are more likely to obtain assistance from employees of the target company when they perform their investigation. Conversely, it is likely the coefficient for VD will be insignificant when the lingering litigation period is tested because it is unlikely that staff investigators will use employees of involuntary disclosing companies because these employees are perceived by staff investigators as lacking the sufficient skills and integrity needed to effectively assist with the company’s investigation. While fraud perpetrators frequently do cooperate with staff investigators in hopes of reducing their punishment, the staff investigators cannot rely much upon testimony received from perpetrators because the staff investigators will perceive the perpetrators as lacking integrity. As a result, I have no prediction for the sign o f the coefficient for VD when testing the lingering litigation period. The coefficients for CO_SIZE, FS, and FD should all be positive when testing both the media exposure period and the lingering litigation period. Whether the SEC is gathering evidence against the company or the fraud perpetrators, the staff investigators R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 54 will still be required to search for evidence within the company as well as from fraud perpetrators. Staff investigators will likely consume more time gathering evidence in large companies than small companies because large companies process a larger volume o f transactions through processes that are likely more complex than transaction processing cycles of smaller companies. I have no prediction regarding the coefficient for REG. One argument is that the media exposure period will be longer because of the need to coordinate with the investigation of other agencies, which will extend the amount o f time the staff investigators needs to complete that investigation. Furthermore, the SEC and its staff investigators may be concerned about how their investigatory procedures and evidence obtained are perceived by the other regulatory agencies. Therefore, staff investigators may take more time when a company is highly regulated to make sure their work is not perceived adversely by others. A contradictory argument suggests that highly regulated companies might have a shorter media exposure period and lingering litigation period because these companies are being investigated by several agencies simultaneously. Consequently, the period staff investigators need to gather evidence to prevail in its case against the company and/or fraud perpetrators will be shorter because other agencies are assisting in the evidence-gathering task. This theory is dependent upon how willing other regulatory agencies and the SEC and its staff investigators are to work together. There are three variables related to independent auditors: AAUD, AUDVIO, and BIG_ X. The coefficients for AAUD and AUDVIO should both be positive. If the company’s independent auditor resigns or is terminated, this likely indicates that their client’s fraud R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 55 is serious. If a company’s independent auditors participate in the fraud or violated generally accepted accounting standards (GAAS) which allowed the fraud to go undetected, then staff investigators would not be able to rely upon the auditors’ work during the investigation and will have to assess the reliability o f evidence whether it was audited or not. Conversely, the coefficient for BIG_ X should be negative because if the company’s independent auditors were one of the Big X firms, the staff investigators not only can rely upon their work, but also will be less reluctant to ask them to assist in their investigation than they would o f auditors o f a smaller, less quality, audit firm.5 TABLE 3.2 SUMMARY OF PREDICTIONS AND VARIABLE TYPE FOR H2 & H3 Hypothesis 2 Variable Coefficient Sign VD Hypothesis 3 Dichotonomous VD Coefficient Sign ? Dichotonomous Variable Yes Yes C O S IZ E + No C O S IZ E + No FS + No FS + No REG ? Yes REG ? Yes FD + No FD + No AAUD + Yes AAUD + Yes AUDVIO + Yes AUDVIO + Yes B IG X - Yes B IG X - Yes MEP No 5 The variable is named ‘B I G X ’ because some o f the sampled companies disclosed their fraud at a time when the largest audit firms were not classified as a ‘Big 4 ’ audit firm. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 56 The coefficient for MEP is expected to be negative because this indicates the SEC will likely try to reduce the investigation pertaining to the fraud perpetrators when the duration of the company’s investigation is longer than normal. This is consistent with the fact that the SEC had limited resources during the sample period and, therefore, would likely try to avoid overall lengthy investigations.6 Another reason why an inverse relationship exists between the media exposure period and the lingering litigation period is that much o f the evidence the SEC needs to effectively prosecute fraud perpetrators may be gathered when investigating the company. Hence, the time needed by staff investigators to gather evidence during the perpetrators’ investigation will likely be shorter the longer the company’s investigation period is. Investigating Civil Penalty and Disgorgement Amounts The SEC has the right to seek civil penalties and disgorgement from corporate fraud perpetrators. The third loss amount investigates the civil penalty and disgorgement amounts imposed upon fraud perpetrators by the SEC. I theorize that no relationship exists between a company’s forthcomingness with fraud and civil penalty and/or disgorgement amounts because the fraud perpetrators likely have nothing to do with the choice o f voluntarily disclosing fraud. There is a possibility, though, that the SEC may inappropriately seek less civil penalties from perpetrators of voluntary disclosing companies because the fraud concealment period is shorter when fraud is voluntarily disclosed by the companies. However, because disgorgement amounts are calculated in a 6 The sampled period is from 1997 to 2004. All SEC investigations began prior to 2002, which is when the enforcement division within the SEC received a significant increase in its budget. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 57 more objective manner than civil penalties, it is likely there is a stronger relationship between civil penalties and disclosure choice than the relationship between disgorgement and disclosure choice. To test to see if civil penalty (CP) and disgorgement amounts (DG) for perpetrators vary based on a company’s forthcomingness with their fraud, the following two models are used: CPi = b0 + b)VD + b2FS +b3FOF + b4TIP + b5AUDVIO + b6DG + e DGi = b0 + bi VD + b2FS + b3FOF + b4TIP + b5AUDVIO + b6CP + e All variables in the above models, other than TIP, were previously defined. TIP is defined below: TIP = The sum of the media exposure period and the lingering litigation period. Unlike the prior tests performed in this study, the coefficient for VD (voluntary disclosure) for these two tests is expected to be insignificant. Although the fraud perpetrators have a shorter fraud concealment period because the company voluntarily disclosed its fraud, it is likely the SEC takes into account the fact that the fraud perpetrators normally had little or nothing to do with the company’s choice to voluntarily disclose its fraud when it calculates civil penalty amounts. Disgorgement amounts likely are not influenced by disclosure choice because ill-gotten gains are objectively calculated. The coefficient for FS (fraud size) is expected to be significant and positive for both models because the larger the fraud, the more likely the SEC will perceive the fraud R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 58 perpetrators as deserving to be punished. Furthermore, the larger the fraud, the more resources and time the staff investigators likely will utilize during the investigation. Also, larger frauds are likely exposed to the public more and, therefore, the SEC may feel it has more o f an opportunity to set an appropriate example by seeking larger civil penalties from fraud perpetrators. I predict the coefficient for FOF (fictitious revenues) to be positive for both models because these fraud types have been found to be significantly associated with litigation and, therefore, the SEC is likely reluctant to terminate an investigation when they believe the work of the staff investigators will be closely scrutinized by others in federal court. Furthermore, the SEC may try to compensate the federal government for more time and resources spent during extended investigations by increasing civil penalty amounts sought from fraud perpetrators.7 The amount o f damages caused by the fraud is taken into consideration by the SEC when determining both civil penalty and disgorgement amounts. The coefficient for TIP (total investigation period) will likely be positive for both models. A significant direct relationship between civil penalty and disgorgement amounts and TIP may indicate the SEC may try to compensate extended time and resources consumed during lengthy investigation periods by seeking more civil penalties and disgorgement from fraud perpetrators. 