An examination of anonymous and non-anonymous fraud reporting channels

Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

Contents lists available at SciVerse ScienceDirect

Advances in Accounting, incorporating Advances in International Accounting journal homepage: www.elsevier.com/locate/adiac

An examination of anonymous and non-anonymous fraud reporting channels Steven E. Kaplan a,⁎, Kurt Pany a, Janet Samuels b, Jian Zhang c a b c

Arizona State University, United States Thunderbird School of Global Management, United States San Jose State University, United States

a r t i c l e

i n f o

Keywords: Whistleblowing Anonymous reporting channels Non-anonymous reporting channels Retaliation Repercussions Fraudulent financial reporting

a b s t r a c t Interest in the role of the reporting channel on whistleblowing has been fostered by the passage of the Sarbanes-Oxley Act of 2002, which requires that audit committees of public companies establish and oversee an anonymous reporting channel for questionable accounting or auditing matters. But only limited information exists as to the likely effectiveness of such a channel as compared to a non-anonymous channel. The purpose of our paper is to report the results of an experimental study examining participants' intentions to report fraud using anonymous and non-anonymous reporting channels given information about the outcomes from a previous non-anonymous whistleblowing incident. The experiment manipulates the outcomes to both the previous whistleblower and to the transgressor. We find that while negative outcomes from the perspective of a previous non-anonymous whistleblower (either the occurrence of retaliation against that person or no negative repercussions to the previous transgressor) lowered participants' non-anonymous reporting intentions, these negative outcomes did not decrease participants' anonymous reporting intentions. But when, no such negative outcomes from the previous whistleblower's perspective have occurred, our participants' reporting intentions did not differ between the anonymous and non-anonymous channels. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction In response to a series of high-profile financial statement frauds, Congress passed the Sarbanes-Oxley Act of 2002, which, in part, made several changes directed towards strengthening public companies' internal controls and expanding the responsibilities of audit committees. Under Section 301, audit committees of public companies are required to establish and oversee a reporting channel for anonymous employee reporting of questionable accounting or auditing matters. Presumably, requiring an anonymous reporting channel strengthens internal control related to fraud detection by ensuring that employees have access to a reporting channel with relatively low personal costs. But establishing and operating an anonymous reporting channel is costly to public companies. Ordinarily, such a reporting channel is not intended to and need not displace other reporting channels, such as reporting non-anonymously and directly to a superior. Indeed, regarding reporting channels, best practices include the availability of multiple channels for employees to report their concerns about illegal and unethical behavior (e.g., ACFE 2005; The Network, Inc., 2006). Thus, even when an audit committee for a public company has established an anonymous reporting channel, informal reporting channels will generally be available which allow employees to non⁎ Corresponding author. E-mail addresses: [email protected] (S.E. Kaplan), [email protected] (K. Pany), [email protected] (J. Samuels), [email protected] (J. Zhang). 0882-6110/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.adiac.2012.02.008

anonymously report questionable acts to a superior. Given multiple channels, circumstances in which such a channel outperforms a non-anonymous channel are uncertain and not addressed by prior research. In this paper we present the results of an experimental study that provides new evidence on whether the availability of an anonymous reporting channel, relative to a non-anonymous reporting channel, improves reporting intentions in settings where individuals may be particularly hesitant to report fraud. 1 We specifically focus on fraudulent financial reporting as the wrongdoing incident because of the large organizational and societal costs related to fraudulent financial reporting and because the anonymous reporting provisions in SOX are specifically intended as a reporting channel for questionable accounting such as fraudulent financial reporting. Participants in the study were online MBA students who were attending a meeting on campus. They received background information about a company and a single scenario describing the discovery of another employee 1 Much of the prior research in this area is field- or survey-based (Mesmer-Magnus and Viswesvaran, 2005). While these methods have a number of advantages, they generally are not able to control for a variety of potentially relevant factors such as the rank of the participant, the organizational and ethical culture of the organization, the opportunity for observing wrongdoing, the severity of the wrongdoing, and the available channels for reporting wrongdoing. Using an experimental approach allows us to isolate the effect of the manipulated variable of interest, while holding constant the information about the fraudulent financial reporting and the related context in which it occurs. See Maines, Salamon, and Sprinkle (2006) for an excellent discussion of advantages and disadvantages of experimental research approaches.

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

engaging in a fraudulent financial reporting act. Participants, assuming they were facing the situation described in the scenario, provided their intention to report the act through (1) a non-anonymous reporting channel and (2) an anonymous reporting channel. In a between participants design, information about the organization's follow-through outcomes from a previous non-anonymous whistleblowing incident was manipulated, as follows: (1) the presence or absence of retaliation taken against a previous whistleblower, and (2) the presence or absence of negative repercussions taken against a previous transgressor. 2 Consistent with our hypotheses, we find that participants' reporting intentions for the nonanonymous, but not the anonymous, reporting channel are significantly lower under either condition that discourages reporting. Furthermore, we find that when a previous whistleblower faced retaliation or when a previous transgressor did not face negative repercussions, non-anonymous reporting intentions were significantly lower than anonymous reporting intentions. But when no such negative outcomes from the perspective of the whistleblower existed, non-anonymous reporting intentions did not differ significantly from anonymous reporting intentions. 3 In the next section of the paper we develop the study's hypotheses and research questions. This is followed by discussions of our research method and results. The final section of the paper addresses the study's implications as well as its limitations.

