Technological Educational Institution of Thessaloniki

Operational Research, 2008

The present study investigates the efficiency of four classification techniques, namely discriminant analysis, logit analysis, UTADIS multicriteria decision aid, and nearest neighbours, in the development of classification models that could assist auditors during the examination of Asian commercial banks. To develop the auditing models and examine their classification ability, the dataset is split into two distinct samples. The training sample consists of 1,701 unqualified financial statements and 146 ones that received a qualified opinion over the period 1996–2001. The models are tested in a holdout sample of 527 unqualified financial statements and 52 ones that received a qualified opinion over the period 2002–2004. The results show that the developed auditing models can discriminate between financial statements that should receive qualified opinions from the ones that should receive unqualified opinions with an out-of-sample accuracy around 60%. The highest classification accuracy is achieved by UTADIS, followed by logit analysis, nearest neighbours and discriminant analysis. Both financial variables and the environment in which banks operate appear to be important factors.

Journal of Business Ethics, 2017

In recent years, after various scandals, the role of auditors has been called into question, even casting doubt on whether their reports reliably reflect the true financial situation of the auditee, especially when this situation is not good. Normative changes in the way auditors have to rate certain questions provide a good opportunity to study this problem. These changes have acquired great relevance among the factors involved in studying audit quality. Thus, the present study analyzed the effect of the normative change that took place in Spain in December 2010, related to opinions modified for going-concern uncertainties. Until that date, the auditor's uncertainty about the company's going-concern status led to a qualified opinion. However, under the new regulation, it became an opinion that included an explanatory paragraph stating the reasons for concern, which was considered less serious. In all, 152 small-and medium-sized enterprises that had begun bankruptcy proceedings were studied. Expert systems were used for their analysis, based on classification trees assembled through boosting and bagging. In addition, the logistic regression was used as baseline to compare previous methods. The main result obtained was that a change in the norm that catalogs the going-concern issue as less serious made auditors more likely to report this situation, thus questioning the audit quality.

Our study evaluates a multiple criteria linear programming (MCLP) and other data mining approaches to predict auditor changes using a portfolio of financial statement measures to capture financial distress. The results of the MCLP approach and the other data mining approaches show that these methods perform reasonably well to predict auditor changes using financial distress variables. Overall accuracy rates are more than 60 percent, and true positive rates exceed 80 percent. Our study is designed to establish a starting point for auditor-change prediction using financial distress variables. Further research should incorporate additional explanatory variables and a longer study period to improve prediction rates.

This paper explores the application of data mining techniques to fraud detection in the audit of financial statements and proposes a taxonomy to support and guide future research. Currently, the application of data mining to auditing is at an early stage of development and researchers take a scatter-shot approach, investigating patterns in financial statement disclosures, text in annual reports and MD&As, and the nature of journal entries without appropriate guidance being drawn from lessons in known fraud patterns. To develop structure to research in data mining, we create a taxonomy that combines research on patterns of observed fraud schemes with an appreciation of areas that benefit from productive application of data mining. We encapsulate traditional views of data mining that operates primarily on quantitative data, such as financial statement and journal entry data. In addition, we draw on other forms of data mining, notably text and email mining. . 5). Data mining involves acquisition, loading and integration of data; application of specialist data mining tools and, finally, human interpretation of the discovered meaning. 2 The decision to incorporate data mining into financial audits is both a firm-level decision for accounting firms and an engagement-level decision. Firm-level decisions preclude engagement-level decisions in that if firm management does not see a beneficial reason to invest resources in software, infrastructure, training, and staffing then data mining will likely not be a cost-effective option for engagement teams. Larger accounting firms and some specialist providers offer a variety of data mining services. Currently, data mining is used in specialized audits (e.g., fraud audits or forensic audits) by expert staff in the professional services firms; however, data mining is seldom used in financial statement audits. When used, it is for identified high-risk clients by the firm's data mining specialists. The majority of this paper is focused on engagement-level data mining activities; however, we revisit firm-level issues and research opportunities in the concluding parts of the paper.

