• ISSN: 2010-0248 (Print)
    • Abbreviated Title: Int. J. Innov.  Manag. Technol.
    • Frequency: Quarterly
    • DOI: 10.18178/IJIMT
    • Editor-in-Chief: Prof. Jin Wang
    • Managing Editor: Ms. Nancy Y. Liu
    • Abstracting/ Indexing: Google Scholar, CNKI, Ulrich's Periodicals Directory,  Crossref, Electronic Journals Library.
    • E-mail: ijimt@ejournal.net
IJIMT 2017 Vol.8(3): 172-179 ISSN: 2010-0248
doi: 10.18178/ijimt.2017.8.3.723

Financial Statements Earnings Manipulation Detection Using a Layer of Machine Learning

B. Dbouk and I. Zaarour

Abstract— This paper applies a layer of machine learning technique such as the Bayesian Naïve Classifier (BNC) to enhance the decision making process in the framework of Earnings Manipulation Detection (EMD). It evaluates and competes Manual Auditors’ Methods versus a mathematical model in EMD such as the Beneish Model. The Data sets consist of fifty-three (53) Financial Statements acquired from largest corporations over four consecutive years. Using the Beneish model, we classified corporations between manipulators and non-manipulators to establish the training set. The manual audit results for each corporation used to establish a test set as the expert set. In testing for EMD under the mathematical model versus the audit methods, and to evaluate results, a new layer of Machine Learning technique introduced such as the BNC. Our results show that mathematical models outperform auditors. They reveal a classification rate of (86.84 per cent) using the Beneish Model and (60.53 per cent) using Manual Auditors’ Methods. Our findings indicate that Manual Auditors’ Methods are difficult to detect Earnings Manipulation of Financial Statements. The Main contribution of this research is to use the Machine Learning as a new layer in the Framework of EMD. This approach broadens the scope for auditors, and other financial experts to use Machine Learning with mathematical models through their audit. The results of this study will help regulators and practitioners to detect accounting manipulations and to add value for the auditing, accounting, and financial professions.

Index Terms— Beneish model, earnings manipulation, machine learning, supervised classification.

Bilal Dbouk is with the Doctoral School of Law, Political, Economic and Administrative Sciences, The Lebanese University (LU), Beirut, Lebanon (e-mail: bilal_dbook@hotmail.com). Iyad Zaarour is with the Faculty of Economics and Business Administration, The Lebanese University (LU), Beirut, Lebanon.


Cite: B. Dbouk and I. Zaarour, " Financial Statements Earnings Manipulation Detection Using a Layer of Machine Learning," International Journal of Innovation, Management and Technology vol. 8, no. 3, pp. 172-179, 2017.

Copyright © 2010-2024. International Journal of Innovation, Management and Technology. All rights reserved.
E-mail: ijimt@ejournal.net
Published by International Association of Computer Science and Information Technology (IACSIT Press)