A inadimplência em um programa de crédito de uma instituição financeira pública de Minas Gerais: uma análise utilizando Regressão Logística

ABSTRACTThe objective of this paper is to propose an econometric model to estimate the risk of default on financing granted by a public financial institution in the state of Minas Gerais. In elaborating this model data from contracts, socioeconomic inputs of partners and guarantors of and economic-financial information of the companies were used from a sample of 9,232 firms drawn from a population of 25,616 cases of financing for micro and small enterprises made between Jun./97 and Dec./05. A Hierarchical Logistic Regression Analysis was used in which the dependent variable was the quality of credit as defined by timely payments or default of the loan transaction. A total of 22 independent variables related to the contract, the company and the partners/guarantors were used of which five emerged as important in predicting insolvency and correctly classifying 88.5% of the companies. In conclusion, one can say that the amount financed and hence the proportion of this funding consumed bring about conditions for default, while the value of the assets of the guarantor in relation to the financing, the value of fixed investments and the age of the company act as protectors against default