The Use of Generalizability Theory to Improve International Lending Decisions

This paper demonstrates the application of generalizability theory, a statistical technique which explicitly recognizes multiple sources of measurement error, to the identification of error or inconsistency in commercial bank loan ratings. The approach allows decision makers to assess the reliability of the credit analysis process used in either international or domestic lending. A sample obtained from a large commercial bank is analyzed to determine whether. the credit analysts' evaluative skills or the rating criteria utilized contribute to variability in the score for each loan. The results indicate that differences in credit analysts' loan evaluations are an important source of error, suggesting a possible tradeoff between increased analysis costs and improved rating consistency.