89

multiple types of wrongdoing, neither explicitly considered situations where individuals, based on the specific contextual factors, may be especially hesitant to report wrongdoing. We contend that the follow-through outcomes on a previous whistleblowing incident will change either the expected costs or expected benefits to employees as they consider using either an anonymous or non-anonymous reporting channel and that this will affect the likelihood of whistleblowing. This is consistent with Mesmer-Magnus and Viswesvaran (2005, p. 281) who suggest that, “It is assumed that a whistleblower's experiences (perceived or actual, reward or retaliation) following a whistleblowing event will have strong effects on others' willingness and likelihood to blow the whistle in the future.” In our study, participants receive information about the outcomes to the whistleblower and transgressor from a previous whistleblowing incident. Specifically, participants received information that retaliation against a previous whistleblower either did or did not occur and that the transgressor of the previous incident either suffered or did not suffer negative repercussions. Because it was important to provide outcome information about a previous whistleblower, the previous incident involved the use of non-anonymous reporting. That is, it may seem unrealistic that retaliation would be taken against a previous whistleblower who reported anonymously.

2.1. Information about retaliation against previous whistleblower 2. Development of hypotheses and research questions The 2010 Association of Certified Fraud Examiners (ACFE) Report to the Nations on Occupational Fraud and Abuse (2010, 16–17) states that tips, most frequently from employees, are the single most frequent source leading to the detection of fraud. Yet, research results suggest that only about half of all employees who observe wrongdoing report it (Miceli et al. 2008, 22). One approach to increase fraud reporting is to provide employees with more than one reporting method. To the extent that an individual's decision on whether to report wrongdoing is based upon a consideration of the costs and benefits (Gundlach, Douglas, & Martinko, 2003; Hooks, Kaplan, & Schultz, 1994; Miceli and Near, 1985), analyzing reporting alternatives, which trigger different costs and benefits, represents an important topic for research. Mesmer-Magnus and Viswesvaran (2005) perform a metaanalysis of previous research and conclude that in general the threat of retaliation is negatively associated with intentions to report wrongdoing. Miceli et al. (2009, p. 380) speculate that increases in the threat of retaliation may influence whether whistleblowers report internally (e.g., to a reporting channel within the company) or externally (e.g., to a reporting channel not under the company's control). Concerning anonymous vs. non-anonymous reporting channels, Ayers and Kaplan (2005) report that reporting intentions were similar across two cases, each involving ethical problems related to an enterprise resource planning software implementation. Kaplan and Schultz (2007) find that the availability of an anonymous reporting channel tended to reduce the likelihood of reporting to nonanonymous channels. While both of these studies considered 2 The overall study also includes a fifth form of the research instrument, not crossed with this design, which is used as a control to describe a situation in which the participant has no information on a previous whistleblower or transgressor. Results relating to that form of the instrument (n = 20) are presented in the “Supplemental Analysis” section of the paper. 3 Miceli et al. (2009) also suggest that increases in the threat of retaliation may influence whistleblowers' reporting channel. In this regard, they suggest that threats of retaliation increases the tendency for whistleblowers to report to the media or others outside the company. Results from Rehg et al. (2008) indicate that the increases in the threat of retaliation increase the external whistleblowing among women but not men. Neither paper, however, discusses the potential relation between threat of retaliation and anonymous internal reporting.

Rehg, Miceli, Near, and Van Scotter (2008, 222) define retaliation against whistleblowers to be “the undesirable action taken against a whistle-blower – and in direct response to the whistle-blowing – who reported wrongdoing internally (i.e., within the organization) or externally (i.e., outside the organization).” 4 While limited, the available data suggests that retaliation against whistleblowers occurs frequently. For example, Miceli et al. (1999) report that 17% to 38% of whistleblowers in random samples of federal employees have faced retaliation. A more recent study places the percentage at 37% for employees of a military base (Near et al., 2004). Dyck, Morse, and Zingalses (2007 62–63), in examining extremely large corporate fraud cases (companies with over $750 million in assets), report that in 13 of 16 non-anonymously reported cases, the individual involved was fired, resigned, or had significantly reduced responsibilities, although in 14 other cases the resolution was unknown. We expect that information about whether a previous whistleblower, reporting non-anonymously, suffered retaliation will influence reporting intentions for a subsequent fraudulent act. Specifically, we suspect that whether retaliation has been suffered by a previous whistleblower will be used as a signal of whether subsequent potential whistleblowers face an elevated prospect of retaliation. Retaliation represents a potential cost of whistleblowing. Other things equal, increases in one's personal costs should decrease one's likelihood of whistleblowing. If a previous whistleblower has suffered retaliation, this signals that one's personal costs from nonanonymous reporting are likely to be very high. Consequently, we expect that information about a previous whistleblower who has suffered retaliation will decrease non-anonymous reporting intentions. But will retaliation against a previous whistleblower who used a non-anonymous channel also affect a subsequent potential whistleblower's anonymous reporting? To the extent individuals believe that anonymous reporting offers a mechanism to protect one's identity, thus minimizing one's exposure to retaliation, retaliation against 4 In a similar vein, Rehg (1998, p. 17) states, “retaliation against whistleblowers represents an outcome of conflict between an organization and its employee, in which members of the organization attempt to control the employee by threatening to take, or actually taking, an action that is detrimental to the well-being of the employee, in response to the employee's reporting, through internal or external channels, a perceived wrongful action.”

