One of the most important aspects of credit scoring is constructing a model that has low misclassification rates and is also flexible enough to allow for random variation. It is also well known that, when there are a large number of highly correlated variables as is typical in studies involving questionnaire data, a method must be found to reduce the number of variables to those that have high predictive power. Here we propose a Bayesian multivariate logistic regression model with both fixed and random effects for small business loan credit scoring and a variable reduction method using Bayes factors. The method is illustrated on an interesting data set based on questionnaires sent to loan officers in Canadian banks and venture capital companies.

Additional Metadata
Keywords Bayes factors, Credit scoring, MCMC, Variable selection
Journal Pakistan Journal of Statistics and Operation Research
Farrell, P, MacGibbon, B. (Brenda), Tomberlin, T.J, & Doreen, D. (Dale). (2011). A bayesian analysis of a random effects small business loan credit scoring model. Pakistan Journal of Statistics and Operation Research, 7(2 SPECIAL ISSUE), 433–449.