Reporting Self-Made Errors: The Impact of Organizational Error-Management Climate and Error Type

In very broad terms, “ethical behavior” has been characterized in the literature as “both legal and morally acceptable to the larger community” (Jones 1991, p. 367; quite similar definitions were offered by Lewis 1985, p. 381, or Treviño et al. 2006, p. 952). In our study, the decision of an auditor to not report an error that he/she has just discovered but which is still unknown by the auditor’s superiors would be considered unethical in light of this norm. The ultimate objective of the audit is to assure that the audited financial statements are free of material misstatement and hence serve the stakeholder needs. In order to ensure that this goal is met, superiors in audit firms need to be informed by their subordinate auditors about all errors, because errors may be connected to misstatements and knowledge about errors is crucial to be able to make appropriate decisions and ultimately issue a correct audit opinion. Hence, audit firms’ “organizational norms and expectations” (Vardi and Wiener 1996, p. 153) and their “legitimate interests” (Marcus and Schuler 2004, p. 648) entail that the reporting of discovered errors by auditors to their superiors would constitute ethical behavior, while non-reporting would violate these norms, expectations, and interests as such non-reporting has high potential to be harmful to the organization or its members (Kaptein 2008, p. 980) and hence be unethical (Sackett and DeVore 2001, pp. 145–146).We reach this conclusion by considering that non-reporting involves potential harm to the organization given the audit firm’s liability for incorrect audit opinions as well as to the client and to the society at large when stakeholders make erroneous decisions based on incorrect financial statements or audit opinions.

Gronewold and Donle (2011) employ a survey to examine whether organizational error climate affects the strategies auditors use in dealing with the discovery of client-committed errors. They report an indirect link between organizational error climate and the strategies auditors employ to deal with the discovery of client-committed errors via the auditor’s personal reflections on how they dealt with their own errors. This study provides evidence as to the relevance of this construct, organizational error-management climate, but does not address any issue in our study.

Absent some third party inspection or court case, which is a relatively rare event (see Palmrose 1991), albeit becoming somewhat less rare with regulatory inspections. Professional standards on engagement quality assurance (ISA 220, IAASB 2010a; ISQC1, IAASB 2010c) are clear that each level of review is not expected to redo all the detailed reviewing of the previous level but instead ensure on the basis of “selected engagement documentation” (ISA 220.20, p. 130; ISQC 1.37, p. 47) that the previous review was done conscientiously and that all matters still open for resolution at the end of the prior level of review are resolved or highlighted for further attention in subsequent sequential reviews prior to the audit report being finalized. In addition, in the explanatory material to ISA 220, we note the following in A18 with respect to engagement partner review, which by its very nature is more comprehensive than quality assurance review “The engagement partner need not review all audit documentation, but may do so. However, as required by ISA 230 (IAASB 2010b), the partner documents the extent and timing of the reviews.” Hence, the discovery of an error is unlikely at the stage of review completion.

The potential of self-discovery and reporting of errors as a contribution to work quality can be expected to be even higher in other organizations that do not have a formal review process. Hence, finding support for the effectiveness of such mechanisms in the audit context should be informative for other institutional settings as well.

The German audit setting at a micro-level is structurally equivalent to Anglo-Saxon audit setting, particularly regarding the audit approaches and procedures applied in practice. Hence, the process of a financial statement audit does not substantially differ. German national standards on auditing do (and did at the time of our study) mirror the International Standards on Auditing (ISAs) and are continuously adapted to any changes that occur to the underlying ISAs. As our case is about a privately held firm, in USA the standards issued by the AICPA’s Auditing Standards Board (ASB) would apply. The US ASB has adopted ISAs as the primary source for its revisions to its Statements on Auditing Standards (SASs). Hence, the standards on auditing both in USA and in Germany are (and were at the time of our study) equivalent in all material respects for a privately held company. Furthermore, the vast majority of our participants worked for Big-4 audit firms who attempt to standardize their methodology around the world to the extent possible. Hence, the ISAs form the framework of all Big-4s’ audit manuals albeit with extensive additions to deal with the US public company Securities and Exchange Commission registrants who are audited under Public Company Accounting Oversight Board (PCAOB) standards. Furthermore, in recent years PCAOB standards have been influenced extensively by the ISAs as can be seen in PCAOB AS #8–15.

Other errors that could be discovered include audit nature and extent judgments that while not conforming to accepted norms or standards provide no direct evidence that the financial statements are in error or misstated (e.g., Peytcheva and Gillett 2012). Furthermore, we do not consider events that violate the professional code of ethics for auditors (e.g., an auditor receiving a job offer from a client during the audit) but have no direct implications for errors in the financial statements (e.g., Taylor and Curtis 2010).

Payne et al. (2010) find such effects of review anticipation for “lower” levels of review (manager review), particularly when interactive review formats such as interviews are expected. This anticipation increases auditors’ efforts to be well prepared for the review and the potential questions asked by the reviewers, and in turn enhances auditors’ ability to detect conceptual and calculation errors (in Payne et al.’s study the errors were in the client’s financial statements).

