: Audit firms are investing billions of dollars into artificial intelligence (AI) technology. These investments have the potential to transform the audit landscape and potentially improve the efficiency and effectiveness of audit processes. These investments might also result in wider benefits for audit firms, as audit firms can leverage this technology to reinforce perceived objectivity and trust in the audit. This article summarizes a recent study by Libby and Witz (2024) that finds that the use of AI can help mitigate auditor liability by limiting the effects of the appearance of auditor independence conflicts on juror negligence assessments. This article discusses the implications of these findings. It discusses how these findings indicate a potential solution to a long-standing dilemma that audit firms have faced, where audit firms may desire to invest in favorable relationships with their audit clients while also being seen as performing objective and unbiased audit work.

In this study, we examine whether the use of artificial intelligence (AI) can reduce the effect of independence conflicts on audit firm liability. In two experiments, we manipulate (1) whether procedures are performed by a human auditor or with AI and (2) whether the audit firm was careful in maintaining the appearance of independence from the audit client. Results of both experiments indicate that the use of AI significantly reduces the impact of the appearance of independence conflicts on jurors' judgments of audit firm liability. When concerns relating to the appearance of independence conflicts are present, the use of AI helps maintain the perceived objectivity of the auditor, which results in jurors maintaining higher overall trust in the audit process. Our study contributes to literature on determinants of auditor litigation risk and how technological change that is likely to grow in prominence might affect audit firm liability. Dans quelle mesure l'intelligence artificielle peut-elle attenuer l'impact des conflits d'independance sur la responsabilite des cabinets d'audit?Dans cette etude, les auteurs examinent si l'utilisation de l'intelligence artificielle peut attenuer l'impact des conflits d'independance sur la responsabilite des cabinets d'audit. Les auteurs realisent deux experiences pour determiner (1) si les procedures sont menees par un auditeur humain ou par une intelligence artificielle et (2) si le cabinet d'audit a veille a preserver l'apparence d'independance a l'egard du client. Les resultats des deux experiences indiquent que l'utilisation de l'intelligence artificielle reduit de facon significative l'impact des conflits d'apparence d'independance sur les jugements des jures en matiere de responsabilite des cabinets d'audit. Lorsque des preoccupations liees a l'apparence de conflits d'independance se creent, l'utilisation de l'intelligence artificielle aide a maintenir l'objectivite percue par l'auditeur, ce qui permet aux jures de conserver une plus grande confiance dans le processus d'audit. L'etude vient enrichir les connaissances sur les determinants du risque de litige des auditeurs et sur la maniere dont les changements technologiques, appeles a prendre de plus en plus d'importance, peuvent determiner la responsabilite des cabinets d'audit.

This paper discusses the role of process evidence in accounting research. We define process evidence broadly as data providing insight into how and why cause-effect relationships occur, and we provide a framework to guide the provision and evaluation of process evidence in accounting studies. Our definition allows for an expanded understanding of techniques for gathering process evidence. The framework highlights the importance of the study's goals and theory in choosing how to provide process evidence, as well as how much process evidence to provide. The paper also outlines the strengths and limitations of three approaches to providing process evidence: mediation, moderation, and multiple-study-based designs. We provide recommendations for best practices for each approach to minimize threats to validity and maximize the value of process evidence.

The overall market for derivative securities is often estimated as more than ten times the World’s GDP and many decry the complexity of derivatives as a main contributor to the subprime financial crisis. In this paper, we investigate whether and why complexity is used as a proxy for risk when evaluating derivatives. We conduct three laboratory experiments to show a robust effect consistent with investors and managers taking complexity as a proxy for risk and deeming more complex, but equally risky, derivatives as worse for risk minimization. We also provide evidence that the effect is driven by negative perceptions (i.e., affect) regarding derivatives. Altogether, our results shed light on how derivatives are evaluated and suggest an important potential source of misinterpretations of disclosures provided under US GAAP (ASC 815-10-50).

We use an experiment with experienced managers to provide more-direct evidence on how reporting goals and firm performance influence language choices. We find that bad news disclosures are less readable than good news, but only when managers have a stronger self-enhancement motive. Our results suggest that this difference is driven mainly by attempts to write more readable good news reports as opposed to intentional obfuscation of poor performance. In order to frame poor performance in a positive light, managers also focus more on the future, provide causal explanations for poor performance, and use more passive voice and fewer personal pronouns.