90

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

a previous non-anonymous whistleblower is not expected to decrease anonymous reporting. However, given the complexity of issues related to whistleblowing decision making (Gundlach et al., 2003; Miceli et al., 2008), a question remains whether individuals will consider anonymous reporting as an effective mechanism to protect against retaliation in an environment known to retaliate against whistleblowers. The above discussion leads to our first hypothesis: Hypothesis 1. The effect of retaliation against a previous whistleblower on reporting intentions will differ between anonymous and non-anonymous reporting channels. Specifically, compared to when retaliation against a previous whistleblower did not occur, retaliation against a previous whistleblower will lower non-anonymous reporting intentions but not anonymous reporting intentions. Hypotheses 1 examines whether follow-through outcome information about a previous whistleblower is expected to differentially influence anonymous and non-anonymous reporting intentions. However, this hypothesis does not directly compare anonymous and non-anonymous reporting intentions when a previous whistleblower did and did not suffer retaliation. The Sarbanes-Oxley Act requires an anonymous reporting channel regardless of company history. If a potential whistleblower has confidence in his or her company having appropriate follow-through, there may be no preference between the anonymous and non-anonymous channels. We pose two research questions addressing the appropriate or positive follow-through situations (retaliation vs. no retaliation against the previous whistleblower): Research Question 1a: When a previous whistleblower has suffered retaliation after using a non-anonymous reporting channel, are anonymous reporting intentions stronger than nonanonymous reporting intentions? Research Question 1b: When a previous whistleblower has not suffered retaliation after using a non-anonymous reporting channel, are anonymous reporting intentions stronger than nonanonymous reporting intentions? 2.2. Information about repercussions to a previous transgressor The company chooses what follow-through procedures (if any) to take relating to the transgressor. For example, at one extreme, followthrough actions may result in the transgressor being terminated or even prosecuted, while at the other extreme no direct action against the transgressor may be taken, or that individual may even be rewarded. 5 Punishments directed toward the transgressor, as well as perceptions about the fairness of these punishments, may be influenced by the seriousness of the wrongdoing, the nature and quality of the evidence that is subsequently found from the whistleblowing incident, and the organizational culture and related policies (Trevino & Weaver, 2001). While individuals may vary in views about the severity of punishment with respect to a transgressor of financial statement fraud, in most cases there is general agreement that engaging in financial statement fraud is serious, and that the transgressor should be punished. We expect that information about whether a previous transgressor suffered negative repercussions will influence reporting intentions for a subsequent fraudulent act. Specifically, we suspect that whether a previous transgressor suffered repercussions will be used as a signal of the extent and thoroughness of the company's expected follow-through actions if a subsequent fraudulent act is reported. To the extent that expected benefits are declining, we expect individuals 5 Trevino and Weaver (2001) define follow-through as “the extent to which a company takes action to deal with ethical issues employees raise, and with violations of the company's formal ethics policies.”

will only report if the expected costs are relatively low. Consequently, information about a previous transgressor who has not suffered repercussions will decrease non-anonymous reporting intentions given the relatively high personal cost of non-anonymous reporting. The above discussion leads to our second hypothesis: Hypothesis 2. The effect of negative repercussions suffered by a previous transgressor on reporting intentions will differ between anonymous and non-anonymous reporting channels. Specifically, compared to when repercussions are suffered, a lack of repercussions will lower non-anonymous reporting intentions but not anonymous reporting intentions. Hypothesis 2 examines whether information about a previous transgressor is expected to differentially influence anonymous and non-anonymous reporting intentions. However, this hypothesis, as is the situation with Hypothesis 1, does not directly compare anonymous and non-anonymous reporting intentions when the previous transgressor did and did not suffer negative repercussions. If the potential whistleblower is concerned that the organization does not sufficiently respond to fraud reports (e.g., a previous transgressor did not face repercussions), we expect anonymous reporting intentions will be stronger than non-anonymous reporting intentions. That is, the costs of reporting should be lower using an anonymous reporting channel, while the benefits are likely to be similar. Alternatively, if a potential whistleblower has confidence in his or her company having appropriate follow-through, there may be no preference between the anonymous and non-anonymous channels. The above discussion leads to our second set of research questions: Research Question 2a: When a previous transgressor did not suffer negative repercussions when non-anonymously reported by a whistleblower, are anonymous reporting intentions stronger than non-anonymous reporting intentions? Research Question 2b: When a previous transgressor did suffer negative repercussions when non-anonymously reported by a whistleblower, are anonymous reporting intentions stronger than non-anonymous reporting intentions? 3. Method We obtain responses from participants placed in an experimental setting described by our research instrument. Below we discuss the design, participants, task, independent and dependent variables for the experiment. 3.1. Design The experiment utilizes a 2 × 2 × 2 repeated measures design with reporting intention as the dependent measure. The fully crossed design uses reporting channel (anonymous vs. non-anonymous) as a within participant repeated measure and information about retaliation to the previous whistleblower (information indicates retaliation or no retaliation) and information about the repercussions to the previous transgressor (information indicates that there were negative repercussions or no negative repercussions) as the between participants independent variables. 6 For simplicity, we refer to the former variable as whistleblower retaliation and the latter as transgressor repercussions. 6 Alternatively, the data collected may be viewed as two separate 2 × 2 analyses of variances with between participant factors information about retaliation against a previous whistleblower and information about repercussions against a previous transgressor and no within participant factor. We also ran the analysis in that manner. Although that analysis makes it impossible to directly consider the interaction between the two reporting channels, other results are similar to those subsequently reported in this paper. Detailed results are available from the authors.

