Speaker: Robert P. Lieli (University of Texas at Austin)
Venue: MNB-Visitor Centre
Time: 15:15 pm, Thursday, March 18, 2010
Abstract:
The paper examines the econometric implications of the decision problem faced by a profit- or utility maximizing lender operating in a simple “double-binary” environment, where there are only two actions available (approve or reject) and there are only two possible states of the world (pay back fully or default). In practice, an approval decision is often arrived at by applying a fixed cutoff to credit scores obtained by estimating a logit model of the conditional probability of default. Following Elliott and Lieli (2005), we argue that this practice may well contradict the goal of profit-maximization and propose the use of “context-specific” cutoffs and an estimation method that explicitly relies on the economic objective of the lender. We also show how the proposed framework can help the lender avoid “legal risks” emanating from the prohibition of disparate treatment and disparate impact. The econometric method is illustrated by an application to German credit data.