The Application of Neural Networks and a Qualitative Response Model to the Auditor's Going Concern Uncertainty Decision*

Abstract

An auditor gives a going concern uncertainty opinion when the client company is at risk of failure or exhibits other signs of distress that threaten its ability to continue as a going concern. The decision to issue a going concern opinion is an unstructured task that requires the use of the auditor's judgment. In cases where judgment is required, the auditor may benefit from the use of statistical analysis or other forms of decision models to support the final decision. This study uses the generalized reduced gmhent (GRG2) optimizer for neural network learning, a backpropagation neural network, and a logit model to predict which firms would receive audit reports reflecting a going concern uncertainty modification. The GRG2 optimizer has previously been used as a more efficient optimizer for solving business problems. The neural network model formulated using GRG2 has the highest prediction accuracy of 95 percent. It performs best when tested with a small number of variables on a group of data sets, each containing 70 observations. While the logit procedure fails to converge when using our eight variable model, the GRG2 based neural network analysis provides consistent results using either eight or four variable models. The GRG2 based neural network is proposed as a robust alternative model for auditors to support their assessment of going concern uncertainty affecting the client company. Subject Areas: Auditing, Decision Processes, and Decision Support Systems.*The authors are thankful to Professor Ming Hung for his suggestions on the use of the GRG2 optimizer, and to the anonymous reviewers for their useful comments and suggestions.

Artificial neural networks in bankruptcy prediction: General framework and cross-validation analysis

“…Hence, the correct classi®cation rates in Table 4 for the large test set are derived directly from the results for both small test sample and training sample. For example, for training sample 1, the total number of correctly classi®ed ®rms in the large test set is 191 which is equal to the best small test result (35) plus the corresponding training result (156).…”

“…Recently, Hung and Denton [27] and Subramanian and Hung [59] have proposed to use a general-purpose nonlinear optimizer, GRG2, in training neural networks. The bene®ts of GRG2 have been reported in the literature for many classi®cation problems [35,42,59]. This study uses a GRG2 based system to train neural networks.…”

“…However, comparing the Madaline with the discriminant analysis model shows no signi®cant advantage of one over the other. Lenard et al [35] ®rst apply the generalized reduced gradient (GRG2) optimizer for neural network training in an auditor's going concern assessment decision model. Using GRG2 trained neural networks results in better performance in terms of classi®cation rates than using backpropagation-based networks.…”

Comparision of some models for estimation of reflectance of hypericum leaves under stress conditions

“…ANN is a computational modeling tool that used to solve many complex real world problems due to its remarkable learning and generalization capabilities [27,28]. ANN is being used in water quality and water resources to estimate evaporation, evapotranspiration, rainfall, runoff, and nutrient transportation [29,30], accounting and fi nance [31], health and medicine [32,33], engineering and manufacturing [34,35], and marketing [36,37]. The ANN and MLR are used to predict the empirical relationships between refl ectance and plant stress levels of hypericum leaves under stress conditions using multiple regression, artifi cial neural network and ANFIS models.…”

Audit‐firm group appointment: an artificial intelligence approach

“…In the fi eld of auditing, relevant studies have only recently employed AI techniques, such as NNs (e.g. Lenard et al, 1995;Gaganis et al, 2007a), support vector machines (Doumpos et al, 2005), nearest neighbours (Gaganis et al, 2007b), DTs and Bayesian belief networks (Kirkos et al, 2007a,b).…”