Microenvironment-specific effects in the application credit scoring model

This paper introduces the improved version of a credit scoring model which assesses credit worthiness of applicants for a loan. The scorecard has a two-level multilevel structure which nests applicants for a loan within microenvironments. In addition, the paper proposes a new type of clustering for a hierarchical two-level structure which is more intuitive and efficient in the application to credit scoring. This structure explores living area-specific effects which are viewed as unobserved determinants of default. The primary benefit of the multilevel scorecard compared to a conventional scoring model is higher accuracy of the model predictions.