ABSTRACT Voluntary assurance represents a costly signal of financial reporting quality in equity crowdfunding markets, yet investors often fail to distinguish audit from review reports. Drawing on signaling theory and construal-level theory, I argue that this reflects a fundamental cognitive processing problem: financial statement users default to a high-level construal mindset, focusing on abstract conclusions over engagement-level details. I posit that platform-embedded priming instructions prompting users to consider how the report’s conclusion was reached can solve this problem by inducing a low-level construal mindset, redirecting attention toward concrete details that carry the signal. In an experiment with 121 participants, only when primed did investors with an audit report show greater reliance and investment willingness than those with a review report. Perceived credibility also mediated this effect. These findings suggest that assurance signal effectiveness depends not only on disclosure content but on how information is cognitively framed through platform design.
We investigate the impact of board gender diversity and the IT expertise of female board members on data breaches. Using data from U.S. firms between 2005 and 2022, we find that gender diverse boards with female IT expertise are significantly less likely to experience data breaches. These results are robust to multiple endogeneity and sensitivity tests. Cross-sectional analysis reveals that the effect is driven by independent female board members, and requires a critical mass of female representation. Furthermore, auditors do not charge increased fees for breached firms when such board structures are present. The channel analysis suggests that gender-diverse boards with female IT expertise enhance cybersecurity investments, thereby mitigating cybersecurity risks. Overall, the findings shed new light regarding the oversight role of boards in cybersecurity risk management.
Social media platforms such as Twitter influence capital markets by rapidly disseminating information; yet, this environment is increasingly shaped by nonhuman bots. Building on theories of investor attention and information salience, we examine whether bots amplify market reactions to earnings news by directing attention toward larger surprises. Using machine learning to classify 12.02 million tweets discussing S&P 1,500 firms in 2018, we measure firm-specific abnormal bot activity and analyze its association with market responses to earnings announcements. We find that bot activity amplifies the relationship between earnings surprises and abnormal returns. Additional analyses reveal that this effect is stronger when bot sentiment is positive but diminishes with excessive positivity, varies by bot type, and is most pronounced for firms with fewer analysts, further supporting our investor attention argument. Our findings highlight bots' roles as "attention amplifiers" and underscore the need for greater scrutiny of algorithmic actors in financial markets.
As cybersecurity breaches against publicly traded firms are increasing, more attention is being paid to how firms disclose a breach. This experimental study examines how two factors influence nonprofessional investors' investment intentions and perceptions of the breached company: (1) corporate social responsibility (CSR) designation and (2) cybersecurity jargon included in the breach disclosure. Overall, our results demonstrate that a firm's CSR designation protects the perception of management following a cyberbreach. Mediation analysis indicates that this strong CSR designation maintains favorable management perceptions among investors, thereby fostering greater investment intentions than a company without a CSR designation. When a breached company does not have a CSR designation, we find that the level of jargon in the disclosure becomes relevant. Our study provides insights into the role of CSR in mitigating negative investor reaction to a cyberattack, as well as the use of jargon in a disclosure.
Retrieval-augmented generation (RAG) systems enhance large language models (LLMs) by integrating dynamic external data, enabling more contextually relevant and accurate outputs. As these systems gain traction in accounting applications (e.g., financial reporting, auditing, and tax), concerns emerge regarding data reliability, control oversight, and auditability. This study applies Design Science Research Methodology to develop a practical control framework tailored to the four core stages of a typical RAG system: External Data Management, User Interaction and Query Flow, Retrieval and Prompt Construction, and LLM Inference and Response. The proposed framework includes control objectives linked to stage-specific risks relevant to financial reporting. Iterative validation was conducted using an innovative multi-LLM consensus process to ensure coverage and reduce hallucination risk. This study contributes a framework for evaluating artificial intelligence (AI)-integrated systems, guidance for auditors and regulators to assess probabilistic AI outputs, extension of existing audit control frameworks, and instructional materials for accounting educators.
