Credit Data Cohort Analysis funded by Fair Issac Corporation

Project Description :

    A common way of summarizing consumer credit data is by showing cross-sectional delinquency rates for a number of historical periods, split up by period of origination. Three causal factors are often isolated: life cycle, new account effects, and portfolio effects. Cohort analysis seeks to explain cohort table data in terms of differences across individuals’ age, cohorts, and time periods.

    In this research we addressed two main issues facing cohort analysis. The first one was the so called identification problem. Regardless of how a cohort table is examined, two of the three effects are confounded with one another. The second one is not really an issue but an extension of the tradition age-period-cohort (APC) approach in cohort analysis. We examined the impact of introducing macroeconomic variables to explain variations in delinquency rates. These variables should be linked to the portfolio effect that we tried to model, which is only imperfectly captured by a fixed period effect.

    The report to Fair Isaac contains two main sections. In the first section, we describe cohort analysis on a more theoretical standpoint, and compare different estimators, under the viewpoint of how they address the identification problem. In the second section, we provide an empirical analysis on the data that was provided to us by Fair Isaac. In the appendix, we describe how to use the code that we developed as well as a full listing of all the code.

Claremont Graduate University | Claremont McKenna | Harvey Mudd | Pitzer | Pomona | Scripps
Proudly Serving Math Community at the Claremont Colleges Since 2007
Copyright © 2018 Claremont Center for the Mathematical Sciences