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

3.2. Participants Participants were on-line MBA students from a major university who were attending a meeting on campus. 7 Eighty-one participants completed the four crossed forms of the instrument, with sixteen participants failing one or both of the manipulation checks (as discussed further in the results section) removed, leaving 65 participants for the analysis. As indicated in Table 1, our participants had a mean age of over 30 and 60% were male. Mean work experience was over eight years and, on average, were employees with large firms (mean over 22,000 employees). Finally, over 43% of the participants reported that they had discovered a person of greater authority engaging in a questionable or wrongful behavior. Neither analysis of variance (applied to age, years of work experience, number of employees in company) nor a chi square analysis (applied to gender and whether the individual had discovered questionable or wrongful behavior) revealed any statistically significant differences among the groups.

91

Table 1 Demographic information. Total (n = 65) Age Mean 30.4 Std. Dev. 5.7 Gender Female 40% Male 60% Years of work experience Mean 8.5 Std. Dev. 5.7 Number of Employees in Company Mean 22,466 Std. Dev. 45,866 Have you discovered a person of greater authority engaging in questionable or wrongful behavior? Yes 43.1% No 56.9%

3.3. Task Participants were given an experimental instrument describing a hypothetical manufacturer of consumable materials. They were then informed of the company's anonymous and non-anonymous exporting channels. Following this was background information about the company indicating that expenses were underreported. Different forms of the instrument then systematically manipulated information about whether a previous non-anonymous whistleblower faced retaliation and information about whether a previous transgressor faced negative repercussions between participants. Participants were asked to provide their intentions to report the fraudulent act using each of the two reporting channels. After providing these reporting intentions, participants responded to a series of statements about their perceptions of the fraudulent act, personal costs and responsibilities of the employee to report the fraudulent act, and anticipated organizational responses if the fraudulent act was reported. Finally, participants answered manipulation check questions and provided information about their gender and background. 3.4. Nature of the fraudulent act We used a fraudulent financial reporting act in our study. The specific fraudulent act involved improper asset valuation, a frequent method of fraudulent financial reporting (Association of Certified Fraud Examiners, 2006, 16), and one that is similar to the WorldCom fraud in that items that should have been expensed were capitalized (Romero & Berenson, 2002; Verrecchia, 2003). The individual involved was convinced that the supervisor had intentionally changed the expense classification and that this was done to enable the firm to meet the aggressive financial targets that had been established. Note that the case, by design, left little room for doubt regarding whether a fraudulent act had occurred. 3.5. Independent variables The experiment includes three independent variables. Whistleblower retaliation and transgressor repercussions are manipulated 7 The varied work experiences of MBA students make the group of particular interest to a study such as ours. Recent studies have used MBA students to examine reporting intentions for questionable acts. Kaplan et al. (2009) and Kaplan and Schultz (2007) used evening MBA students as participants in their study examining the use of an anonymous reporting channel to report questionable events. Ayers and Kaplan (2005) used day MBA students to address employee reporting intentions relating to wrongdoing by consultants. More generally, experimental research in accounting commonly uses MBA students (Kaplan & Mauldin, 2008) or Masters of Accounting students (Ugrin & Odom, 2010) as participants.

between participants and reporting channel is manipulated within participants. We discuss each in turn. 3.5.1. Whistleblower retaliation Two levels of whistleblower retaliation are included in the experiment. Under the retaliation present condition, the case included information indicating that a previous non-anonymous whistleblower faced retaliation. Specifically, the manipulation read, in part, “You also have heard of a similar situation from last year in which an employee reported a questionable act directly to his supervisor's supervisor…, you did hear that shortly thereafter, the reporting employee resigned, amid rumors that he ‘felt uncomfortable over the incident’, in part, because he had been informally told that he ‘would never be promoted.’” Under the retaliation not present condition, the case included information that a previous non-anonymous whistleblower did not face retaliation. The manipulation read, in part, “You also have heard of a similar situation from last year in which an employee reported a questionable act directly to his supervisor's supervisor. … you did hear that the reporting employee was commended for his action and has continued to advance within the company.” 3.5.2. Transgressor repercussions Two levels of transgressor repercussions are included in the experiment. Under the repercussions present condition, the case included information indicating that a previous transgressor faced negative repercussions. Specifically, the manipulation read, in part, “… you also heard that the supervisor who committed the act (a division manager) resigned, apparently over this incident.” Under the not present condition, the case included information that a previous transgressor did not face negative repercussions. The manipulation read, in part, “…you also have heard that the supervisor who committed the act (a division manager) faced no repercussions as he continued in his position and is well compensated.” 3.5.3. Reporting channel Participants provided fraud reporting intentions via two reporting channels: a non-anonymous and an anonymous reporting channel. 8 The non-anonymous reporting channel involved reporting to one's 8 To familiarize participants with the task, we first asked participants to provide reporting intentions with regards to the individual in the case who was aware of the questionable act. Those results, available from the authors, were similar to those reported in the paper, with the interaction between reporting channel and whistleblower retaliation somewhat stronger and that between transgressor repercussions somewhat weaker. We do not provide details of that analysis here as our primary intent was to obtain participant personal reporting intentions.

