Investigating the Performance of Regression and Deep Learning Approaches to Detect Financial Statement Fraud, Focusing on Pressure/Motivation and Opportunity Dimensions

Articles in Press

Document Type : Research Paper

Authors

1 Ph.D. student, Department of Accounting, South Tehran Branch, Islamic Azad University, Tehran, Iran.

2 Professor, Department of Accounting, South Tehran Branch, Islamic Azad University, Tehran, Iran.

3 Associate Professor, Department of Accounting, South Tehran Branch, Islamic Azad University, Tehran, Iran.

10.22051/jera.2025.47003.3261

Abstract

Purpose: The purpose of this research is to identify the factors affecting the performance of internal audit and to investigate the performance of regression and deep learning methods to detect fraud in financial statements, focusing on the dimensions of pressure/motivation and opportunity.

Method: The spatial domain of this research is the companies accepted in the Tehran Stock Exchange and the temporal domain is the years between 1391 and 1400. In order to collect the data needed for this research, the library method was used. Regression method was used in the investigation of factors affecting internal audit performance. Deep learning and feedforward neural network were also used in fraud detection.

Findings: The results showed that deep learning and feedforward neural network models performed better in simulating and predicting financial statement fraud compared to regression methods. In particular, deep learning was able to identify a better relationship between pressure/motivation dimensions and opportunity dimensions and showed higher performance than regression models. In this regard, these methods were able to simulate more hidden features in the data and identify more complex factors that traditional models could not.

Conclusion: The current research has specifically addressed the value of using deep learning techniques in identifying fraud in financial statements and has shown that these techniques can simulate and identify more complex dimensions of pressure/motivation and opportunities that were less noticed before. do This research has helped to fill the gap in traditional fraud prediction models and has also taken an effective step in identifying the key factors that affect internal audit performance. Thus, the use of deep learning can significantly increase the accuracy and performance of financial statement fraud prediction methods.

Keywords

Main Subjects

Empirical Research in Accounting

Articles in Press, Accepted Manuscript
Available Online from 05 January 2025

Files

History

  • Receive Date: 17 June 2024
  • Revise Date: 25 November 2024
  • Accept Date: 05 January 2025

Share

How to cite

Statistics