We test whether investors react more strongly to narrative disclosures when the CEO’s presence or association with the message is more salient in the disclosure, holding all other information constant. In our first experiment, we manipulate whether a CEO uses more personal pronouns (e.g., “I” and “our” rather than “the company” and “its”) in an assertion about whether the firm is “likely” or “unlikely” to win a lawsuit. We find investors’ beliefs about the outcome of the lawsuit align more closely with the CEO’s assertion when the disclosure contains more personal pronouns. Experiments 2 and 3 manipulate the extent of the CEO’s association with the message and whether the disclosure contains good or bad news. In the second experiment, we manipulate whether a disclosure uses more personal pronouns. In the third experiment, we manipulate whether a disclosure does or does not contain a photo of the CEO. Both manipulations of association with the message lead to stronger reactions from investors in between-subjects tests. That is, when news is good (bad), including either more personal pronouns or the CEO’s photo leads to more positive (negative) assessments of firm value. We also find that, within-subjects, both manipulations are perceived as indicating greater association with the message, but participants do not expect an effect on investment evaluations. In a fourth experiment, we provide additional evidence that personal pronoun usage affects investor reactions by increasing the perceived credibility of the disclosure.

Investments that are classified as Level 2 within the fair value hierarchy account for approximately 92 percent of US banks' fair value assets. We report an experiment that examines how experienced auditors apply current PCAOB guidance when auditing portfolios of these assets. We hypothesize and find that, depending on how overstatement is distributed within a portfolio, current PCAOB guidance leads auditors to make adjustments that are predictably larger or smaller than the aggregate overstatement in the portfolio. Auditors are more likely to follow PCAOB guidance when doing so leads to lower audit adjustments and higher client income. We also predict and find that auditors identify some patterns of overstatement as indicative of management bias, but not others. However, management-bias assessments do not affect auditors' adjustment decisions as standards imply they should, even when auditors are prompted to consider management bias. Together, these results highlight a potential deficiency in current auditing guidance that managers could exploit by strategically locating overstatements within securities with larger book values or by spreading those overstatements across many securities within a portfolio. We suggest changes to current PCAOB guidance which may reduce these effects.

In this paper, we examine the growing number of behavioral studies of how financial reporting, auditing, and other corporate governance regulations affect earnings management and accounting choice-related decisions of managers, auditors, and directors. We first describe how experimental and survey studies can add unique insights into our understanding of earnings management and accounting choice. We then organize our review of the literature by the type of regulation (financial reporting, auditing, or corporate governance) and secondarily by which of the three parties are affected. Finally, we point out useful directions for future research and discuss key methodological choices faced by those who will conduct that future research.

This paper independently replicates the results of the survey of experienced financial managers reported in section 4 of Libby and Rennekamp (2012). Using the same questions as Libby and Rennekamp (2012), we survey 110 experienced managers to examine their beliefs about the relationship between self-serving attribution bias, overconfidence, and management’s issuance of earnings forecasts. As in Libby and Rennekamp (2012), we find that participants agree that, in general, managers are overconfident. They also agree that other managers likely overestimate the extent to which they contribute to positive firm performance, and that both optimism about the firm’s future performance, as well as managers’ confidence in their ability to predict future performance, increase the likelihood that earnings forecasts are provided. Finally, participants agree that providing earnings forecasts amplifies the costs of negative outcomes and the benefits of positive outcomes, consistent with the experimental assumption in Libby and Rennekamp (2012). These findings suggest that the experimental findings in Libby and Rennekamp (2012) match experienced managers’ beliefs about how real-world earnings forecast decisions are made.

Prior research indicates that greater transparency in reporting formats facilitates the detection of earnings management. The current study hypothesizes and demonstrates that greater transparency in comprehensive income reporting also reduces the likelihood that managers will engage in earnings management in the area of increased transparency. In our experiment, 62 financial executives and chief executive officers decide which available-for-sale security to sell from a portfolio. We manipulate the transparency of comprehensive income reporting and the relationship of projected earnings to the consensus forecast in a 2×2 between-subjects design. When projected earnings are below (above) the consensus forecast, participants sell securities that increase (decrease) earnings. However, the rarely used, more transparent format for reporting comprehensive income significantly reduces both income-increasing and income-decreasing earnings management. Participants in the less transparent setting indicate that earnings management attempts will not be obvious to readers, will improve stock prices, and have no effect on management's reputation for reporting integrity. Conversely, respondents in the more transparent condition suggest that earnings management will be obvious to readers, harmful to stock prices, and damaging to reporting reputation. Results of this study suggest that more transparent reporting requirements will reduce earnings management in the area of increased transparency or change the focus of earnings management to less visible methods.

Investments that are classified as Level 2 within the fair value hierarchy account for approximately 92% of US banks’ fair value assets. We report an experiment that examines how experienced auditors apply current PCAOB guidance when auditing portfolios of these assets. We hypothesize and find that, depending on how overstatement is distributed within a portfolio, current PCAOB guidance leads auditors to make adjustments that are predictably too large or small relative to the aggregate overstatement in the portfolio. We also predict and find that auditors identify some patterns of overstatement as indicative of management bias, but not others. However, management-bias assessments do not affect auditors’ adjustment decisions as standards imply they should, even when auditors are prompted to consider management bias. Together, these results highlight a potential deficiency in current auditing guidance that managers could exploit by strategically locating overstatements within securities with larger book values or by spreading those overstatements across many securities within a portfolio. We suggest changes to current PCAOB guidance which should reduce these effects.

