The bank loan approval decision from multiple perspectives

This paper develops a probability model to evaluate the predictive validity of two-way classification schemes in the context of personal credit scoring and bank loan applications. The Bayesian decision model provides a structure for identifying classification rules that lead to optimal-maximum expected payoff or minimum expected cost-classifications. Using payoffs from multiple perspectives allows identifying conditions where the various perspectives produce contradictory classifications generating either profit premiums or cost penalties depending on the perspective.

► A customer whose credit score remains constant may no longer pass approval criteria due to dropping real estate values.
► Under certain (realistic) conditions, banks approve loans not in the best interest of a customer.
► The synergistic effect of changing economic conditions and high risk mortgages lead to the financial crisis.