Algunos aspectos sobre el análisis empírico de "credit scoring"
ISSN: 2255-5471
Year of publication: 1992
Issue: 5
Type: Working paper
More publications in: Documentos de trabajo de la Facultad de Ciencias Económicas y Empresariales
Abstract
A common problem in empirical analyses of credit scoring is that collected samples are not random, but instead they arise from a selection rule. In the general case, to correct the sample selection bias and obtain consistency a censored sample (including both granted and denied credits) needs to be used and a bivariate model with partial observability has to be estimated. In this article, we focus on this problem and its main contribution is to propose a procedure to obtain consistency in the restricted but frequent case in which the available sample is truncated (including only granted credits).