
Yes, You Can Predict Loan Default from Loan Application Text
This study shows that analyzing text from online loan requests can improve predictions of loan default by revealing insights into borrowers' psychological and personality traits. Incorporating Linguistic Inquiry and Word Count (LIWC) analysis into predictive models provides a more nuanced view of borrower risk and significantly enhances accuracy, potentially boosting lenders' ROI by up to 5.75% compared to models relying solely on financial and demographic data.