7 The civil penalties and disgorgement amounts are not funds for the SEC’s use. Rather, civil penalties may flow to the U.S. Treasury or, at the discretion o f the SEC, be set aside in a trust to be distributed to injured shareholders after a class action lawsuit settlement. Disgorgement always is set aside in a trust to be distributed to injured shareholders upon settlement of a class action lawsuit. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 59 I predict the coefficient for AUDVIO (auditor violation) to be positive. Whenever a company’s independent auditors either participate in the fraud or failed to detect the fraud because they did not follow GAAS, the SEC and most market participants will likely perceive the situation as market failure. As a result, the SEC may try to set an example and increase the amount of civil penalties they seek. In this situation, the coefficient for AUDVIO would be positive. An audit violation would also likely increase disgorgement amounts imposed because when more perpetrators are involved, it is likely that ill-gotten profits increase as well. Furthermore, an audit violation may consist of the auditors helping management conceal fraud, which increases the duration and opportunity fraud perpetrators have to obtain ill-gotten profits. I have no prediction for the direction of the sign for the coefficient for DG (disgorgement) in the model to test H4 and the sign o f the coefficient for CP tested in the model used to test H5. A positive relationship would indicate that the fraud perpetrators are imposed larger amounts of disgorgement if civil penalties are greater, and similarly, if civil penalties increase, then disgorgement will increase. Table 3.3 presents a table o f predictions for the coefficients o f variables used to test H4 and H5: R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 60 TABLE 3.3 SUMMARY OF PREDICATIONS AND VARIABLE TYPE FOR H4 & H5 Hypothesis 4 Hypothesis 5 Dichotonomous Variable VD Sign o f coefficient ? No FS + No + Yes FOF + Yes TIP + No TIP + No AUDVIO + Yes AUDVIO + Yes DG ? No CP ? No Dichotonomous Variable Variable VD Sign of coefficient ? Yes FS + FOF Variable Yes Sample and Data Collection The SEC summarizes both fraudulent and nonfraudulent accounting related enforcement actions in Accounting and Audit Enforcement Releases (AAERs). Only companies with settled enforcement actions for fraud were tested for this study. To identify these companies, AAERs were searched using key words “fraud”, “irregularity” and “settle”. For an observation to be included in the sample used for this study, the following three criteria must have been met: (1) the exact initial disclosure date was determinable, (2) stock price data was available during the eleven day window being investigated in the Center for Research in Security Prices (CRSP) database, and (3) companies could not have two or more consecutive enforcement actions for fraud during R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 61 a five year period.8 All the SEC actions taken against the sample companies were during the period from 1997 to 2004. The following four sources were examined to identify a company’s initial disclosure: ■ Wall Street Journal articles ■ Company press releases ■ SEC filings (8-Ks, 10-Ks, and proxy filings) ■ AAERs All of the companies included in the final sample were classified as either voluntary disclosing or involuntary disclosing companies. A company was considered a voluntary discloser if it disclosed its fraud before the SEC initiated a formal investigation and the fraud was not found or disclosed by an external party. Table 3.4 illustrates the derivation o f the final sample used to test this study’s hypotheses. Table 3.5 presents information regarding the source o f the initial fraud disclosure for the sample companies. Most of the initial fraud disclosures for the voluntary and involuntary disclosing companies were made in press releases. None o f the voluntary disclosing companies made their initial disclosure in any SEC document, such as filings or AAERs.9 There are 11 involuntary disclosing companies that had their initial fraud disclosed in AAERs. This meant that that it was actually the SEC that made the 8 When a company has had consecutive frauds, the fact that one fraud soon follows another is likely, in itself, going to make both market participants and the SEC react more adversely than if this is a first time fraud for a company. 9 Obviously, no initial fraud disclosure is made in AAERs by voluntary disclosing companies because part o f the criteria used to classify a company as voluntary disclosing is that the company disclose the fraud prior to it being informed the SEC will be formally investigating it. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 62 disclosure, not the companies. Furthermore, it is likely that more detail o f these companies’ fraud were reported in their disclosures relative to the other companies’ initial disclosures. This is because by the time AAERs are disclosed to the public, the SEC investigation is likely to be at least mostly, if not fully completed. Hence, more information is available at this point to report in the AAERs. TABLE 3.4 DERIVATION OF FINAL SAMPLE Panel A: Derivation of final sample Companies with enforcement actions for fraud 359 Companies with indeterminate initial disclosure dates (122) Companies missing stock price data (138) Companies with two or more enforcement actions during a five year period (6) Companies included in the final sample 22 Panel B: Classification of companies as voluntary or involuntary disclosing Companies with initial disclosure after the SEC began formal investigation10 19 Companies with fraud found or disclosed by an external party 37 Companies that fell in both of the above two categories11 16 Companies classified as involuntary disclosing 22 Companies classified as voluntary disclosing 21 10 Because companies frequently do not disclose exact dates the SEC begins a formal investigation, this number may be understated. 11 Observations included in this amount are not included in the preceding two amounts. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 63 TABLE 3.5 SOURCE OF INITIAL FRAUD DISCLOSURE Wall street journal Press releases SEC filings AAER Total Voluntary disclosers 5 16 0 0 21 Involuntary disclosers 9 48 4 11 72 Total Sample 14 64 4 11 93 H2 and H3 examine the duration of the investigation related to the company and fraud perpetrators, respectively. H2 seeks to provide evidence that the duration o f the media exposure period is less for companies that voluntarily disclose versus involuntarily disclose fraud. To test this theory, the number of days in the media exposure period was calculated for each company. The beginning date for the media exposure period is the date the company initially discloses fraud. The ending date for the media exposure period is the date the company fully settled with the SEC, which can be obtained from 12 AAERs . H3 seeks to provide evidence regarding the duration o f the lingering litigation period, i.e. it should not be less for voluntary disclosing companies than for involuntary disclosing companies. Thus, the number of days will be obtained for the period beginning the date the company settles its case with the SEC to the date the last fraud perpetrator(s) settles their case with the SEC. Some frauds included in the final sample do not have a media exposure period and/or a lingering litigation period. Table 3.6 illustrates the frequency of media exposure periods and lingering litigation period for the companies include in this study’s sample. 12 The term fully settled is used because occasionally the SEC will partially settle cases. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 64 TABLE 3.6 COMPANIES WITH OR WITHOUT MEDIA EXPOSURE PERIODS AND/OR LINGERING LITIGATION PERIODS Voluntary disclosing companies: Number % Have only a media exposure period 6 28% Have only a lingering litigation period 5 24% 10 48% 0 0% Have both a media exposure period and lingering litigation period Have neither a media exposure period and lingering litigation period Total voluntary disclosing companies 21 Involuntary disclosing companies: Number % Have only a media exposure period 24 34% Have only a lingering litigation period 20 28% Have both a media exposure period and lingering litigation period 22 31% 6 7% Have neither a media exposure period or a lingering litigation period Total involuntary disclosing companies 22 Companies having only a media exposure period can arise from one o f two reasons. One reason is that both the company and fraud perpetrators may settle with the SEC on the same day. Another reason is that the SEC took action only against the company and not any of the fraud perpetrators. However, because companies cannot perpetrate its own fraud, it is rare that the SEC takes action against only the company. