International Journal of Accounting Information Systems最新文献

We develop an analytical model intended as the first stage in the development of expert systems to improve auditor knowledge in, and assist in the decision process of, Going Concern Opinions (“GCOs”). Our approach is consistent with a design science approach to developing information systems, resulting in an initial artifact, the mathematical model, which can, through iterative design science and behavioral research, inform a technology-based expert system. Based on Bayesian networks, our model provides insights about auditors’ revision, or inflation, of the probability to issue a GCO based on the interrelationship that forms with the incremental existence of one, two, or three publicly observable financial statement risk factors – net operating loss, negative cash flows from operations, and negative working capital. We calculate the revised probabilities using empirical data of GCOs from 2004 to 2015. Results reveal that the incremental relationship (one, two, or three factors present) effectively models expert auditors’ decisions to issue a GCO, and suggests the existence of these measurable inflation factors that represent situational and auditor-specific factors. We also find that Non-Big Four auditors inflate these factors differently than Big Four auditors to arrive at a decision to issue a GCO.

Reviewing literature is a demanding task that calls for effective tools. After examining literature review (LR) practices, this paper develops an innovative approach in representing literature. In particular, the paper introduces depictions based on a three-dimensional structure that enables the generation of unitary and highly informative literature representations, thereby improving the completeness and reducing the mismatch between field complexity and the traditional figures and tables of LRs. The viability of the designed solution is shown by replicating a published LR on a representative topic of the accounting information systems (AIS), namely a literature review on enterprise resource planning systems in SMEs. Its benefits arise from comparing the literature representation developed here with the original figures. The paper has methodological and practical implications. Indeed, it proposes a structured approach to improve the representation of multifaceted literature, but it can also be suitable for other types of analysis. The efficiency and efficacy benefits that emerged from the evaluation activity make the output of the paper useful both for researchers and academic research users. In the light of developments in AIS research, the article also comments on potential applications in this area by examining some recently published LRs.

Machine Learning (ML) applied to Robotic Process Automation (RPA) and chatbot interfaces can generate significant value for many business processes. However, these technologies generate the intended return only with a carefully planned deployment. Current literature only contains a small number of case studies about how the adoption of ML-based automation services impacts employees’ behavior. In particular, no case studies look into the automation of manual tasks related to accounting management. This article reports a study conducted to understand users’ perceptions of an ML-enabled service to automate repetitive management tasks. The service was developed in a partnership between Unisinos University and Dell Inc. The study was conducted with a group of ten highly skilled employees from Dell with expertise in accounting processes and with IT background that frequently would use the automation service. The group participated in a presentation about the service and its interface and voluntarily answered a Technology Acceptance Model (TAM) questionnaire to evaluate the usability and ease of use. Results show that 10 out of 10 users agree that the service was easy to use. Also, 8 of them agree that its output is useful to reduce the manual labor required for statutory reconciliation. Furthermore, employees with an accounting management background were given access to the service, and three voluntarily answered an open-ended survey. In summary, employees agree that an automation service can reduce the time required to conduct management tasks but questioned the long-term usefulness and the ability to incorporate the process’s particularities. These results provided insights leading to ten lessons related to user experience, training and awareness, and service development.

This paper is a response to Weber’s (2020) call for further debate on the (potential) contribution of agential realism for the understanding of the content and functioning of accounting information systems (AIS) (Weber, 2020). Contrary to Weber’s conclusions, we suggest that agential realism can make important contributions to AIS studies. In order to realize such contributions we have to acknowledge that agential realism is a metaphysical framework rather than a theory or epistemology. Its potential contributions should be set against the potential contributions stemming from the metaphysical framework that is dominant in AIS research: representationalism. In representationalism, for the purpose of creating a knowledge base for a distant ‘knower’ who acts on the basis of such knowledge, accounting information systems are assumed to represent or provide information about a reality that is ‘out there’. In agential realism, accounting is assumed to be performative in (re)configuring local and temporal boundaries between meaningful positions for humans and non-humans. Rather than putting the human subject (the knower) at the centre of the stage, an agential realist account of accounting foregrounds accounting practices in their interrelation with other practices. It acknowledges that accounting participates in the (re)configuring of a sociomaterial world.

The first-two digits ExcessMAD test was created in 2016, allowing to evaluate whether a certain data set conforms to Benford’s Law (BL). The purpose of this study is to explore some questions that remained open: develop the exact and approximate mathematical formulation of the first and second digit ExcessMAD test and study the type I error of these tests when applied to different sample sizes conforming to BL and to the uniform distribution, due to its wide use in accounting data. The importance of this study is to make available to accountants, auditors and researchers the first and second digit ExcessMAD tests, which will make it possible to conduct further investigations involving BL, especially for smaller samples. In addition, the relevance of the type I error analysis stems from the reduction of unnecessary additional studies for the investigation of non-conformity, in the case of the erroneous rejection of the null hypothesis stated as conforming to BL. The application of the second digit ExcessMAD test in the uniform distribution reveals that the close proximity between the uniform and BL distributions can lead to misinterpretations. Based on the exact and approximate mathematical formulations of the three ExcessMAD tests and the use of the Monte Carlo simulation technique, samples were generated in accordance with the BL and uniform distributions, with sizes between 100 and 3,500 elements, which allowed the study of type I error and the comparison of the tests applied to those distributions. This paper seeks to cover three gaps in the literature on ExcessMAD tests. In the previous studies, the following approaches were not found: the exact and approximate mathematical formulation of the first and second digit ExcessMAD tests; the analysis of type I error in these tests and the comparison of such results in the BL and Uniform distributions.