We examine whether information in footnotes might lack reliability because auditors permit more misstatement in disclosed, as opposed to recognized, amounts. In both the stock-compensation and lease settings, audit partners require greater correction of misstatements in recognized amounts than in the equivalent disclosed amounts. Debriefing questions indicate that the partners make these decisions knowingly, even though they expect greater client resistance to correcting recognized amounts, because they view recognized amounts as more material. Partners also spend more time on correction decisions for recognized information. While prior literature suggests that amounts are often relegated to footnotes because they are less reliable, our results suggest that the actual choice to disclose versus recognize can also reduce information reliability. These results have implications for the interpretation of prior research on the reliability of recognized and disclosed numbers, for financial-accounting standard setters who may want to consider the reliability effects of their recognition versus disclosure decisions, and for auditing standard setters who may wish to clarify auditors' responsibilities for preventing misstatements in disclosed amounts.

The above article has been retracted at the request of authors Robert Libby and Hun‐Tong Tan, in agreement with the Editor‐in‐Chief, Patricia C. O'Brien, the copyright holder, the Canadian Academic Accounting Association (CAAA), and Wiley Periodicals, Inc.Bentley University conducted an investigation confirming that Dr. J. E. Hunton, while a faculty member at Bentley University and without the knowledge of his co‐authors, engaged in research misconduct, specifically data fabrication. Dr. Hunton provided the data used in the above study, and did not respond to a request for comment. Based on Bentley's initial report and further investigation into details specific to this paper, we conclude that the validity of the data cannot be confirmed. To correct the academic literature and maintain standards of academic integrity, we therefore retract the paper.The article was published online in Contemporary Accounting Research on 13 May 2010, in Wiley Online (wileyonlinelibrary.com).ReferenceTan, H.‐T., R. Libby, and J. E. Hunton. 2010. . Contemporary Accounting Research 27 (): 187–208. doi: 10.1111/j.1911‐3846.2010.01006.x

We examine the effects of increased capital market pressure and disclosure frequency-induced earnings/cash flow conflict on myopic behavior. In our experiments, experienced financial managers choose between projects where a conflict exists between near-term earnings and total cash flow. Managers more often choose projects that they believe will maximize short-term earnings (and price) as opposed to total cash flows in response to increased capital market pressure resulting from a pending stock issuance, holding constant agency frictions and other stock market pressures. When faced with increased capital market pressure, changes in disclosure frequency cause managers to behave more or less myopically depending on the impact of the change on the pattern of earnings and the resulting earnings/cash flow conflict. Our study provides insights into managers' beliefs about stock market pressures, mandatory reporting, and the availability of alternative communications channels, and contributes to literature on managerial myopia and earnings management, as well as current debates over disclosure frequency.

This study examines how the form of management's earnings guidance (point, narrow range, wide range) affects analysts' earnings forecasts. Results from two experiments demonstrate that: (1) guidance form has no effect on analysts' forecasts made immediately after the guidance; and (2) after the actual earnings announcement, guidance form and the relationship of the earnings guidance to actual earnings (guidance error) interact in their effect on analysts' forecasts. After the actual earnings announcement, guidance error leads to higher (lower) analysts' forecasts for firms with downwardly (upwardly) biased guidance; this effect of guidance error is magnified by a narrow range and reduced by a wide range, compared to a point estimate. These results suggest that treating the mean of the range endpoints as equivalent to a point estimate and failing to consider effects after the release of actual earnings may paint an incomplete picture of how management guidance affects analysts and investors. It also offers useful information to managers who issue earnings guidance, and presents a challenge to the psychology literature regarding the effects of information precision on judgment and decision making.

We survey recent (mainly US) research on the effects of earnings presentation attributes on manager and user behavior. The literature we discuss relates to three primary earnings presentation attributes: (1) disaggregation (vertical and horizontal), (2) location (recognition vs. disclosure, which statement for recognized items, and within statement classification, labeling, and subtotals), and (3) narrative attributes (location of key amounts within narratives, readability, medium, and timing of disclosure). We show that disaggregation operates mainly by directly affecting information content. Location operates mainly by indirectly affecting information content through changes in managers' actions and by affecting ease of processing. Narrative presentation attributes operate mainly by affecting ease of processing. These differences in mechanisms determine the implications of the presentation attributes for contracting and valuation uses of accounting information. They also have implications for future research and standard setting.