92

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

Table 2 Analysis of reporting channel, occurrence of retaliation to previous whistleblower, and occurrence of repercussions to previous transgressor on reporting intentions. Panel A: ANOVA Source

Mean square

F-value

p-value (1-tailed)*

Reporting channel Retaliation to previous whistleblower Repercussions to previous transgressor Reporting channel ⁎ retaliation Reporting channel ⁎ repercussions Retaliation ⁎ repercussions Reporting channel ⁎ occurrence of retaliation ⁎ occurrence of repercussions Error

36.40 5.382 3.935 35.66 12.66 .96 .125 .299

7.77 1.80 1.32 7.62 2.70 .322 .027

.004 .092 .128 .004 .053 .286 .436

Present (resigned)

Not present (continued to advance)

Total

3.52 (2.49) 27 5.85 (1.54) 27 4.69 (2.36) 54

5.16 (1.87) 38 5.13 (1.92) 38 5.15 (1.92) 76

4.48 (2.28) 65 5.43 (1.79) 65 4.96 (2.10) 130

Panel B: reporting intention means, (Std Dev), and N for whistleblower retaliation and reporting channel Reporting channel

Non-anonymous

Anonymous

Total

Whistleblower retaliation

Panel C: reporting intention means, (Std Dev), and N for previous transgressor and reporting channel Reporting channel

Non-anonymous

Anonymous

Total

Transgressor repercussions Not present (continued)

Present (resigned)

Total

3.89 (2.42) 35 5.60 (1.68) 35 4.75 (2.24) 70

5.17 (1.91) 30 5.23 (1.92) 30 5.20 (1.91) 60

4.48 (2.28) 65 5.43 (1.79) 65 4.96 (2.10) 130

*Participants provided their assessments using a 7-point scale (from 1 to 7) with endpoints labeled “Very Unlikely” and “Very Likely” as to “How likely is it that you would report this instance of questionable behavior to the anonymous ethics hotline” and “your supervisor's supervisor.” Significant p-values are in bold.

supervisor or, if the supervisor was suspected (the situation in the experiment), to that person's supervisor. Supervisors were described as having a general responsibility for follow-up on reports of suspected illegal practices or instances of questionable behavior by employees under their authority. The channel required reporters to identify themselves, and consequently, report using this channel nonanonymously. The anonymous reporting channel involved reporting via a telephone hotline. 9 The case indicated that all information reported to the hotline is anonymous and confidential. 3.6. Dependent measures The research instrument uses a 7 point scale (ranging from very unlikely to very likely) and requires participants to respond to the anonymous ethics hotline and a supervisor above the individual involved with the misstatement. These dependent measures are similar to previous research (Ayers & Kaplan, 2005; Chiu, 2003; Kaplan, Pany, Samuels, & Zhang, 2009; Kaplan & Whitecotton, 2001; Schultz, Johnson, Morris, & Dyrnes, 1993; Singer, Mitchell, & Turner, 1998). 9 Anonymous reporting may take various forms, including web-based, telephone, fax, or written correspondence. We selected the telephone hotline because it is commonly used in practice. We designed the telephone hotline based on information from the Association of Certified Fraud Examiners (2006) continuing professional education course on hotlines and the websites of the two leading independent company hotline administrators—The Network (www.tnwinc.com/index.asp) and Global Compliance (2007) (www.globalcompliance,com). We also interviewed Friedman (2007) then vice president of Global Compliance on matters relating to the actual hotlines.

4. Results 4.1. Manipulation check We included two manipulation check questions to determine whether participants attended to our manipulations. First, respondents were asked whether the individual who reported the earlier questionable act “Continued to advance in the company” or “Resigned.” Second, a question was included on whether the individual who committed the earlier questionable act “Was still employed by the company” or “Resigned.” As indicated previously, sixteen participants incorrectly answered the manipulation check and were dropped from the statistical analysis. However, our results are qualitatively similar when responses from all participants are analyzed.

4.2. Tests of hypotheses and research questions Analysis-of-variance (ANOVA) and within-participant t-tests are used to test our hypotheses and research questions. Reporting intentions are the dependent measure in each analysis. For the ANOVA, reporting channel (anonymous or non-anonymous) is a within participants independent variable, and whistleblower retaliation and transgressor repercussions are between participants independent variables. Statistical results for the ANOVA are presented in Table 2 Panel A and related descriptive statistics are presented in Table 2 Panel B.

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

93

Table 3 Means of participants' perception about the fraudulent act by level of whistleblower retaliation and transgressor repercussions.

1. 2. 3. 4. 5. 6. 7. 8. 9.

b

Act is morally wrong. Act seriousness. Act is unethical (general consensus). Act Fairness. Personal cost reporting anonymously. Personal cost of reporting to a supervisor. Responsibility to inform. Act will be thoroughly investigated. Supervisor (transgressor) in this situation will be forced to resign.

Previous whistleblower

Previous transgressor

Retaliation

Negative repercussions

Not Present

Present

p-value

Not Present

Present

p-valuea

7.50 7.45 7.55 4.41 4.16 7.13 8.11 4.82 4.95

7.37 7.33 7.59 3.73 5.62 8.11 8.19 4.63 4.67

.781 .760 .913 .271 .028 .026 .773 .263 .616

7.40 7.33 7.57 4.46 4.10 7.00 8.30 5.47 5.77

7.49 7.46 7.57 3.86 5.29 8.00 8.00 4.27 4.03

.852 .737 .989 .318 .073 .021 .271 .089 .001

a

Significant p-values (2-tailed) are in bold. Participants provided their assessment using a 9-point scale (from 1 to 9). Questions 1, 3, and 9 were labeled Strongly Disagree to Strongly Agree; Question 8 was labeled Strongly Disagree to Strongly Agree, but for presentation purposes we have reversed the scale to be consistent with the other questions. Questions 2, 5, 6, and 7 were scaled from Very Low to Very High. Question 4 was scaled from Very Unfair to Very Fair. b

