Risco de crédito: uma abordagem utilizando análise discriminante, regressão logística e redes neurais artificiais
Year: 2016
Type: dissertation
Abstract: Considering the relevance of researches concerning credit risk, model diversity \nand the existent indicators, this thesis aimed at verifying if the Fleuriet Model \ncontributes in discriminating Brazilian open capital companies in the analysis of \ncredit concession. We specifically intended to i) identify the economic-financial \nindicators used in credit risk models; ii) identify which economic-financial \nindicators best discriminate companies in the analysis of credit concession; iii) \nassess which techniques used (discriminant analysis, logistic regression and \nneural networks) present the best accuracy to predict company bankruptcy. To \ndo this, the theoretical background approached the concepts of financial \nanalysis, which introduced themes relative to the company evaluation process; \nconsiderations on credit, risk and analysis; Fleuriet Model and its indicators, \nand, finally, presented the techniques for credit analysis based on discriminant \nanalysis, logistic regression and artificial neural networks. Methodologically, the \nresearch was defined as quantitative, regarding its nature, and explanatory, \nregarding its type. It was developed using data derived from bibliographic and \ndocument analysis. The financial demonstrations were collected by means of the \nEconomática \n® \nand the BM$FBOVESPA website. The sample was comprised of \n121 companies, being those 70 solvents and 51 insolvents from various sectors. \nIn the analyses, we used 22 indicators of the Traditional Model and 13 of the \nFleuriet Model, totalizing 35 indicators. The economic-financial indicators \nwhich were a part of, at least, one of the three final models were: X1 (Working \nCapital over Assets), X3 (NCG over Assets), X4 (NCG over Net Revenue), X8 \n(Type of Financial Structure), X9 (Net Thermometer), X16 (Net Equity divided \nby the total demandable), X17 (Asset Turnover), X20 (Net Equity Profitability), \nX25 (Net Margin), X28 (Debt Composition) and X31 (Net Equity over Asset). \nThe final models presented setting values of: 90.9% (discriminant analysis); \n90.9% (logistic regression) and 97.8% (neural networks). The modeling in \nneural networks presented higher accuracy, which was confirmed by the ROC \ncurve. In conclusion, the indicators of the Fleuriet Model presented relevant \nresults for the research of credit risk, especially if modeled by neural networks.
Author: José Willer do Prado
Institution: Universidade Federal de Lavras
Language: Portuguese
Subfield: Accounting
Domain: Social Sciences
Keywords: credit risk, Fleuriet model, credit risk modeling, bankruptcy prediction, financial ratios, financial and economic indicators, open capital companies, lending, discriminant analysis, logistic regression, artificial neural networks, model comparison, Brazilian listed companies, working capital, net equity, debt structure, asset turnover, net profit margin, financial structure
Open Access status: green
OpenAlex ID: W2397067415
