Credit Scoring Models with Truncated Samples and Their Validation

The main object of this paper is to develop a credit scoring methodology for Argentine bank commercial obligors based on information available in the Public Credit Registry of the Central Bank of Argentina (Central de Deudores) as a reference tool to assess credit risk in local banks. Previous experience in this field has shown promising results; in this paper, we focus on two innovative aspects: firstly, the potential bias introduced by the fact that a considerable number of obligors are removed from the database for no traceable reason, and secondly, the application of validation techniques to the resulting models as proposed by the document recently published by the BCBS.