Proposta de um modelo probabilístico de risco de inadimplência em uma cooperativa de crédito, com a aplicação da técnica de regressão logística

Abstract

The current view is of magnifying of operations of credit destined to physical persons. This growing justifies by the contract of financings and loans with larger terms of paying, increasing of income and job. Being thus, in the same way as if it expands the search for credit in the financial institutions, it occurs the increasing of the index of insolvency of these operations. With this, it appears the necessity of an efficient management and control of risk making that these institutions try to perfect the quantitative techniques used in its process of analyze of concession of credit and administration of risk. Thus, the objective of this research was to propose a model of risk of credit for to esteem the probability of insolvency in the segment of physical persons in a Credit Cooperative, located in Pelotas/RS. It was identified too the profile of the cooperated from information registered in cadastre that themselves relate to the institution, trying by means of them to verify what are the variables able to explain the suggested model of risk. It was developed a study of case, based on the period of December, 2010 and it was used for analyzing data the multivaried statistic technique of multiple logistic regression. It was obtained a sample of 400 cooperated and it was tested 29 co-varied of the respective bank of data of the Unicred Automation System (UAS). In this study, the variable dependent corresponds to the situation or no of the insolvency of the cooperated in the operations of credit. The results gotten from the adjusted model indicate that only two predatory variables were statistically significant (p<0,05) for esteeming the probability of the insolvency , that are: profession dentist (OR = 7,147) and if it has the monthly income over 10 thousand until 20 thousand Reals (OR = 4,339).