Indicadores contábeis financeiros e a previsibilidade de insolvência das distribuidoras de energia elétrica no Brasil

• Autor(a) principal: Silva, Shirley Fernandes Pereira da
• Data de Publicação: 2020
• Outros Autores: [e-mail]
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UERJ
• Texto Completo: http://www.bdtd.uerj.br/handle/1/18344
• topic: Electric Power Distributors; Insolvency; Accounting Indicators; Regression Model; Panel Data; Distribuidoras de Energia Elétrica; Insolvência; Indicadores contábeis; Modelo de Regressão Logística; Dados em painel; Serviços de eletricidade – Contabilidade – Brasil; Serviços de eletricidade – Falência – Brasil; CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO::CIENCIAS CONTABEIS

This research aims to analyze the predictive capacity of financial accounting indicators in the probability of insolvency of electricity distributors. Through the analysis of the financial statements published on the website of the National Electric Energy Agency (ANEEL) of 63 distribution concessionaires, in the period 2011-2018. This research has the characteristic of being a bibliographic, descriptive and quantitative study. The collected data were submitted to the calculations of the indicators participating in the research, as established by accounting theory. Soon after, the data were submitted to an exploratory analysis, proceeding to a screening of the 18 indicators selected for the research (LC, LG, ENDAT, ENTOT, LOLB, LODF, EBTDF, EBITDA, EBTRL, ROA, ROE, MO, LLRL, DEFIN, INFIN, PL AND PNT), where all indicators were presented with statistically significant results. However, the indicators ENDF, ENTOT, ROE, EBTDF, had unexpected results from the perspective of accounting theory, and thus were discarded. In view of this, 14 indicators were considered for the continuity of the modeling. These 14 indicators were separated into insample (used to adjust the model) and outsample (used to assess the predictive capacity of adjusted models). Then, a search was made for all logistic regression models in which the regression coefficients were statistically significant at the level of 10%. For the selection of the best models, the AIC criterion, ROC curve, hit rate and Akaike Weights were used. Then, 6,479 models were fitted, but only 105 models presented statistically significant regression coefficients and with the expected signs, according to accounting theory, in the period 2011-2015. Thus, the indicators were identified: LLRL, DEFIN, LG, INFIN and PNT as relevant indicators to predict insolvency. To establish the robustness of the modeling, data from the 2011-2017 period were used, where 6,479 adjusted models were obtained, having 134 models with statistically significant regression coefficients and with the expected signs, according to accounting theory. During this period, the indicators were identified: DEFIN, INFIN, PNT, LG, LLRL, LODF, EBITDA, PL and ROA. Finally, after analyzing these two periods, 5 relevant indicators were identified to predict the insolvency of electricity distributors: DEFIN, INFIN, PNT, LG and LLRL.