This study explores auditor adoption of GenAI and its impacts on audit quality through semistructured interviews with 37 Big 4 audit professionals. Results show that GenAI supports various audit tasks and enhances communication and skill development. However, overreliance without verification, lack of transparency, confidentiality and security risks, and inability to exercise skepticism and judgment can have an adverse impact on audit quality. We also note that junior and experienced auditors use GenAI differently. GenAI adoption for auditing is influenced by several factors including, the audit profession's risk aversion, GenAI experience, perceived usefulness, ease of use, perceiving GenAI as a job demand or resource, job insecurity due to GenAI, social influence, facilitating conditions, perceived risks, technology awareness, maturity, and attitudes. Auditing experience and time constraints for the audit engagement alter the adoption patterns. Our findings inform audit firms' strategic GenAI incorporation in auditing with a focus on improving audit quality.
This study examines auditors' perceptions of how task-specific artificial intelligence (AI) impacts the effectiveness and efficiency of auditing corporate disclosures, particularly management reports. Based on a survey of employees of a Big 4 audit firm in Germany, we analyze experiences with an AI tool designed to assist in detecting misstatements in management reports, e.g., by automatically identifying and matching disclosure requirements with reported content. The results indicate that the AI tool enhances audit effectiveness and efficiency although this result is less pronounced in supporting the detection of more complex qualitative issues. Perceptions of the AI tool's impact are shaped not only by its technological features but also by auditors' roles, expertise, and engagement with digital transformation. Auditors' perceptions of potential deskilling effects and level of trust in AI outputs also vary across these attributes, emphasizing the relevance of implementation strategies, training, and transparent communication when integrating AI into audit workflows.
Internal audit is increasingly expected to move from a retrospective "after the fact" stance to a proactive, risk-mitigating role. We introduce a predictive process-monitoring approach that lets internal auditors forecast late-payment risk while preserving the role separation mandated by the Three Lines Model. Using the 2018 BPI Challenge event log and a synthetic corporate invoice log, we (1) transform traces into deadline-centered, time-bucketed prefixes; (2) compare five classifiers; and (3) embed the resulting accuracy-timeliness profiles in a four-phase decision framework that aligns intervention timing with organizational context and governance guard rails. The study contributes by showing (1) a concrete use case for predictive monitoring, (2) statistical evidence clarifying the accuracy-timeliness tradeoff for audit use cases, and (3) a reusable, governance-anchored decision framework.
International Financial Reporting Standards (IFRS) adopt a principles-based approach, giving managers greater discretion in interpreting accounting standards and preparing financial statements. Given this flexibility, this study examines how the 2017 IFRS XBRL reporting mandate affects SEC monitoring costs and capital market outcomes for IFRS filers. We find that IFRS filers are less likely to receive SEC comment letters, suggesting reduced SEC monitoring costs due to lower information acquisition and processing costs. Additionally, we document that stock liquidity and intraperiod timeliness improve post-XBRL adoption. However, these benefits diminish for firms using more custom XBRL tags. Despite SEC guidelines advising IFRS filers to refer to the U.S. GAAP taxonomy when appropriate IFRS tags are unavailable, we find that IFRS filers rarely do so. Collectively, our study highlights how IFRS filers respond to the IFRS XBRL reporting mandate and the implications for regulatory monitoring and capital market outcomes.
In this study, I examine board attributes associated with firm transparency of governance over cybersecurity risk (GCR). Using a sample of firms that report cybersecurity as a material risk, I hand collect GCR data from 8,384 proxy statements filed from 2019 to 2021. Surprisingly, I find that only 57 percent of the filings provide stakeholders information about GCR. In multivariate analysis, I find that a firm's GCR disclosure is positively associated with board size, independence, information technology expertise, and those boards with a higher proportion of "busy" directors. I further find that boards with longer serving directors provide less GCR disclosure. Finally, I find increasing levels of GCR disclosure over the three-year period, suggesting that firms are responding to stakeholder demand for increased transparency. These findings should be helpful to firms looking to benchmark disclosure practices and to the SEC in evaluating the efficacy of existing cybersecurity disclosure guidance.