Hypothesis 1 predicts that the effect of whistleblower retaliation on reporting intentions will differ between anonymous and nonanonymous reporting channels. Specifically, compared to when retaliation against a previous whistleblower did not occur, retaliation against a previous whistleblower will lower non-anonymous reporting intentions but not anonymous reporting intentions. It is tested first by the interaction between reporting channel and retaliation and then by individual comparisons of the related means. Table 2, Panel A indicates that the interaction between reporting channel and retaliation is significant (p b .004). To further test hypothesis one, we first compare non-anonymous reporting intention means (see Table 2, Panel B) between the present condition (3.52) and the not present condition (5.16) of whistleblower retaliation; a one-way ANOVA relating to these means is significant (F = 9.2, p. b .003). 10 A second comparison is made for anonymous reporting intention means (Table 2, Panel B) also between the present condition (5.85) and the not present condition (5.13) of whistleblower retaliation. This result is consistent with hypothesis one since the mean anonymous reporting intention is not significantly different from that under the present condition of whistleblower retaliation. Thus, the results support Hypothesis 1. Research question 1a asks whether anonymous reporting intentions will be stronger than the non-anonymous reporting intentions when previous whistleblower retaliation occurred. The anonymous reporting intention mean (5.85) is significantly stronger than the non-anonymous reporting intention mean (3.52) (t = 3.9, p b .001, within-participants t-test). This indicates that when a previous whistleblower faced retaliation, anonymous reporting intentions are significantly stronger than non-anonymous reporting intentions. Research question 1b addresses whether anonymous reporting intentions will be stronger than non-anonymous reporting intentions absent retaliation to the previous whistleblower. The respective means from Table 2 (5.13 vs. 5.16) do not differ significantly. Thus, in summary, the anonymous reporting channel only outperforms the nonanonymous reporting channel in circumstances in which the previous whistleblower faced retaliation. Hypothesis 2 predicts that the effect of transgressor repercussions on reporting intentions will differ between anonymous and nonanonymous reporting channels. Specifically, compared to when a previous transgressor suffered repercussions, non-anonymous reporting intentions, but not anonymous reporting intentions, will be lower when a previous transgressor did not suffer repercussions. Similar

10

Two-tailed statistical tests are presented.

to hypothesis 1, we initially consider the interaction term between reporting channel and repercussions and then by individual comparisons of the related means. As shown in Table 2, Panel B the interaction between reporting channel and transgressor repercussions is significant (p b .053). The patterns of means (Table 2, Panel C) are also consistent with hypothesis two. First, non-anonymous reporting intention means are significantly lower under the not present condition (3.89) compared to the present condition (5.17) of transgressor repercussions (F = 5.5, p b .03). Second, anonymous reporting intention mean is not significantly different from the not present condition (5.60) compared to the present condition (5.23) of transgressor repercussions. Thus, the results support hypothesis 2. Research question 2a compares non-anonymous and anonymous reporting intentions when a previous transgressor has not suffered negative repercussions. The anonymous reporting intention mean (5.60) is significantly bigger than the non-anonymous reporting intention mean (3.89) under the not present condition of transgressor repercussions (t = 3.0, p b .01, within-participants t-test). Finally, research question 2b addresses whether anonymous reporting intent will be stronger than non-anonymous reporting intent under negative repercussions to a previous transgressor. The means for when repercussions were present (5.23 v. 5.17) do not differ significantly. In summary, as is the situation with retaliation to a previous whistleblower (research questions 1a and 1b), reporting intent for the anonymous reporting channel only exceeds the non-anonymous reporting channel when transgressor repercussions were not present relating to the previous whistleblowing incident. 4.3. Supplemental analysis 4.3.1. Perceptions related to the fraudulent act Our research instrument also included a series of nine statements about participants' perceptions of the fraudulent act, personal costs and responsibilities of the employee to report the fraudulent act, and expected organizational responses if the fraudulent act is reported. These statements represent factors that previous research has identified as potentially influential in whistleblowing (Miceli and Near 1992; Ponemon, 1994; Schultz et al., 1993; Smith, Keil, & DePledge, 2001). Participants' mean response for each statement is presented in Table 3 for each level of whistleblower retaliation. The first four statements in Table 3 relate to perceptions of the fraudulent act. The means for these first three statements indicate that the act is considered morally wrong, serious, and unethical (e.g., well above the scale mid-point), and that the fourth statement is near the scale mid-point. Further, means for these four measures

