Banca de DEFESA: Elmo Dias da Silveira

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : Elmo Dias da Silveira
DATE: 05/12/2023
TIME: 10:30
LOCAL: Ambiente Virtual Teams (ID: 285 299 009 656 Senha: cAsGjs / Link: https://abrir.link/1uZ6e)
TITLE:

Use of artificial intelligence as a way to improve the performance of traditional earnings management detection models

KEY WORDS:

Artificial intelligence; Machine Learning; Earnings Management; Accruals; Artificial neural networks.

PAGES: 87
BIG AREA: Ciências Sociais Aplicadas
AREA: Administração
SUBÁREA: Ciências Contábeis
SUMMARY:

This thesis constitutes a comprehensive investigation into earnings management, focusing on the assessment and enhancement of traditional models through innovative approaches. The first article, a theoretical essay, highlights Artificial Neural Networks (ANN) as a promising tool to address econometric issues associated with earnings management models. The research addresses concerns related to the lack of direct observability of management's discretionary accruals and argues that the application of ANN can significantly improve the accuracy and specificity of these models. The second empirical article focuses on the reversal of distortions in accruals and their application in traditional earnings management models in Brazilian publicly traded companies. The results reveal the limitations of accrual-based models, highlighting the complexity of non-discretionary accruals and the challenges associated with the accurate measurement of earnings management. The third empirical article explores the use of machine learning techniques in earnings management detection. The study overcomes the deficiencies of traditional models, especially regarding the direct measurement of earnings management. The results indicate that machine learning algorithms, such as the Decision Trees Classifier, offer a viable solution to significantly increase the explanatory power of these models. Overall, this thesis contributes to the understanding and improvement of earnings management, emphasizing the importance of innovative approaches such as ANN and machine learning in addressing persistent challenges in traditional models. These findings promote better integration between accounting and information technology, paving the way for future research exploring a variety of artificial intelligence techniques in accounting information quality.

COMMITTEE MEMBERS:
Externo à Instituição - LUIZ PAULO LOPES FÁVERO - USP
Externo à Instituição - FABIO MORAES DA COSTA - FUCAPE
Interno - 2571490 - MARCELO DRIEMEYER WILBERT
Presidente - 2221870 - PAULO AUGUSTO PETTENUZZO DE BRITTO
Interno - 1554042 - RODRIGO DE SOUZA GONCALVES