Financial reports, including 10-K and 10-Q filings, are a primary source of textual data in business disciplines. However, extracting specific sections from these lengthy documents remains a challenge. Custom code development by each research team to parse these files leads to redundancy, inefficiency, and inconsistencies and is especially challenging for teams lacking technical expertise. We address this by offering raw textual data from MD&A, risk factors, and business description sections, and financial statement notes, for all firms from 2008 onward. We share Python code to facilitate download and parsing. We also provide pre-calculated textual metrics, such as word counts, readability measures, and several bags of word metrics, including negative sentiment, forward-looking statements, and R&D. Additionally, we contribute two new word lists, COVID-19 and human capital, developed using a novel approach based on disclosure shocks. Our goal is to streamline research processes, ensure consistency, and enable further advances in the field.
This paper examines factors contributing to perceptions of success within small-and medium-sized accounting firms (SMAFs). Although numerous practitioner articles and podcasts discuss success factors, empirical evidence that controls for confounding effects and supports these determinants of success remains scarce. To address this gap, we survey 192 firms and identify being ahead in technology adoption as the most significant factor associated with success. Specifically, firms that demonstrate perceived leadership in leveraging technology outperform their peers across multiple data collections. Exceeding customer expectations also emerges as a contributing factor, albeit to a less robust extent. We also find some evidence that culture and organizational values are positively related to success but no evidence linking advisory services relative to compliance (i.e., assurance and tax) to firm success. These findings offer actionable insights for SMAFs, highlighting the strategic value of investing in accounting technology and maintaining a client-centric focus to drive firm performance.
Accounting firms are increasingly investing in information technology (IT) to enhance audit quality. This study examines the impact of IT personnel investment on audit quality, drawing on the resource-based view and human capital theory to argue that specialized IT expertise improves audit effectiveness. The results show that investment in IT personnel is associated with higher audit quality, as measured by lower discretionary accruals and fewer financial restatements. The positive effect is more pronounced in high-growth accounting firm offices, where resource constraints are more common, and becomes stronger following the STEM OPT extension policy, which expands opportunities for IT capability development in accounting firm offices. Furthermore, higher compensation for IT professionals appears to incentivize improved audit quality. The findings are robust across various measures and methods addressing potential endogeneity concerns. This study contributes to audit quality literature and provides practical insights as audit firms continue integrating AI into their processes.
Technological advancements have enabled client organizations to replace human labor with autonomous technology that processes transactions independent of direct human intervention. Theory suggests that autonomous technology introduces an unconscious bias that reduces the perception of human control over negative events when systems use more autonomous technology. In a between-participants experiment, we test whether jurors reduce their assessments of auditor control over an undetected misstatement in the presence of a more, compared to a less, autonomous client system. Although jurors should focus on audit evidence quality and not on the nature of the client's system, we find they do so only in the presence of a less autonomous client system and only when their implicit belief that "more technology is good" is lower. Overall, technology-based perceptions can change perceptions of auditor accountability for audit failures.
We examine the extent to which internal auditors' data analytics (DA) use is associated with improved financial reporting reliability and timeliness. Using data from a large survey among chief audit executives and data on public firms, we document a reduction in the likelihood of material weaknesses in internal controls over financial reporting and discretionary accruals at firms extensively using DA in internal audit tasks. Additionally, we find reductions in both the likelihood of restatements and the number of material weaknesses, as well as timelier earnings announcements. Next, we investigate and document a positive association between the use of several DA techniques in internal audit and financial reporting reliability and timeliness. Our findings inform researchers, practitioners, and standard setters by providing initial empirical evidence about the benefits of internal auditors' use of DA in terms of financial reporting quality.