94

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

did not significantly differ by level of whistleblower retaliation. The next three statements relate to perceptions about oneself in connection to the fraudulent act. Statements 5 and 6, regarding the personal costs of reporting to a report recipient, significantly differ by level of whistleblower retaliation. Mean personal costs of reporting were significantly higher across report recipients when a previous whistleblower suffered retaliation and when a previous transgressor did not suffer repercussions, as contrasted to when there was no retaliation and when there were previous repercussions. Mean responses for statement 7 indicate participants perceived a responsibility to inform and that the means did not significantly differ by level of whistleblower retaliation or transgressor repercussions. Statements 8 and 9 differed significantly in the case of the transgressor manipulation. As expected, when a previous transgressor had suffered negative repercussions, respondent considered it more likely that the act would be thoroughly investigated and that the supervisor would be forced to resign. Next, we investigated whether any of the significant variables served as mediator variables, variables intended to increase understanding of the process by which individuals “transform predictor or input variables” (Baron & Kenny, 1986, 1179). For both previous whistleblower retaliation and previous transgressor repercussions only statement 6 serves as a mediator variable. For previous whistleblower retaliation the mediation is partial (F = 5.71, p. b .02) and for transgressor repercussions the effect is stronger (F = 2.70, p b .11). Overall these results indicate that assessments of one's personal costs of reporting to a supervisor mediate the relationship between both whistleblower repercussions and transgressor repercussions. 4.3.2. No information about a previous whistleblowing event As indicated in an earlier footnote, the study also included a fifth form of the instrument which did not include any information about the outcomes a previous whistleblowing incident. The fifth form of the instrument only included the background and other information included in the other four versions of the instrument. Thus, the fifth form of the instrument was identical to the other four forms, except for the exclusion of information about a previous whistleblowing incident. We refer participants receiving this fifth form (n = 20) as the control group. The mean (standard deviation) anonymous reporting intention is 4.9 (2.2), and the mean non-anonymous reporting intention is 5.4 (1.6). We compare reporting intentions from the control to treatment means for whistleblower retaliation and then to treatment means for transgressor repercussions. Thus, we compare reporting intentions from the control group to each of the two treatment means for whistleblower retaliation and the two treatment means for transgressor repercussions. These comparisons will provide evidence on whether participants receiving a particular level of an independent variable formed reporting intentions that differed from the reporting intentions formed by participants in the control group. We ran four separate 2 × 2 ANOVAs. The independent variables were group, manipulated between participants at two levels (e.g., control group vs. one of the other four treatment groups), and reporting channel, manipulated within participants at two levels (e.g., an anonymous reporting channel and anonanonymous reporting channel). Reporting intentions for the control group did not significantly differ from the reporting intentions of either the not present condition of whistleblower retaliation or the present condition of transgressor repercussions. This suggests that participants generally did not find information indicating that a previous whistleblower did not face retaliation or that a previous transgressor did suffer repercussions informative with respect to their reporting intentions. However, reporting intentions for the control group did significantly differ from the reporting intentions of the present condition of whistleblower retaliation and the not present condition of transgressor

repercussions. This suggests that participants generally did find information indicating that a previous whistleblower faced retaliation or that a previous transgressor did not suffer repercussions informative with respect to reporting intentions.

5. Discussion This study provides evidence on intentions to report fraud in settings where individuals may be particularly hesitant to report fraud. Our results indicate that participants' non-anonymous reporting intentions were significantly lower when a previous (non-anonymous) whistleblower experienced retaliation compared to when a previous (non-anonymous) whistleblower did not experience retaliation. This finding is consistent with previous research and hardly unexpected (e.g., Gundlach et al., 2003; Miceli and Near, 1992; Miceli et al., 2008). However, the design of our study also allows us to address whether retaliation against a previous (non-anonymous) whistleblower had a similar effect on anonymous reporting intentions. Our findings indicate that retaliation against a previous (non-anonymous) whistleblower was not significantly associated with participants' anonymous reporting intentions. Our results for the effect of repercussions to a previous transgressor on participants' reporting intentions parallel our retaliation results. In this regard, reporting intentions were significantly lower under the non-anonymous reporting channel but not the anonymous reporting channel when the transgressor did not face repercussions. These findings suggest that the availability of an anonymous reporting channel only increases reporting intentions in circumstances in which previous whistleblowing results had been negative from the perspective of the whistleblower. We believe our findings are relevant to policy makers or others concerned with evaluating the whistleblowing provisions of the Sarbanes-Oxley Act. While establishing and providing oversight for anonymous reporting channels is costly, the implicit expectation of this mandate is that the availability of an anonymous reporting channel would increase the likelihood of the reporting of existing fraud, at least, in some circumstances, by lowering the personal costs of reporting. Consistent with this implicit expectation, the results of our study show that the availability of an anonymous reporting channel increased such reporting intentions, relative to non-anonymous reporting intentions, in each of two different circumstances where individuals may be particularly hesitant to report fraud— (1) when a previous whistleblower faced retaliation and (2) when a previous transgressor did not face negative repercussions. Further, for both circumstances, assessment of one's personal costs of reporting to a supervisor at least partially mediates the relationship. Interestingly, for both circumstances, assessments of one's personal costs of reporting anonymously also increased. Overall, our findings suggest that the anonymous whistleblowing requirements under Section 301 of the Sarbanes-Oxley Act appear to increase internal reporting. Specifically, when outcome information from a previous incident was negative from the whistleblower's perspective, the availability of an anonymous reporting channel increased the likelihood of internal reporting. The results of this study must be viewed in light of its limitations. First, using an experimental approach, with participants responding to a single hypothetical incident, is not the same as discovering a fraudulent act in their actual work environment. Second, although our MBA participants are working professionals, they may not represent a cross-section of the entire relevant work force. We note, however, that Table 1 reveals that over 43% of the respondents had discovered a person of greater authority engaging in some form of questionable or wrongful behavior. Third, the study does not address possible reward for reporting illegal acts such as those implemented by the Securities and Exchange Commission in implementing the Dodd-Frank Act (SEC, 2011). Finally, our case addresses a situation

