Created an underwriting decisioning engine for a foreign education loan product
Challenge
The client was underwriting educational loans for students going to courses in foreign markets for post graduation studies. Since, the principal borrowers have limited financial and credit information, the lender was struggling to underwrite effectively.
Solution
Since for students, the historical credit information is sparse, we focussed our efforts on determining the employability and employment potential of candidates. Apart from financial data for historical applications, we leveraged the information provided on past details – educational details, scores like GMAT/ GRE, work experiences, and future details – course details of the future educational program. However, the client had limited capabilities for differentiating applicants based on these information and had traditionally relied on subjective judgements of the underwriter.
We created several datasets based on publicly available information – such as ranking of undergraduate and post-graduate schools by department. We collected data for selectiveness of schools (% of applications accepted), course completion rates, employment level within 3 months of passing school, average salaries, distance from major employment centers, etc.
We created risk underwriting models which predicted risk based on applicant’s credit bureau information, applicant’s educational and employment details (from the previously collected data), and co-applicant’s credit bureau ratings.
We finally created a decision engine which rates applicants based on limited details.
Impact
The scorecard was able to differentiate effectively between low risk and high risk candidates. We also leveraged the models to determine the loan to value and collateral requirement.
The decision engine helped reduce TAT by 300%.