Jurnal Mantik

Auditors who have appropriate competence are needed to carry out certification. The amount of experience a person has before becoming an Auditor determines the competence possessed by an Auditor. To be assigned to carry out certification, an auditor needs special competencies such as areas of expertise, work experience, and the Schemes and Standards that become the reference for certification. Auditors owned by LSUP need to be classified so that they can be assigned according to their competence. The C4.5 Decision tree algorithm is one of the methods that can be used to create a model used to classify and predict auditor competence. 65 data from 221 data are used as training data to form a model in the form of a tree structure. This model consists of 3 attributes that form a Node in a tree structure, namely the Workplace, Skill and Standard attributes. Each attribute has a value that forms a branch. The model was evaluated using the Confusion Matrix with a matrix size of 7x7 which r...

Indonesian Journal of Contemporary Accounting Research

This research work examines the factors that determine the selection of external auditor among Global Brand listed companies in Nigeria. The variables of Audit Big 4 firms which include Deloitte, Klynveld Peat Marwick Goerdeler, (KPMG) and Price Water House Coopers (PWC) where selected as choice target for this research work. Big four audit firm of Ernst’s & Young (E&Y) was excluded in the sample as it is found not to have provided audit services to any of the selected global brand listed company in Nigeria during the period covered by this study. The study employed the methodology of Multinomial Probit Regression (MPR) analysis in X-raying the data. Empirical evidence shows that there is a significant positive relationship between firm size and choice of external auditor’s selection for the big four audit firms of KPMG and Deloitte. The study also provides a revelation that the relationship between the variable of firm size is negative and statistically significant with audit firm ...

SSRN Electronic Journal, 2012

Nowadays data mining, which is used in various accounting and financial applications, has received a great deal of attention. One of these applications is predicting and identifying the audit opinion type. The objective of research is to help auditors identify audit opinions by using a support vector machine from data mining methods. The system receives the data from financial reports and identifies the type of audit opinions. This approach combines support vector machine with a decision tree that can understand and interpret the obtained results. In this paper, a novel approach for rule extraction from support vector machine and decision tree is presented and its application is shown in the prediction of audit opinions. The research result is 30 rules that predict the audit opinions.

Journal of Artificial Intelligence and Data Mining

Abstract Predicting financially false statements to detect frauds in companies has an increasing trend in recent studies. The manipulations in financial statements can be discovered by auditors when related financial records and indicators are analyzed in depth together with the experience of auditors in order to create knowledge to develop a decision support system to classify firms. Auditors may annotate the firms’ statements as “correct” or “incorrect” to add their experience, and then these annotations with related indicators can be used for the learning process to generate a model. Once the model is learned and tested for validation, it can be used for new firms to predict their class values. In this research, we attempted to reveal this benefit in the framework of Turkish firms. In this regard, the study aims at classifying financially correct and false statements of Turkish firms listed on Borsa İstanbul, using their particular financial ratios as indicators of a success or a manipulation. The dataset was selected from a particular period after the crisis (2009 to 2013). Commonly used three classification methods in data mining were employed for the classification: decision tree, logistic regression, and artificial neural network, respectively. According to the results, although all three methods are performed well, the latter had the best performance, and it outperforms other two classical methods. The common ground of the selected methods is that they pointed out the Z-score as the first distinctive indicator for classifying financial statements under consideration.

Intelligent Systems in …, 2007

This study investigates the efficiency of k-nearest neighbours (k-NN) in developing models for estimating auditors' opinions, as opposed to models developed with discriminant and logit analyses. The sample consists of 5276 financial statements, out of which 980 received a qualified audit opinion, obtained from 1455 private and public UK companies operating in the manufacturing and trade sectors. We develop two industry-specific models and a general one using data from the period 1998-2001, which are then tested over the period 2002-2003. In each case, two versions of the models are developed. The first includes only financial variables. The second includes both financial and non-financial variables. The results indicate that the inclusion of credit rating in the models results in a considerable increase both in terms of goodness of fit and classification accuracies. The comparison of the methods reveals that the k-NN models can be more efficient, in terms of average classification accuracy, than the discriminant and logit models. Finally, the results are mixed concerning the development of industryspecific models, as opposed to general models.

The tax administration is required to audit some or all its taxpayers to check the evasion of tax and ensure compliance. Conducting of audits involves costs to the tax department as well as to the taxpayer. Thus, audit is not a very welcome procedure both for the taxpayers as well as the economy. Tax administration agencies must therefore use their limited resources very judiciously to achieve maximal taxpayer compliance, minimum intrusion and minimum costs. This paper analyzes the use of data mining algorithms as the best cost effective option to make audit selection more efficient and effective.