This mechanism has been studied in various areas, including verbal monitoring (e.g., Ganushchak and Schiller 2006, 2008a, b; Hartsuiker and Kolk 2001; Levelt 1983, 1989; Nooteboom and Quené 2008; Postma 2000) and action and performance monitoring (e.g., Desmurget and Grafton 2000; Luu et al. 2000; Rodríguez-Fornells et al. 2002).

For the positive effect of communicating about errors on effective organizational learning see also Homsma et al. (2009).

Peytcheva and Gillett (2012) document significant between-office differences in willingness to suppress audit evidence that is indicative of previous inaccurate assessment of internal control strength and between-office differences in reputation effects of reporting errors discovered.

For the general importance of motivational factors in psychological decision theories that give reasons to engage in self-protection and take actions to maintain a positive self-image and sense of competence see Larrick (1993); for organizational antecedents that may entail self-presentational and impression-management behaviors see Ashforth and Lee (1990).

We do not predict a main effect of error type on reporting willingness, for two reasons. First, we had no compelling theoretical reason for such a prediction. While a variety of theoretical perspectives might apply (e.g., availability heuristic) none of them lead to a compelling rationale for making a prediction. Second, the focus of our article is on the error-management climate effect and hence knowing that these two error types are well documented in the audit review literature, we focus on examining the moderating effects of error type on the relationship between error-management climate and reporting willingness.

This is in addition to the general requirement in Germany that companies have what is referred to as a statutory audit in the Anglo-Saxon world.

The SNBAD is a standard audit working paper, in which all financial statement errors discovered during an audit are recorded, including the proposed adjustments for their correction, their absolute amounts, and overall effect on client firm earnings. Hence, the SNBAD provides crucial information for the decision of the responsible audit managers and partners whether the financial statements overall are still correctly stated (i.e., if total adjustments for errors are still below the defined materiality level and thus “tolerable”) or are materially misstated (i.e., if total adjustments exceed materiality).

Studies show that for behaviors with moral overtones, people’s prediction of what others would do is more accurately correlated with others’ actual behavior than people’s prediction of what they themselves would do (Epley and Dunning 2000). The social desirability bias literature also indicates that indirect questions asking people to predict others’ behavior suffer from less social desirability bias compared to direct questions asking people to predict their own behavior (Fisher 1993; Fisher and Tellis 1998; Sherwood 1981). Prior studies have measured social desirability bias as the difference between people’s report of what they say others would do and what they personally would do (Cohen et al. 2007; Cohen et al. 1993; Chung and Monroe 2003). People exhibit more social desirability bias when they perceive the behaviors in question as more socially undesirable (Cohen et al. 1993; Chung and Monroe 2003; Fisher 1993; Tourangeau and Yan 2007). Hence, we use this approach when eliciting a variable that had substantial social desirability overtones.

We did this so that response differences would be less likely to be subject to primacy effects due to different initial error amounts (e.g., Yates and Curley 1986) being noted by the participants.

This adjustment allowed for the final error projected in each treatment condition to be the same (i.e., within €1,000 of the total misstatement of approximately €1,020,000), allowing us to reduce the likelihood of recency effects (e.g., Baddeley and Hitch 1993) due to the final total misstatement being different.

There were no significant differences in dependent variable responses between firms.

Participants’ audit experience ranged from 3 to 160 months. Only 10 % of the participants had <12 months of experience. The median was 23 months, the mode 24 months.

Eliminating the control variable altogether from the analysis reported in the article only strengthens the significance level of the results. Furthermore, two items that were highly correlated with the control variable employed, self-reported own error-management climate experience and self-reported experience with statistical audit sampling, varied significantly across experimental treatments. Using either of these variables as the control variable makes no difference to the reported results. Furthermore, neither variable has explanatory power of its own when the treatment variables are removed from the analysis. Finally, neither variable interacts with any of the treatment variables. The low response rate to own error-management climate experience is due to one of the participating firms declining to include this measure in the post-test debriefing questions.

The seven questions are based on items from the underlying psychology literature on error-management climate (e.g., Van Dyck et al. 2005). Three questions were about climate issues directly manipulated in the case description and four questions were about broader inferences about error-management climate that were not found in a 1:1 fashion. Those questions were taken from the “organizational error management culture” scale developed by Van Dyck et al. (2005), which has already been successfully validated in an auditing setting (Gronewold and Donle 2011).

The seven items loaded onto a single factor that captures 69.88 % of the variance of the items, so we computed an average index of the seven. In addition to the exploratory factor analysis, a reliability analysis revealed a high Cronbach’s alpha (0.92), indicating that the single identified factor measures the underlying construct with a high degree of consistency.

A total of 12 (14) participants answered the error-management climate (error type) manipulation in the wrong direction. After omission of the participants who responded erroneously to the manipulation checks we re-ran the ANCOVA with a sample of n = 151. All F and p values as well as the pattern of the interaction are equivalent with our original results.

We thank participants at various workshops, especially those with recent audit experience, for suggesting this explanation for our results.

Remarkably, Kuenzi and Schminke (2009) note that this lack of research even holds for organizational work climates in general, given their review “failed to uncover a single study aimed at modeling and testing the processes by which such changes in work climates emerge” (p. 707).