Um modelo de previsão de solvência utilizando regressão logística

Abstract

RESUMOO processo de entrada de novos bancos estrangeiros no mercado brasileiro, aliado à estabilidade monetária do país, está requerendo do sistema financeiro uma mudança em seu perfil de atuação, principalmente na área de crédito. Neste sentido, este artigo representa uma importante contribuição ao apresentar os resultados do teste e comprovação de uma nova técnica (regressão logística) para avaliar o risco de crédito. Esta ferramenta estatística se mostrou mais robusta em relação a outras técnicas utilizadas em trabalhos desta natureza. Ademais, as características da amostra (tamanho, perfil e origem) consubstanciam-se como diferenciais deste estudo em relação aos anteriores, pois foram utilizados 323 clientes de uma instituição financeira, identificados como empresas do setor industrial. Em função da revisão da literatura, foram selecionados 49 indicadores financeiros para a análise de solvência. Através da aplicação da análise estatística, conhecida como regressão logística, foi obtido o modelo econométrico de previsão de solvência, composto por 5 variáveis. A precisão deste modelo foi bastante alta, pois 94,85% das empresas foram classificadas corretamente. A validação do modelo foi realizada por meio do método conhecido como crossvalidation, ou seja, a subdivisão da amostra original: uma para a definição do modelo e outra para a sua validação.Palavras-chaves: bancos; crédito; regressão logística. ABSTRACTThe process of entry of foreign banks in the Brazilian financial system, together with the achievement of monetary stability, are requiring changes in the profile of action of Brazilian banks, especially in the area of credit. In this sense, this article represents an important contribution in testing a new technique (logistic regression) in evaluating credit risk. The statistical technique used shows itself more robust than other methods that have been used in previous studies of this nature. Additionally, characteristics of the sample used (size, profile, and source) differentiate this study from others. 323 clients from the industrial sector of a financial institution were used. Through the application of the statistical technique known as logistic regression, an econometric model of solvency prediction was developed that includes five variables. The precision of the model was relatively high, in that 94.85% of the sample was classified correctly. The validation of the model was performed by the method known as cross-validation, dividing the original sample in two: one part for constructing the model and the other for validating the model.

Predicting financial distress in publicly-traded companies

“…Then, Ohlson (1980) used the logit model (logistic regression) with financial indicators for predicting company bankruptcy and determined that the factors related to probable bankruptcy within the space of a year were company size and measures of financial structure, performance, and liquidity. Minussi, Damacena, and Ness (2002) report that the advantage of logistic regression compared with multivariate discrimant analysis lies in its coverage of possibilities, given that it is not necessary to guarantee the normality of residues nor the existence of homogeneity of the variance. Moreover, the logistic regression models enable the likelihood of a company going into bankruptcy to be estimated (Balcaen & Ooghe, 2004).…”

Analysis of credit risk faced by public companies in Brazil: an approach based on discriminant analysis, logistic regression and artificial neural networks

“…Results diverging from those of Minussi et al (2002) concerned the fact that only 4 solvent companies in the sample had Type I Financial Structure 'Excellent', i.e., they presented positive WC and BT, and negative NWC (Table 8). On the other hand, most solvent companies presented a 'solid' Type 2 Financial Structure (32 companies, positive WC, NWC and BT) or 'dissatisfactory' Type 3 (22 companies, positive WC and NWC, and negative BT), wherein NWC was positive.…”

Financial distress in Brazilian banks: an early warning model,

“…In the main Brazilian journals there are studies on solvency, generally related to publicly-traded Brazilian companies; however none covering Brazilian banks in their sample. ese studies include those from Brito and Assaf Neto (2008), Brito, Assaf Neto, and Corrar (2009), Guimarães andAlves (2009), Minardi (2008), Minussi, Damacena, and Ness Jr. (2002), Onusic, Nova, and Almeida (2007), and Bressan, Braga, and Bressan (2004), with the latter analyzing insolvency risk in credit cooperatives from the state of Minas Gerais. e study from Liu (2015), also published in a Brazilian journal, addresses factors determining nancial di culties in banks from various countries, but in its sample it does not explain which observations were used, as well as obtaining a low predictive power in the models.…”