Two case studies are presented which focus on model building and validation. Model building is described in detail, and seeks to clarify the main issues concerned with model building, how to use statistical tools and interpret results. Univariate, bivariate and multivariate stages of model building are discussed outlining the need to merge the knowledge of quant people with that of bank practitioners.
The validation case study looks a document for the internal validation unit, summarizing the process of building a shadow rating for assessing financial institutions creditworthiness is presented and analyzed.
Conclusions are drawn on the use of policies in order to gain leverage on potentialities and managing limits of statistical based ratings.