Previsão de estresse fiscal em municípios de Minas Gerais: um olhar especial para o Triângulo Mineiro e Alto Paranaíba

• Ano de defesa: 2023
• Autor(a) principal: Garruti, Daniel Vitor Tartari
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso embargado
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Administração
• Palavras-chave em Português: Condição Financeira Governamental; Government Financial Condition; Revisão Sistemática; Systematic Review; ProKnow-C; ProkNow-C; Inteligência Artificial; Artificial Inteligence; Condição Fiscal; Fiscal Condition; CNPQ::CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO::ADMINISTRACAO PUBLICA::CONTABILIDADE E FINANCAS PUBLICAS; Administração; Finanças regionais - Triângulo Mineiro/Alto Paranaíba (MG : Mesorregião); Política governamental - Triângulo Mineiro/Alto Paranaíba (MG : Mesorregião); Administração financeira - Triângulo Mineiro/Alto Paranaíba (MG : Mesorregião)
• Link de acesso: https://repositorio.ufu.br/handle/123456789/37238; http://doi.org/10.14393/ufu.di.2023.7006

During periods of financial difficulties, Local Governments tend to suffer impacts. In view of this, groups in the United States of America and the academy began to be interested in the subject in the last century, developing the Government Financial Condition. In Brazil, the topic is still recent and began to be debated after the difficulties faced in 2016 and in previous years, such as the negative GDP. Given this scenario, the objective of this paper is to adapt the model by Trussel & Patrick (2018) for predicting fiscal stress in municipalities in Minas Gerais, using Random Forest, in order to verify regional potentials of the Mesoregions of the Triângulo Mineiro (Mina’s Triangle) and Alto Paranaíba (Paranaíba Highlands) compared to others that differ in results. For this, Brazilian studies were considered, as well as the factors by Groves et al. (1981). This is the first research that makes such a prediction in municipalities in this country, as well as the use of Artificial Intelligence in papers on the subject at the national level. In general, the average of correct answers for the model was 64%, being better to predict the problem in periods without the influence of the Covid-19 Pandemic, and when that data was influenced by this sanitary crisis, different results were obtained, improving the forecast of non-fiscal stress and worsening that of stress. It is also noted that the environmental and organizational factors showed low importance, as well as the analyzed mesoregions. The most important variables are related to liquidity and debt cash was the aspect that varied the most in the two analyzed periods. Such results showed some improvements in relation to previous paper, however, it obtained low results for the area under the Receiver Operating Characteristic curve. In addition, it was possible to corroborate some authors and contradict others.