This paper surveys the published work on how blockchain technology will impact accounting in general, but AI-enabled auditing specifically. The purpose is to investigate how blockchain technology can improve transparency and trust in accounting practice and how professionals can use blockchain data to improve decision-making, based on the qualities of immutability, append-only, shared, verified, and agreed-upon (i.e., consensus-driven) blockchain data. The multi-party validation of blockchain protocols adds real-time trusted data for the AI systems used by auditors to improve assurance and efficiency. This review summarizes four themes emerging from the literature focusing on how blockchain technology has changed record-keeping in accounting: event approach to accounting; real-time accounting; triple entry-accounting and continuous auditing. The research interprets the findings using agency theory and stakeholder theory to advance how using blockchain to mitigate information asymmetry and improve stakeholder collaborations is understood. The investigation also summarizes the challenges and clarifies organizations’ reasons to be cautious about adopting blockchain. Lastly, the study suggests that future researchers use this study in two ways that enrich blockchain literature: first, to apply the themes and answer the questions identified within this review to improve the business methods of practitioners and policymakers; and second, to encourage stakeholders such as practitioners, system designers/developers, and policymakers to collaborate in designing blockchain ecosystems that suit accounting and auditing as they transform digitally.

Over the last decade, the use of different artificial intelligence (AI) tools has increased. To shed some light on the emerging trend of AI disclosure, the aim of this paper is to analyse the current practices of major Western European companies regarding the automated decision-making (ADM) disclosure in their annual or sustainability reports. This paper proposes a methodology based on bigrams that enables the automatic extraction of the information on ADM that companies disclose. The sample consisted of 962 annual/sustainability reports, published in 2018 and 2019, of 337 companies listed on 13 Western European countries’ stock markets. Our findings show that ADM disclosure is still at an early stage and that the first adopters are mostly companies operating in the financial sector.

We investigate the feasibility of machine learning methods for attributional content and framing analysis in corporate reporting. We test the performance of five widely-used supervised machine learning classifiers (naïve Bayes, logistic regression, support vector machines, random forests, decision trees) in a top-down three-level hierarchical setting to (1) identify performance-related statements; (2) detect attributions in these; and (3) classify the content of the attributional statements. The training set comprises manually coded statements from a corpus of management commentary reports of listed companies. The attributions include both intra- and inter-sentential attributional statements. The results show that for both intra- and inter-sentential attributions, F1-scores of our most accurate classifier (i.e., support vector machines) vary in the range of 76% up to 94%, depending on the identification, detection and classification levels and the content characteristics of attributions. Additionally, we assess the hierarchical performance of classifiers, providing insights into a more holistic classification process for attributional statements. Overall, our results show how machine learning methods may facilitate narrative disclosure analysis by providing a more efficient way to detect and classify performance-related attributional statements. Our findings contribute to the accounting and management literature by providing a basis for implementing machine learning methodologies for research investigating attributional behavior and related impression management.

We examine the effects of information technology material weaknesses on a firm’s reputation by examining how management’s actions before and after disclosure influence investors’ trust in management and perceptions of investment risk. Specifically, we look at the influence of: 1) management taking responsibility for an information technology material weakness, and 2) replacing the CFO with someone with technology expertise. We find that management taking responsibility for a material weakness does not lead to increased trust in management before or after remediation. However, investors perceive more favorable market reactions to remediation when management had previously taken responsibility for the control weakness. Further, we find that replacing the CFO with someone who has technology expertise results in increases in investor trust and improvements in perceptions of investment risk after control weakness remediation. This suggests the importance of sending clear signals to investors that the company is hiring managers with appropriate technology expertise.

This study demonstrates a way of bringing an innovative data source, social media information, to the government accounting information systems to support accountability to stakeholders and managerial decision-making. Future accounting and auditing processes will heavily rely on multiple forms of exogenous data. As an example of the techniques that could be used to generate this needed information, the study applies text mining techniques and machine learning algorithms to Twitter data. The information is developed as an alternative performance measure for NYC street cleanliness. It utilizes Naïve Bayes, Random Forest, and XGBoost to classify the tweets, illustrates how to use the sampling method to solve the imbalanced class distribution issue, and uses VADER sentiment to derive the public opinion about street cleanliness. This study also extends the research to another social media platform, Facebook, and finds that the incremental value is different between the two social media platforms. This data can then be linked to government accounting information systems to evaluate costs and provide a better understanding of the efficiency and effectiveness of operations.