S.E. Kaplan et al. / Advances in Accounting, incorporating Advances in International Accounting 28 (2012) 88–95

in which participants were given no history related to effects of anonymous hotline reporting. References Association of Certified Fraud Examiners (2005). Detecting & Deterring Fraud Using Hotlines. Austin, Texas: Continuing Education Course. Association of Certified Fraud Examiners (2006). Report to the Nation on Occupational Fraud and Abuse. Austin, Texas: ACFE. Association of Certified Fraud Examiners (2010). Report to the Nations on Occupational Fraud and Abuse. Austin, Texas: ACFE. Ayers, S., & Kaplan, S. E. (2005). Wrongdoing by consultants: an examination of employees' reporting intentions. Journal of Business Ethics, 57, 121–137. Baron, R., & Kenny, D. (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. Personality and Social Psychology, 51, 1173–1182. Chiu, R. K. (2003). Ethical judgment and whistleblowing intention: examining the moderating role of locus of control. Journal of Business Ethics, 43(1), 65–74. Dyck, A., Morse, A., & Zingalses, L. (2007). Who blows the whistle on corporate fraud? Working paper 12882. : National Bureau of Economic Research (February 2007). Friedman, Adam (2007). Telephone interview. (July 10, 2007). Global Compliance (August 20, 2007). Alert line, allegation reporting. Website. Gundlach, M. J., Douglas, S. C., & Martinko, M. J. (2003). The decision to blow the whistle: a social information processing framework. Academy of Management Review, 28(1), 107–123. Hooks, K. L., Kaplan, S. E., & Schultz, J. J., Jr. (1994). Enhancing communication to assist in fraud prevention and detection. Auditing: A Journal of Practice & Theory, 86–117 (Fall). Kaplan, S. E., & Mauldin, E. G. (2008). Auditor rotation and the appearance of independence: evidence from non-professional investors. Journal of Accounting and Public Policy, 27(2), 177–192. Kaplan, S. E., Pany, K., Samuels, J., & Zhang, J. (2009). An examination of the effects of procedural safeguards on intentions to anonymously report fraud. Auditing: A Journal of Practice & Theory, 28, 273–288. Kaplan, S. E., & Schultz, J. J. (2007). Intentions to report questionable acts: an examination of the influence of anonymous reporting channel, internal audit quality, and setting. Journal of Business Ethics, 71, 109–124. Kaplan, S. E., & Whitecotton, S. M. (2001). An examination of auditors' reporting intentions when another auditor is offered client employment. Auditing: A Journal of Practice & Theory, 20, 45–64. Maines, L., Salamon, G., & Sprinkle, G. (2006). An information economic perspective on experimental research in accounting. Behavioral Research in Accounting, 18, 85–102. Mesmer-Magnus, J., & Viswesvaran, C. (2005). Whistleblowing in organizations: an examination of correlates of whistleblowing intentions, actions and retaliation. Journal of Business Ethics, 62, 277–297. Miceli, M., & Near, J. (1985). Characteristics of organizational climate and perceived wrongdoing associated with whistle-blowing decisions. Personnel Psychology, 38, 525–544.

95

Miceli, M., & Near, J. (1992). Blowing the Whistle: The Organizational and Legal Implications for Companies and Employees. New York: Lexington Books. Miceli, M., Near, J., & Dworkin, T. (2008). Whistle-blowing in Organizations. Routledge Taylor & Francis Group (New York). Miceli, M., Near, J., & Dworkin, T. (2009). A word to the wise: how managers and policy makers can encourage employees to report wrongdoing. Journal of Business Ethics, 86(3), 379–396. Miceli, M., Rehg, P., Near, M., & Ryan, K. (1999). Can laws protect whistle-blowers? Results of a naturally occurring field experiment. Work and Occupations, 26(1), 129–151. Near, J., VanScotter, J., Rehg, M., & Miceli, M. (2004). Does type of wrongdoing affect the whistle-blowing process? Business Ethics Quarterly, 14(2), 219–242. Ponemon, L. (1994). A comment on 'whistleblowing' as an internal control mechanism: individual and organizational considerations. Auditing: A Journal of Practice & Theory, 13, 118–130. Rehg, M. T. (1998). An examination of the retaliation process against whistle-blowers: A study of federal government employees. Indiana University, Bloomington: Unpublished doctoral dissertation. Rehg, M. T., Miceli, M. P., Near, J. P., & Van Scotter, J. R. (2008). Antecedents and outcomes of retaliation against whistle-blowers: gender differences and power relationships. Organization Science, 19, 221–240. Romero, S., & Berenson, A. (June 26, 2002). WorldCom says it hid expenses, inflating cash flow $3.8 billion. : New York Times. Schultz, J. J., Johnson, D. A., Morris, D., & Dyrnes, S. (1993). An investigation of the reporting of questionable acts in an international setting. Journal of Accounting Research, 75–103 Supplement. Securities and Exchange Commission (2011). Implementation of the whistleblower provisions of Section 21F of the Securities Exchange Act of 1934. (Release No. 34–64545; File No. S7-33-10). Singer, M., Mitchell, S., & Turner, J. (1998). Consideration of moral intensity in ethicality judgments: its relationship with whistle-blowing and need-for-cognition. Journal of Business Ethics, 17, 527–541. Smith, H. J., Keil, M., & DePledge, G. (2001). Keeping mum as the project goes under: toward an explanatory model. Journal of Management Information Systems, 18, 189–227. The Network, Inc. (2006). Corporate Governance and Compliance Hotline Benchmarking Report. Trevino, L., & Weaver, G. (2001). Organizational Justice and ethics program “followthrough” influences on employees' harmful and helpful behavior. Business Ethics Quarterly, 11(4), 651–671. Ugrin, J., & Odom, M. (2010). Exploring Sarbanes-Oxley's effect on attitudes, perceptions of norms, and intentions to commit financial statement fraud from a general deterrence perspective. Journal of Accounting and Public Policy, 29(5), 439–458. Verrecchia, R. E. (2003). Why all the hoopla about Enron? Journal of Accounting and Public Policy, 22(2), 99–105.