Blockchain technology can reshape financial accounting operations by altering how transactional information is recorded, verified, and shared across organizations. Prior research has largely examined the technology's effects on operational efficiency, but less is known about its influence on accounting transaction costs. Using transaction cost theory, we examine how blockchain affordances affect search, bargaining, and monitoring costs in financial accounting operations. We began by drawing on the accounting literature on blockchain technology to identify four affordances associated with this technology: decentralization, automation, confidentiality, and data integrity. Next, we conducted semistructured interviews with financial industry professionals and blockchain experts and analyzed their responses based on transaction cost determinants of bounded rationality, asset specificity, uncertainty, opportunism, and transaction frequency. Our results develop a conceptual model relating the identified four blockchain technology affordances to transaction costs. The results further indicate that blockchain technology's organizational effects depend on contextual alignment rather than inherent technological capabilities.
Experimental studies reveal that speakers' vocal and facial expressions convey additional insights beyond the transcript. Building on these findings, we examine how verbal and nonverbal features extracted from the forward-looking part of CEO interviews correlate with a firm's future performance. Our results show that nonverbal cues, particularly negative facial expressions, dominate in signaling a firm's future performance. Particularly, negative vocal and facial cues collectively diminish the positive sentiment conveyed by verbal content. Furthermore, we examine the determinants of CEOs' communication styles presented in interviews and find that male CEOs display fewer happy expressions than female CEOs, whereas those with longer tenures, dual roles, Ivy League backgrounds, or broader networks exhibit greater pessimism or sadness during forward-looking discussions. Finally, we find that managers' forward-looking statements remain largely short-term in focus (i.e., next quarter), even in fourth-quarter interviews, when they likely possess more private information about longer term prospects.
Our study draws on self-organization theory to develop a framework for explaining financial executives' shadow information technology (IT) behavior (i.e., executive adoption of a technology without the IT department's knowledge). Understanding executive shadow IT usage is important, because it can significantly increase business risk related to loss of control and information security. Although directly involving shadow IT, our research also attempts to clarify mixed results in the broader usage policy literature. We investigate shadow IT usage via an experiment and separate interviews with financial executives from Germany and Italy, representing autonomous and autogenic environments, respectively. Results indicate higher shadow IT propensity in autonomous contexts and stronger adherence to stringent IT usage policies in autogenic contexts. Interview data further suggest that internal control strength and executive alignment with the IT department are key factors shaping shadow IT behavior. Unique findings add an international component to shadow IT and usage policy literatures.
Extracting specific items from 10-K filings is challenging because of variations in document formats and item presentation. This study aims to improve traditional rule-based approaches by introducing and comparing two advanced item segmentation methods: (1) GPT4ItemSeg, employing a novel line-ID-based prompting mechanism to utilize a large language model, ChatGPT-4o, for item segmentation, and (2) BERT4ItemSeg, combining a pre-trained language model, BERT, with a Bi-LSTM model in a hierarchical structure to overcome context window constraints. Trained and evaluated on 3,737 annotated 10-K reports, BERT4ItemSeg achieves a macro-F1 of 0.9825, surpassing GPT4ItemSeg (0.9567), conditional random field (0.9818), and rule-based methods (0.9048) for core items (1, 1A, 3, and 7). These approaches enhance item segmentation performance, improving text analytics in accounting and finance. BERT4ItemSeg offers satisfactory item segmentation performance, whereas GPT4ItemSeg can easily adapt to regulatory changes. Together, they provide an extensible framework for 10-K item segmentation that supports reliable and reproducible results.
Email is a prevalent communication medium between auditors and clients for obtaining evidence and performing inquiry. Client email responses can contain affective statements that increase likability. These responses may contain mixed evidence, meaning some client responses align with audit evidence obtained and others do not. We investigate how auditors are influenced by positive affect in a mixed evidence multiple email exchange setting. In Experiment 1, results indicate that auditors judge clients as more reliable and assess a lower risk of material misstatement when emails contain positive affective cues. Interestingly, positive affect is most influential when clients provide evidence-inconsistent responses. In Experiment 2, positive affective statements do not influence auditors when the client's first response is inconsistent. This suggests that affect primarily influences auditor judgment if a positive "first impression" is established. These results establish a boundary condition of positive affect and show when client positive affective cues are more influential.