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 65 Companies having only a lingering litigation period can arise from one of two situations: (1) the SEC targets only the fraud perpetrators and not the company, or (2) the company’s settlement is disclosed in the initial fraud disclosure and this is followed by at least one fraud perpetrator settlement. A company having neither a media exposure period nor a lingering litigation period can only result when a company’s initial fraud disclosure, company settlement (if applicable), and all fraud perpetrator settlement(s) are disclosed in the same AAER. The reason why the voluntary disclosing sample does not have any cases that fall under this category is because a company’s voluntary disclosure o f fraud can never be made in an AAER, only in the Wall Street Journal, company press release, or SEC filing. To test H4 and H5, civil penalty and disgorgement amounts are needed. Civil penalty and disgorgement amounts were obtained from settlement AAERs.13 Sometimes companies are imposed prejudgement interest related to either civil penalties or disgorgement. Prejudgement interest is added to either civil penalties or disgorgement amounts, depending upon to which it relates to.14 13 The initial complaint AAERs sometimes report the amount o f civil, penalty and disgorgement the SEC is seeking while the settlement AAERs report the amounts imposed. Only the amounts o f civil penalty and disgorgement amounts imposed were used. 14 The primary reason the prejudgement interest is included with civil penalty or disgorgement is because sometimes AAERs add the prejudgement interest to one or both o f the amounts and it is impossible to separate. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. CHAPTER IV RESULTS Introduction This chapter is organized into two sections. The first section discusses descriptive information about the sample companies, perpetrators, and type o f frauds committed. The second section presents the results o f the five research questions investigated in this study. Company Descriptive Information Table 4.1 presents characteristics of the companies classified as voluntary and involuntary disclosers that were included in the final sample. The total number of companies included in the final sample is 93. O f these 93 companies, 21 companies were classified as voluntary disclosers and the remaining 72 companies were classified as involuntary disclosers. The two most significant variations between the voluntary and involuntary disclosing companies reported in Table 4.1 are the companies’ size (average assets) and average stockholders’ equity. The average size for voluntary disclosing companies is $1.3 billion, whereas the average size for involuntary disclosers is $19.5 billion. The average stockholders’ equity for the voluntary and involuntary disclosers is $321 million and $2,085 million, respectively. The involuntary disclosing companies have much 66 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 67 larger asset size and average stockholders’ equity than the voluntary disclosing companies. The average fraud amount is fairly close. The duration of the fraud for both the voluntary and involuntary disclosing companies is approximately two years for both samples, which is consistent with evidence found by prior research (ACFE, 2004; Bonner and Palmrose, 1998; Calderon and Green, 1998/1999). Regarding the exchange traded on, more of the involuntary disclosing companies, as opposed to the voluntary disclosing companies, traded on the NASDAQ as compared to the NYSE. However, the difference appears relatively proportional to their numbers. Very few o f either category traded on the OTC exchange. Table 4.1 also reports that 19 of the 21 voluntary disclosing companies are audited by a Big X audit firm, while 65 o f the 72 involuntary disclosing companies were audited by Big X firms. Only 9 of the 93 companies were audited by small or medium size audit firms. The industry classification between the voluntary and involuntary companies was widely spread, with a concentration in manufacturing for the voluntary disclosing companies, and a concentration in manufacturing and technology for the involuntary disclosing companies. The finding that there is such a high concentration in manufacturing in both samples is consistent with the results o f the ACFE’s Report to the Nation on Occupational Fraud and Abuse (2004), where it was found that the manufacturing industry has the highest industry of occupational frauds among those industries included in the ACFE study. The fraud types were also widely spread. The two most common types of frauds R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 68 TABLE 4.1 DESCRIPTIVE CHARACTERISTICS OF SAMPLED COMPANIES Company characteristics A verage total assets (in millions) Voluntary disclosing companies (N =21) $1,327 Average total revenue (in millions) Involuntary disclosing companies (N=72) $19,515 Totals Averages for Sample (N =93) $15,577 Significance o f difference between sample means t-values 4.780*** D-values .000 1,127 6,093 4,983 -1.079 .284 Average total equity (in millions) 321 2,085 1,649 9.755*** .000 Average fraud amount (in millions) 193 230 221 1.281 .203 Fraud duration (mean years) 2.3 1.9 2.2 .188 .851 Exchanee traded on at time o f initial disclosure N A SD A Q companies N Y S E companies OTC companies Total companies 39 29 4 72 49 39 5 93 Size o f comDanv’s audit firm at time o f fraud detection 19 Big X audit firm 2 Small or medium audit firm 21 Total companies 65 7 72 84 _9 93 Industry classification o f comoanv Technology Financial services Manufacturing Communications Services Retailer/Wholesaler Healthcare Other Total companies 3 1 6 1 4 3 3 0 21 19 6 15 6 5 5 11 _5 72 22 7 21 7 9 8 14 _5 93 12 8 12 37 22 25 49 30 37 10 17 5 7 24 24 Fraud TvDes Fictitious revenues Tim ing differences Improper asset valuation Understatement o f liabilities and/or expenses Improper disclosures Other 10 10 1 21 27 29 31 * Statistically significant at the . 10 level ** Statistically significant at the .05 level *** Statistically significant at the .01 level R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 69 for both the voluntary disclosing and involuntary disclosing companies are fictitious revenues and improper asset valuation. Results for the Market’s Reaction Testing Table 4.2 presents average abnormal returns for the period {-5 to 5}, with {0} being the date of the initial fraud disclosure. The results indicate that the voluntary disclosing companies’ cumulative abnormal returns decreased by 18.5% on days {-1, 0}, whereas the involuntary disclosing companies’ cumulative abnormal returns declined only 11.5%. The fact that both groups of companies experienced significant declines in abnormal returns indicates that the market reacts adversely when fraud is initially disclosed. This finding extends those by Feroz et al. (1991); they found a significant market reaction to both fraudulent and nonfraudulent SEC enforcement actions.1 Table 4.2 shows that there is a significant difference at the .10 level between the means o f abnormal returns of the voluntary and involuntary disclosing samples on the day the fraud is initially disclosed and that the returns of the voluntary disclosing sample are less than the returns o f the involuntary disclosing sample. This indicates that the market reacts more adversely to companies that voluntarily disclose fraud, as opposed to companies that involuntarily disclose fraud, upon the initial disclosure o f fraud, which contradicts the original prediction that voluntary disclosing companies will experience less o f a decline in market value than companies that involuntarily disclose fraud. There 1 O f the 58 companies used by Feroz et al. (1991) to test the market’s reaction to SEC enforcement actions, 35 o f the actions were for fraud. They also found that whether or not the enforcement actions were for fraud was not a significant cause of the market’s reaction. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 70 was no significant market differences between the two groups on the other dates examined. TABLE 4.2 AVERAGE ABNORMAL RETURNS AROUND THE INITIAL DISCLOSURE OF FRAUD l © oo Voluntary Disclosers Involuntary Disclosers N = 21 N = 72 Day Mean /-value Mean /-value -5 .002 -.003 -0.71 0.75 -4 .004 -.006 -1.67** 1.51* -3 -.002 -6.80*** -0.45 -2 -.008 -.009 -2.53*** -3.40*** -1 -.054 -.043 -7.84*** -5.74*** 0 -.131 -.072 -5.53*** -5.03*** 1 .011 -.025 1.50 -7.70*** 2 -.029 -20.30*** .017 6.42*** 3 -.009 -3.07*** .000 .00 4 -.003 -.003 -0.70 -1.13 5 -.013 -.014 -6.30*** -5.29*** * Statistically significant at the . 10 level ** Statistically significant at the .05 level *** Statistically significant at the .01 level Statistical Difference Between Two Means /-value p-value -.177 .863 -.354 .731 .566 .584 .035 .973 -.389 .705 -2.090* .063 1.270 .233 -1.630 .134 -.319 .756 .000 1.000 .035 .972 Both the voluntary and involuntary disclosing groups experience significant negative average abnormal returns at least three o f the four days preceding the initial disclosure of fraud. This finding corroborates the findings o f Elayan et al. (2002) where it was found that information regarding their sampled companies’ frauds was leaked out up to four days prior to the initial fraud announcement. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 71 Table 4.3 presents the results for the full model regression used to test hypothesis 1. The regression model used is shown below. CARj = b0 + b,VD + b2CO_SIZE+ b3FS + b4FD + b5TERM + bgSOURCE + b7SECINV + b8SUIT + bgAAUD + b10AMGMT + bnAUDVIO + b12FOF + bnREG + b 14BIG_ X + e Where, CAR! = Cumulative abnormal returns for days {-5, -4, -3, -2, -1, 0, 1,2, 3, 4, 5}, where 0 equals the date fraud is initially disclosed. VD = 1 if a company disclosed fraud before it was notified that it was being investigated by the SEC and the fraud was not announced by an external party prior to the company’s announcement, 0 if otherwise. CO_SIZE = The log o f total assets at the end o f the last year o f the fraud.2 FS = 1 if the fraud amount is disclosed during the initial fraud disclosure window, 0 if otherwise.3 FD = 1 if the fraud duration is disclosed during the initial fraud disclosure window, 0 if otherwise. TERM = 1 if the term ‘fraud’ or ‘irregularity’ was used to describe the act reported during the initial fraud disclosure, 0 if otherwise.4 SOURCE = 1 if the fraud was initially reported in an AAER, 0 if otherwise. SECINV = 1 if a SEC investigation was pending during the initial fraud disclosure window, 0 if otherwise. SUIT = 1 if a fraud-related lawsuit was pending during the initial fraud disclosure window, 0 if otherwise. 1 if an independent auditor resignation or termination was reported during the initial fraud disclosure window, 0 if otherwise. AAUD = AMGMT = 1 if the initial fraud disclosure reports that the company president, chairman o f the board, CFO, or CEO is fired, retires, or takes a leave o f absence, 0 if otherwise. AUDVIO = 1 if an audit violation o f audit standards was reported during the initial fraud disclosure window, 0 if otherwise. 2 If the total asset amount was not available at the end o f the last year o f the fraud, the last reported total asset amount was used as long as it was reported sometime during the fraud. 3 The fraud amount scaled by total stockholders’ equity would have been used here, but a majority o f the initial fraud disclosures did not disclose the amount o f the fraud. 4 To be included in the sample, the act must ultimately be concluded as fraud. However, the initial disclosures do not have to state the terms ‘fraud’ or ‘irregularity’. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 72 FOF = 1if the fraud involved the recognition o f fictitious revenues, 0 if otherwise. REG = 1if a company is in the financial services, insurance, or utilities industry, 0 if otherwise. BIG_ X = 1if the company’s independent auditors were a large audit firm when the fraud is d etected , 0 i f o th e r w ise. Voluntary disclosing companies do not experience less o f a market value decline when fraud is initially disclosed as opposed to involuntary disclosing companies. As indicated by the results in Table 4.3, the significance of the VD variable is .917. The size of the company (CO_SIZE) and the size (FS) and duration (FD) o f the fraud are all insignificant factors associated with the decline in market value a company experiences when they disclose fraud. This finding contradicts with prior evidence found by Feroz et al. (1991), which found that the income effect is a significant factor that adversely effects the market’s reaction to unacceptable accounting practices. The coefficients for AMGMT and SUIT are positive and significant at the .10 level. The fact that the sign of the coefficient for AMGMT is positive indicates that when a company reports the termination or retirement o f at least one member o f management, market participants likely perceive this as good news when this news is simultaneously disclosed with the initial fraud announcement. The positive coefficient found for SUIT contradicts with prior evidence found by Hedge et al. (2003), which found that market participants react adversely when class action lawsuits are announced after a company discloses fraud. The coefficient for SOURCE is negative and significant at the .10 level. This finding indicates that market participants will react more adversely when a company’s R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. TABLE 4.3 CROSS-SECTIONAL ASSOCIATION BETWEEN CUMULATIVE ABNORMAL RETURNS AND INDEPENDENT VARIABLES (N = 93) V ariable and E xpected Sign C oefficien t f-value p -value VD + .012 .105 .917 C O S IZ E - -.111 -.846 .400 FS - .097 .839 .404 FD - .028 .237 .814 TER M - -.032 -.275 .784 SO U R C E - -.219 -1.666* .100 SE C IN V - .029 .216 .829 SUIT - .201 1.763* .082 AAUD - .050 .440 .661 AM GM T + .195 1.713* .091 A U D V IO - -.124 -1.087 .280 FOF - .023 .190 .849 REG - .162 1.316 .192 BIGX + -.064 -.557 .579 * Statistically significant at the . 10 level ** Statistically significant at the .05 level *** Statistically significant at the .01 level Regression Model Statistics p-value R1 F-value .421 1.043 .158 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 74 fraud is initially disclosed in an AAER, as opposed to other media sources. This is likely because AAERs provide more detailed information regarding the fraud and fraud perpetrators. The remaining variables in the model - TERM, SECINV, AAUD, AUDVIO, FOF, REG, and BIG X - are all insignificant. A likely reason why some o f these variables are insignificant is because these items may not be disclosed, or fully disclosed, until sometime after the SEC investigation begins. For example, an initial fraud announcement may disclose only a small portion of a company’s fraud, with the remaining portion having not been detected and reported yet until the investigation is performed. The significance of the three variables - SOURCE, SUIT, and AMGMT - may be questionable because the model was insignificant with a F-value o f 1.043 (p = .421). To ensure that the SOURCE, SUIT, and AMGMT were actually significant and the signs on the coefficients were accurate, each variable was run separately against cumulative abnormal returns. The results are presented in APPENDIX A. Results for the Duration of the Companies’ Investigation Period Table 4.4 presents the results pertaining to the effect voluntary disclosing fraud has on the duration of the media exposure period - the period beginning the date the fraud is initially disclosed and ending the date the company settles its case with the SEC. This is the period of the SEC’s investigation that is primarily focused upon the company, rather than the fraud perpetrator(s). The model used to test this relationship is below: R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 75 MEPi = bo + b|VD + b2CO_SIZE + b3FS + b4REG + b5FD + b6AAUD + b7AUDVIO + b8BIG_X + e Where, MEP = The number o f days in the period beginning the date fraud is initially disclosed and ending the date the company settles its case with the SEC. VD = 1 if a company disclosed fraud before it was notified that it was being formally investigated by the SEC and it was not announced by ah external party prior to the company’s announcement, 0 if otherwise. CO_SIZE = The log o f total assets at the end o f the last year o f the fraud. FS = Total dollar amount o f the fraud scaled by the company’s total stockholders’ equity at the end o f the last year o f the fraud. REG = 1 if a company is in the financial services, insurance, or utilities industry, 0 if otherwise. FD = The number o f days o f the fraud duration period. AAUD = 1 if an audit firm resigned or was terminated because o f the fraud, 0 if otherwise. AUDVIO = 1 if it is announced that the company’s independent auditors were involved in the fraud or violated audit standards, 0 if otherwise. BIG X = 1 if the company’s independent auditors are a large audit firm, 0 if otherwise. Voluntary disclosing companies do not experience a shorter duration during their SEC investigation than the involuntary disclosing companies do. The results in Table 4.4 show that the the F-value for the model is 2.272 (p = .030) and the coefficient for VD is .892. This finding contradicts with the original prediction that the voluntarily disclosing companies will benefit by having shorter investigations because the SEC will be less reluctant to ask management to participate in the investigation than it would management of involuntary disclosing companies. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 76 The coefficient for FS is positive and significant at the .01 level which suggests a direct relationship exists between the size of a company’s fraud and the duration o f the company’s investigation. The coefficient for FD is also positive and significant at the .01 level. This suggests that the longer a company’s fraud is concealed, the longer the company’s investigation will be. Frauds that have continued for a long period o f time are likely to have been well concealed and, therefore, staff investigators need more time to identify evidence needed to win the SEC’s lawsuit against the company. The coefficient for AAUD is positive and significant at the .05 level. This may indicate that when an auditor resigns or is terminated, the fraud is more serious, increasing the length of the SEC investigation. Another alternative explanation is that the auditor may serve as a valuable resource to the staff investigators when they perform an investigation of a company. Thus, when the auditors are no longer readily available to assist staff investigators, the investigation takes longer. All other variables in the model CO-SIZE, REG, AUDYIO, and BIG X - are all insignificant. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 77 TABLE 4.4 CROSS-SECTIONAL ASSOCIATION BETWEEN THE DURATION OF THE MEDIA EXPOSURE PERIOD AND FRAUD DISCLOSURE CHOICE (N = 93) Variable and Expected Sign Coefficient 1-value p-value VD - -.012 -.136 .892 COSIZE + .082 .717 .475 FS + .274 2 7 7 3 *** .007 REG ? -.068 -.668 .506 FD + .434 4 500*** .000 AAUD + .184 2 .0 1 1 ** AUDVIO + -.054 -.539 .592 BIGX - .081 .871 .386 .048 * Statistica ly signi: leant at the .10 level ** Statistica ly signil leant at the .05 level *** Statistica ly signil leant at the .01 level Regression Model S1tatistics F-value p-value R1 2.272 .030 .178 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 78 Results for the Duration of the Fraud Perpetrators’ Investigation Period Table 4.5 presents the results o f the effect voluntary disclosing fraud has on the duration of the lingering litigation period, where the focus o f the investigation is primarily upon the fraud perpetrators. The model used to test this relationship is below: LLPj = bo + b, VD + b 2CO_SIZE + b3FS + b4REG + b5FD + b6AAUD + byAUDVIO + b 8BIG_X + b9MEP + e Where, LLP = The number o f days in the period beginning the date the company settles its case with the SEC and ending the date the last perpetrator’s case is settled with the SEC. VD = 1 if a company disclosed fraud before it was notified that it was begin formally investigated by the SEC and it was not announced by an external party prior to the company’s announcement, 0 if otherwise. CO SIZE = The log o f total assets at the end o f the last year o f the fraud. FS = Total dollar amount of the fraud scaled by the company’s total stockholders’ equity at the end o f the last year o f the fraud. REG = 1 if a company is in the financial services, insurance, or utilities industry, 0 if otherwise. FD = The number o f days of the fraud duration period. AAUD = 1 if an audit firm resigned or was terminated because o f the fraud, 0 if otherwise. AUDVIO = 1 if it is announced that the company’s independent auditors were involved in the fraud or violated audit standards, 0 if otherwise. BIGX = 1 if the company’s independent auditors are a large audit firm, 0 if otherwise. MEP = The number o f days in the period beginning the date fraud is initially disclosed and ending the date the company settles its case with the SEC. The duration of the investigation related to the fraud perpetrators o f voluntary disclosing companies is not affected by the companies’ choice to voluntarily disclose R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 79 fraud. The results in Table 4.5 show that the F-value for the model is 3.703 (p = .003) and the coefficient for VD is insignificant. However, because no relationship was found between voluntary disclosure o f fraud and the media exposure period, no conclusion can be drawn regarding whether or not fraud perpetrators o f voluntary disclosing companies have shorter investigations than involuntary disclosing companies do. There are three factors which were found to affect the duration o f the fraud perpetrators’ investigation period. The coefficient for FS is positive and significant at the.05 level. This may suggest that staff investigators need more time to gather evidence related to the fraud perpetrators actions when the size of fraud is large. The coefficient for FD is positive and significant at the .01 level. This may suggest that the longer a company’s fraud was concealed, the more time staff investigators need to obtain sufficient evidence to prevail in the SEC’s case against fraud perpetrators. The coefficient for MEP is negative and significant at the .01 level which suggests that the longer the company’s investigation is, the shorter the investigation period for fraud perpetrator(s)’ will likely be. This finding could mean a couple o f things. First, because the enforcement division had limited resources at the time these frauds were investigated, it is likely that the staff investigators could spend limited time on the entire investigation. Second, staff investigators likely gather large amounts o f evidence against the fraud perpetrators during the investigation that is mostly focused on the company. Therefore, the longer the media exposure period is, the less amount o f evidence staff investigators will have to gather during the lingering litigation period. Finally, the coefficient for REG is negative and significant at the .10 level. This R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 80 TABLE 4.5 CROSS-SECTIONAL ASSOCIATION BETWEEN THE DURATION OF THE LINGERING LITIGATION PERIOD AND FRAUD DISCLOSURE CHOICE (N = 93) Variable and Expected Sign Coefficient /-value p-value VD + .003 .032 .975 COSIZE + .116 1.290 .201 FS + .207 2.231** .028 REG ? -.178 FD + .524 AAUD + .022 .230 .819 AUDVIO + -.054 -.494 .622 - .106 1.299 .198 BIGX MEP — -.298 -1.941* 5 7 7 7 *** -3.659*** .056 .000 .000 * Statistica ly signil leant at the .10 level ** Statistica ly signil leant at the .05 level *** Statistica ly signil leant at the .01 level Regression Model Statistics p-value R^ F-value 3.703 .003 .205 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 81 finding suggests that the more regulated a company is, the shorter the duration of the fraud perpetrator(s) investigation is. This may be because if two or more agencies regulate a company, the work of one agency may be utilized by another agency. The coefficients for CO SIZF, FD, REG, AAUD, AUDYIO, and B I G X are insignificant. Results for the Civil Penalty and Disgorgement Amounts Table 4.6 presents the results o f the effect voluntarily disclosing fraud has on civil penalty amounts imposed upon fraud perpetrators. The model used to test this relationship is shown below: CP; = b0 + bjVD + b2FS +b3FOF + b4TIP + bsAUDVIO + b6DG + e Where, CP = The dollar amount o f civil penalties imposed upon fraud perpetrators VD = 1 if a company disclosed fraud before it was notified that it was begin formally investigated by the SEC and it was not announced by an external party prior to the company’s announcement, 0 if otherwise. FS = Total dollar amount o f the fraud scaled by the company’s total stockholders’ equity at the end o f the last year o f the fraud. FOF = 1 if the fraud involved the recognition o f fictitious revenues, 0 if otherwise. TIP = The sum o f the media exposure period and the lingering litigation period. AUDVIO = 1 if it is announced that the company’s independent auditors were involved in the fraud or violated audit standards, 0 if otherwise. DG = The dollar amount o f disgorgement fraud perpetrators) must pay upon their settlement with the SEC. As expected, the coefficient for VD is insignificant, which indicates no relationship exists between the choice to voluntarily disclose fraud and the amount o f R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 82 civil penalties imposed upon fraud perpetrators. The F-value for the model is 4.398 (p = .001). The coefficient for AUDVIO is positive and significant at the .10 level when civil penalties imposed upon auditors were excluded from the total civil penalties imposed upon fraud perpetrators .5 While 13 of the 92 observations sampled had auditor violations, in only two of these situations were civil penalties imposed upon auditors. However, the amount o f civil penalties imposed were significantly larger than civil penalties imposed upon non-auditor fraud perpetrators. During this period o f time, the findings indicate that the SEC seldom seeks civil penalties from auditors, but when they do take this action, they seek an amount significantly greater than amounts sought from non-auditor fraud perpetrators. The coefficient for DG is positive and significant at the .01 level. This finding suggests that the SEC seeks civil penalties from fraud perpetrators if they are also seeking disgorgement. In fact, the results indicate that civil penalties and disgorgement amounts have a direct relationship, meaning the more the SEC seeks in disgorgement, the more it will seek in civil penalties as well. All other variables tested in H4 - FS, FOF, and TIP are insignificant. 5 The coefficient for AUDVIO is positive and significant when the civil penalty amounts imposed upon fraud perpetrators excluded civil penalty amounts imposed upon the only two audit firms in both the voluntary and involuntary disclosing samples. The two companies where auditors were imposed civil penalties were Warnaco and Waste Management. Wamaco’s auditors were PricewaterhouseCoopers and were required to pay a $2.4 million civil penalty. Waste Management’s auditors were Arthur Andersen and were required to pay a $7 million penalty. When the civil penalty amounts imposed upon audit firms were included in the civil penalty amounts, the coefficient for AUDVIO is still positive, but significant at the .05 level rather than the . 10 level. In the final sample o f companies, there were 56 fraud perpetrators that were imposed civil penalties and the average imposition amount was $551,000. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 83 TABLE 4.6 CROSS-SECTIONAL ASSOCIATION BETWEEN CIVIL PENALTIES IMPOSED UPON FRAUD PERPETRATORS AND FRAUD DISCLOSURE CHOICE (N = 93) Variable and Expected Sign Coefficient f-value p-value VD ? -.047 -.494 .622 FS + -.171 -1.389 .169 FOF + -.012 -.112 .911 TIP + .031 .305 .761 AUDVIO + .212 1.917* .059 DG ? .454 4 3 5 7 *** .000 * Statistically significant at the .10 level ** Statistically significant at the .05 level *** Statistically significant at the .01 level Regression Model Statistics F-value p-value R2 4.398 .001 .235 Table 4.7 presents the results of the effect voluntarily disclosing fraud has on the disgorgement amount fraud perpetrator(s) must pay upon settlement with the SEC. The model used to test this relationship is shown below: DGi = b0 + b, VD + b2FS + b3FOF + b4TIP + b5AUDVIO + b6CP + e R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 84 Where, DG = The dollar amount o f disgorgement fraud perpetrator(s) must pay upon their settlement with the SEC. VD = 1 if a company disclosed fraud before it was notified that it was begin formally investigated by the SEC and it was not announced by an external party prior to the company’s announcement, 0 if otherwise. FS = Total dollar amount o f the fraud scaled by the company’s total stockholders’ equity at the end o f the last year o f the fraud. FOF = 1 if the fraud involved the recognition o f fictitious revenues, 0 if otherwise. TIP = The sum o f the media exposure period and the lingering litigation period. AUDVIO = 1 if it is announced that the company’s independent auditors were involved in the fraud or violated audit standards, 0 if otherwise. CP = The dollar amount o f civil penalties imposed upon fraud perpetrators There is no relationship between a company’s fraud disclosure choice and disgorgement imposed upon fraud perpetrators. The F-value for the model is 6.998 (p = .000). As predicted, the coefficient for VD is insignificant. The coefficient for FS is positive and significant at the .01 level. This finding is reasonably expected considering fraud perpetrators are likely to profit more from larger frauds than smaller frauds. The coefficient for CP is positive and significant at the .01 level. This finding suggests a positive relationship between CP and DG. As fraud perpetrators’ illegally obtained profits increase, so does the civil penalties and disgorgement amounts imposed upon them. The SEC likely increases civil penalty amounts to punish fraud perpetrators for the harm they caused others. All other variables tested in H5 - FOF, TIP, and AUDVIO are all insignificant. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. TABLE 4.7 CROSS-SECTIONAL ASSOCIATION BETWEEN DISGORGEMENT IMPOSED UPON FRAUD PERPETRATORS AND FRAUD DISCLOSURE CHOICE (N = 93) Variable and Expected Sign Coefficient f-value p-value VD ? -.001 -.013 .989 FS + .376 3 4 5 3 *** .001 FOF + .020 .190 .850 TIP + -.050 -.527 .599 AUDVIO + -.016 -.146 .884 CP ? .399 4.357*** .000 * Statistically significant at the . 0 level ** Statistically significant at the .C)5 level *** Statistically significant at the .C)1 level Regression Model Statistics F-value p-value Rz 6.998 .000 .328 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. CHAPTER V CONCLUSIONS This chapter provides an overall summary of the study and implications o f the results. The sections included in this chapter are: Summary and Implications, Limitations, and Future Research. Summary and Implications There is a vast amount o f literature dealing with accounting disclosure choices. While there has been some research dealing with the consequences o f disclosing fraud, there is little research investigating the effects of voluntarily disclosing fraud. Since the explosion of corporate frauds over the past several years, the SEC has been searching for ways to motivate companies to disclose fraud immediately upon detection. This study investigates whether a reduction in losses associated with fraud disclosure is an incentive for companies to voluntarily disclose fraud. The three fraud losses investigated were (1) the decline in company market value, (2 ) the duration of both the companies’ and fraud perpetrators’ investigation period, and (3) civil penalty and disgorgement amounts imposed on fraud perpetrators. The results confirm selected prior findings found by other researchers who investigated the market’s reaction to fraud disclosures. First, this study and Hedge et al. 86 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 87 (2003) both found that companies will experience a significant decline in market value upon the initial disclosure o f fraud. These results are extended by providing evidence that a significant decline in market value will occur whether or not companies voluntarily or involuntarily disclose their fraud. Furthermore, this study finds that companies will have more o f a decline in market value if the initial fraud disclosure is made in an AAER, rather than other media sources, and that disclosing the firing or retirement o f fraud perpetrators in the initial fraud disclosure may mitigate some o f the company’s decline in market value. This study also confirms evidence found by Elayan et al. (2002) that news of fraud is leaked out to market participants up to four days prior to the initial disclosure o f fraud and extends it by finding that this holds true for both voluntary and involuntary disclosing companies. This four-day leakage found in both voluntary and involuntary disclosing samples gives rise to the potential for future investigations dealing with management’s intentions behind voluntarily disclosing fraud. In addition to evaluating market losses, this study examined the impact of voluntary disclosing fraud on the SEC investigations related to the company and any lingering investigations related to the fraud perpetrators. The results indicate that neither the companies’ investigation nor the fraud perpetrators’ investigation period is reduced when fraud is voluntarily disclosed. These results, taken collectively with the results of the market reaction tests, indicate that companies have not benefited by having a significant reduction o f losses associated with voluntary disclosure. There are several possible explanations o f these results. One possibility, suggested by the leakage four days prior to the initial fraud R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 88 disclosure, is that voluntary disclosing companies are viewed as disclosing because o f the leakage rather than on a truly voluntary basis. Because o f the heavy losses involved in any disclosure o f fraud, no fraud disclosure is viewed as really voluntary. Another possibility is that neither market participants nor the SEC can distinguish whether fraud is voluntary or involuntary disclosed. There may be so much noise in the financial markets or a dirge of information such that at the time of disclosure whether the fraud is voluntarily or not is not clearly discemable. That may be more true o f the market than the SEC since the market measure is at a point in time whereas the SEC measure is over a period of time. Also, if anyone is able to know if a company is voluntarily disclosing it should be the SEC, yet their investigation period is the same for voluntary and involuntary disclosing companies. This leads to another possible explanation o f results. Perhaps the SEC takes the perspective that all fraud, whether voluntarily or involuntarily disclosed, is bad and must be punished and dealt with as swiftly and effectively as possible. If dealt with otherwise, the SEC may feel a strong enough example will not be set. The SEC may also distinguish between voluntary and involuntary disclosing companies in a way that were not measured in this study, i.e. other than the length o f the investigation. A public policy implication o f the findings is that the market and the SEC appears to be very punishment oriented when a company discloses fraud. While it is difficult to make any recommendations for the market, perhaps it may be useful for the SEC to use the carrot as well as the stick in encouraging voluntary behavior by companies in disclosing fraud by providing some kind o f positive incentives. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 89 There are several other important findings resulting from testing the duration o f the two investigation periods. First, the length o f a company’s fraud is positively associated with the duration of both the companies’ investigation, but not the fraud perpetrators’ investigation. This suggests that frauds that last longer will likely result in staff investigators needing more time to gather evidence during the companies’ investigation. A second important finding is that a company’s investigation will be longer if their auditor retires or is terminated by the company. This result is likely due to auditor changes being associated with more serious frauds. No relationship was found between a change in a company’s auditors and the length o f the investigation o f fraud perpetrators, i.e. the lingering litigation period. Another important finding is that there is a significant inverse relationship between the duration of the companies’ investigations and the duration of the fraud perpetrators’ investigations. This result may be explained in a couple o f ways. One, the staff investigators may obtain most of the evidence needed to prosecute fraud perpetrators during the company’s investigation. Evidence needed to prosecute fraud perpetrators may be gathered concurrently while performing the company’s investigation. Hence, staff investigators will need less time after the company’s investigation is completed. An alternative explanation is that the SEC may try to mitigate overspending during an investigation by accelerating the fraud perpetrators’ investigation if the company’s investigation took longer than expected. This conclusion would be consistent with the fact that during most of the sampled period, the SEC’s enforcement division had limited resources to perform their investigations. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 90 The imposition of civil penalties and disgorgement upon fraud perpetrators was examined to determine if fraud perpetrators o f voluntary disclosing companies are inappropriately imposed less civil penalties and disgorgement because the fraud concealment period is shorter than if their company involuntarily disclosed. As expected, no association was found between a company’s disclosure choice and the imposition of either civil penalty or disgorgement amounts. Several findings were identified when testing civil penalty and disgorgement imposed upon fraud perpetrators. First, audit violations are positively associated with civil penalty amounts, but no association exists between audit violations and disgorgement amounts. This finding may be attributable to the SEC wanting to set an example when auditors either participate or help conceal a company’s fraud. While only two audit firms paid civil penalties in the total sample examine in this study, the amounts imposed were much larger than civil penalty amounts imposed upon non-auditor fraud perpetrators. Furthermore, a positive association was found between civil penalty amounts and disgorgement amounts. This result may suggest the SEC will seek more from fraud perpetrators if they profited from the fraudulent funds. Another result found when testing disgorgement amounts is that the larger the fraud is, the more disgorgement the fraud perpetrators must pay. This finding is likely attributable to fraud perpetrators having more opportunity to reap greater benefit when the fraud is larger. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. Limitations AAERs were used as a proxy for fraud occurrences. The SEC normally targets companies based upon the severity of the fraud and whether it believes it will obtain sufficient evidence to prevail in a legal action against the company. Hence, any sample taken o f AAERs may not represent the entire population of fraud occurrences, but rather what currently is on the SEC’s agenda. Because it is fairly rare for companies to voluntarily disclose fraud, the voluntary disclosing sample was small relative to the involuntary disclosing sample. It would have been preferable to have two samples of approximately the same size, as well as having a larger overall sample. However, due to the unavailability o f data (i.e. stock prices, SEC filings, etc.) and the fact that most fraud announcements are not voluntarily made, obtaining equal sample sizes was not possible. Another limitation is that fraud size is inconsistently reported in initial disclosures, if it is reported at all. If the first source of disclosure is an AAER, the specific accounts and related amounts are likely disclosed, such as revenues or expenses. However, if the initial fraud disclosure is made in some other media source, information regarding fraud size may not be reported, or reported with less detail. This inconsistency in reported fraud sizes causes a difficulty in interpreting the degree o f association between cumulative abnormal returns and fraud size. Future Research Whether the Sarbanes-Oxley Act (2002) has increased the number o f R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 92 companies voluntarily disclosing fraud may be o f interest to future researchers. The Act has established harsher penalties associated with fraudulent reporting, which causes uncertainty regarding whether or not management now has an incentive or a disincentive to report fraud immediately upon detection. The increased responsibility associated with certifying their company’s financial statements may accelerate the detection o f fraud because CEOs and CFOs may choose to more thoroughly investigate the reliability o f their external reporting. Hence, this may give rise to more companies disclosing their fraud before it is informed of a SEC investigation or fraud is disclosed by a third party. Conversely, because o f harsher penalties which include extensive prison sentences, CEOs and CFOs may be more reluctant to disclose fraud in fear that they may be perceived as the ultimate bearers o f responsibility for the fraud and, consequently, may be the ones these larger penalties are imposed upon. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. REFERENCES Aboody, D., & Kasznik, R. (2000). CEO Stock Option Awards and the Timing of Corporate Voluntary Disclosures. Journal o f Accounting and Economics, 29 (3): 73-100. Albrecht, C.O., & Albrecht, S.W. (2004). Fraud Examination and Prevention. Mason, Ohio: South-Western. Ameen, E. & Guffey, D. (1993/1994). An Investigation o f the Effect o f Qualified Audit Opinions on the Trading Volume and Bid-Ask Spread o f Over-the-Counter Companies. Review o f Financial Economics, 3 (Fall/Spring): 41-51. Association o f Certified Fraud Examiners (ACFE). (2004). Report to the Nation on Occupational Fraud and Abuse. Austin, TX. Atiase, R.K. (1985). Predisclosure Information, Firm Capitalization and Security Price Behavior Around Earnings Announcements. Journal o f Accounting Research, 23 (1): 21-36. Barnes, P. & Lambell, J. (2003). Oranizational Susceptibility to Fraud: Does Fraud Strike Randomly or are There Organizational Factors Affecting Its Likelihood and Size? Working paper. Nottingham Trent University. Beneish, M. D. (1999). Incentives and Penalties Related to Earnings Overstatements that Violate GAAP. The Accounting Review, 1A {A): 425-457. Bonner, S. & Palmrose, Z. (1998). Fraud Type and Auditor Litigation: An Analysis o f SEC Accounting and Auditing Enforcement Releases. The Accounting Review, 73 (4): 503-532. Botosan, C. (1997). Disclosure Level and the Cost o f Equity Capital. The Accounting Review, 72 (3): 323-349. Calderon, T. & Green, B. (1998/1999). SEC Acounting and Auditing Enforcement Actions in the Banking Industry. Journal o f Applied Business Research, 15 (Winter): 69-83. 93 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 94 Cox, J. D., & Thomas, R.S. (2003). SEC Enforcement Actions for Financial Fraud and Private Litigation: An Empirical Inquiry. Vanderbilt Law and Economics Research Paper No. 03-08. Dechow, P. M. , Sloan, R., & Sweeney, A. (1992) An Examination of the Reasons for and Stock Market Reaction to Voluntary Disclosure o f Auditor Resignations. Working paper. University o f Southern California. __________ , Sloan, R. G., & Sweeney, A. P. (1996). Causes and Consequences of Earnings Manipulation: An Analysis o f Companies Subject to Enforcement Actions by the SEC. Contemporary Accounting Research, 13 (1): 1-36. Defond, M. & Smith, D. (1991). Discussion o f the Financial and Market Effects o f the SEC Accounting and Auditing Enforcement Releases. Journal o f Accounting Research, 3 (Supplement): 143-148. __________ , & Smith, D. (1992). An Examination o f the Reasons for and Stock Market Reaction to Voluntary Disclosure o f Auditor Resignations. Working paper. University o f Southern California. Dempsey, S. J. (1989). Predisclosure Information Search Incentives, Analyst Following, and Earnings Announcement Price Response. The Accounting Review, 64 (4): 748-757. Eberhart, A.C., Moore, W. T., & Roenfeldt, R. L. (1990). Security Pricing and Deviations from the Absolute Priority Rule in Bankruptcy Proceedings. The Journal o f Finance, Volume XLV (5): 1457-1469. Elayan, F. A., Li, J., & Meyer, T. O. (2002). Accounting Irregularities, Management Compensation Structure, and Information Asymmetry. Working paper. Massey University. Feroz, E. H., Park, K., & Pastena, V. S. (1991). The Financial and Market Effects o f the SEC’s Accounting and Auditing Enforcement Releases. Journal o f Accounting Research, 29 (3): 107-142. Hedge, S. P., Malone, C. B., & Finnerty, J. D. (2003). The Financial Impacts o f Fraud and Securities Class Action Suits. Working paper. University o f Connecticut. Lobo, G.J, & Mahmoud, A.W. (1989). Relationship Between Differential Amounts o f Prior Information and Security Return Variability. Journal o f Accounting Research, 21 (I): 116-134. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 95 Mahoney, C. P., Walker, C. W., Schwartz, E. T., Gambino, C. J., & Greenstein, L. D. (2002). The SEC Enforcement Process: Practice and Procedure in Handling an SEC Investigation. Washington, D.C. The Bureau o f National Affairs, Inc. Palmrose, Z., & Scholtz, S. (2000). Restated Financial Statements and Auditor Litigation. Working paper. University o f Southern California. __________ , Richardson, V. J., & Scholz, S. (2001). Determinants o f Market Reactions to Restatement Announcements. Working paper. University o f Southern California. Reason, Tim. (2005). The Limits o f Mercy. CFO, 21 (5): 61-68. Shores, D. (1990). The Association Between Interim Information and Security Returns Surrounding Earnings Announcements. Journal o f Accounting Research, 28 (1): 164-181. Shumway, T. (1997). The Delisting Bias in CRSP Data. The Journal o f Finance, 52 (March): 327- 340. Skinner, D. J. (1994). Why Firms Voluntarily Disclose Bad News. Journal o f Accounting Research, 32 (Spring): 38-60. Watts, R., & Zimmerman, J. (1986). Positive Accounting Theory. Englewood Cliffs, NJ: Prentice Hall. Wells, D. W. & Loudder, M. L. (1997). The Market Effects o f Auditor Resignations. Auditing: A Journal o f Practice and Theory, 16 (Spring): 138-144. Yermack, D. (1997) Good Timing: CEO Stock Option Awards and Company News Announcements. Journal o f Finance, 52 (2): 449-476. R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. APPENDIX A SUPPLEMENTARY TESTING OF MARKET’S REACTION TEST 96 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 97 APPENDIX A SUPPLEMENTARY TESTING OF MARKET’S REACTION TEST This section presents the results of supplementary testing where only the VD variable is included in the model as an independent variable while regressing it on the dependent variables used in all five hypotheses. This testing mitigates adverse effects associated with excessive usage o f independent variables in the models. Hypothesis 1 is tested using the following model: CARi = b0 + biVD + e, Where CAR is cumulative abnormal returns for the eleven day period {-5 -4, -3, -2, -1, 0, 1, 2, 3, 4, 5} and VD is a dichotonomous variable that equals one if a company discloses its fraud if it was notified that it was being investigated by the SEC and the fraud was not announced by an external party prior to the company’s announcement, 0 if otherwise. The results o f this test are similar to the results obtained when the original model was run. The coefficient for VD is insignificant. This finding, corroborated with results presented in Table 4.3, suggest that companies do not obtain a benefit o f a lesser decline in market value if they chose to voluntarily disclosing their fraud. The results o f this test are presented below: Variable and Expected Sign + VD Coefficient t Value Significance .038 .367 .714 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 98 As the results indicate below, the three variables that were significant in H I are also significant when run individually against cumulative abnormal returns. Therefore, the same conclusions that were stated previously in chapters four and five still apply. TABLE A. 1 UNIVARIATE REGRESSION ANALYSIS BETWEEN CUMULATIVE ABNORMAL RETURNS AND INDEPENDENT VARIABLES (N = 93) Variable Variable Statistics t-value p-value SOURCE -2.032 .045 SUIT 1.737 .086 AMGMT 1.743 .085 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. APPENDIX B CLASSIFICATION OF COMPANIES INCLUDED IN FINAL SAMPLE 99 R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 100 APPENDIX B CLASSIFICATION OF COMPANIES INCLUDED IN FINAL SAMPLES Companies Classified as Voluntary Disclosers ABS Industries Itex Amazon.com Material Science Corporation Paracelsus Healthcare Corporation Boston Scientific Comptronix Corporation Pepsi-Cola Photran Corporation Cylink Corporation Rent-Way First Merchants Acceptance Gateway, Inc. Sunrise Medical Gibson Greetings Thor Industries Guilford Mills Vari-L Waste Management Insignia Solutions International Research & Development Companies Classified as Involuntary Disclosers Fischer Imaging Corporation Advanced Technical Products Galileo Corporation American International Group (AIG) Aon Corp. Gunther International Aura Systems Healthsouth Aurora Foods Home Depot Baker Hughes, Inc. ICN Pharmaceuticals BI, Inc. Immunogen, Inc Indus International Brightpoint California Micro Distributors Informix International Business Machines (IBM) Cendant (CUC International) Cronos Group J.P. Morgan Cyberguard Kansas City Southern Dain Rauscher Kurzweil Applied Intelligence Digital Lightwave Lee Pharmaceuticals Lernout & Hauspie Donnkenny, Inc. Lotus Development Corp. DTE Energy Group Dynamic Healthcare Technologies McKesson HBOC Emulex MCN Energy Group Medaphis Corp. (Per Se Technologies) Engineering Animation Mercury Finance Escalade, Inc. Micro Warehouse Fabri-Centers of America MicroStrategy FastComm NationsBank (Bank America Corp.) Ferrofluidics R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. 101 Companies Classified as Involuntary Disclosers (Continued1) Northstar Health Services Scholastic Corp. Oak Industries Scorpion Technologies Peregrine Systems Sensormatic Electronics Peritus Software Signal Technologies Pier 1 Sony Corporation Premier Laser Spectrum Information Technologies Presstek Sunbeam Prime Capital Corp. Symbol Technologies Regency Health Services System Software Associates Rite-Aid Transcrypt International RSA Securities Warnaco Group Sabratek Xerox SafTLok Yes Entertainment R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission. [...]... also reduce the amount o f damages a suing party can claim If market participants’ thinking is consistent with evidence found by Skinner (1994) that managers have an incentive to disclose bad news voluntarily to reduce future litigation losses, then an inference can be made that market participants might perceive voluntary disclosure of fraud as a strategic maneuver to mitigate losses associated with. .. voluntarily disclose fraud, companies will have an incentive to immediately disclose fraud Consequently, shareholders will benefit because losses associated with their stock holdings will likely be reduced due to management choosing to disclose fraud immediately upon detection In addition, if companies are given an incentive to voluntarily disclose fraud, a broad R ep ro d u ced with p erm ission o... The only benefit significant enough to outweigh the cost o f disclosing fraud early is the reduction in company market value incurred because o f uncertainties involving the companies future prospects There are several factors associated with management’s choice o f voluntarily disclosing fraud versus involuntarily disclosing fraud One motivation management may have is to try to reduce the risk of future... imposed by the SEC for frauds with long duration periods and those perpetrated at the highest levels Voluntary Disclosure of Fraud Fraud is defined by Albrecht and Albrecht (2004) as: a generic term that embraces all the multifarious means which human ingenuity can devise, which are resorted to by one individual, to get an advantage over another by false representations No definite and invariable rule can. .. to experience a significant decline in market value when bankruptcy announcements are made (Eberhart et al., 1990) may be associated with bankruptcy for various reasons Fraud For example, fraud may be committed to conceal and/or enhance failing financial performance, while on other occasions the fraud may be so material and cause such significant losses to the company that the fraud itself forces the... capital markets, (2) voluntary disclosure, (3) corporate securities regulation, and (4) other fraud literature Capital Markets Literature An important purpose of the study is to investigate the market’s reaction to fraud to determine if the decline in market value a company incurs can be reduced by voluntarily disclosing its fraud It is challenging to measure the market’s reaction to fraud because o f... frequently occurring frauds are significantly associated with auditor litigation, in this study I will only test frequently occurring frauds because it is too difficult to determine if fraud was concealed by fictitious transactions in the initial fraud disclosure Fictitious revenues will serve as a proxy for frequently occurring frauds because it is the type o f fraud that has been found to be most vulnerable... 1 Initial fraud disclosure SEC informal investigation SEC formal investigation I Company settles with SEC J I Final settlement between SEC and perpetrators Figure 1: The Corporate Fraud Process In addition, this study tests the association between companies forthcomingness with fraud and the amount of civil penalties and disgorgement imposed upon fraud perpetrators by the SEC If the SEC... corporate fraud by establishing the Public Company Oversight Board (PCAOB) to help monitor management activities, investigate fraud, impose sanctions upon fraud perpetrators, and make Chief Financial Officers (CFOs) and Chief Executive Officers (CEOs) more costs from companies and their auditors, it will be several years before it can be determined if the Act has helped reduce corporate fraud Therefore,... mitigate corporate fraud and concealment If it can be determined that companies that voluntarily disclose fraud experience less of a market value decline than involuntary disclosers, an incentive may be created to motivate managers to disclose fraud immediately upon detection Unfortunately, while the SEC has been informing publicly-held companies during the past several years that they can avoid civil