Reject inference in credit operations based on survival analysis

Reject inference is the problem of inferring the good and bad properties of those customers who have not been accepted for a loan. We present a reject inference method based on the confidence interval of a median survival time to delayed repayment. In this method, survival model is built based on the accepted applicant data and is applied to the rejected applicant. If the lower limit of the 90% confidence interval of the median life of the rejected applicant is longer than the median life of the accepted applicants, we consider the rejected applicant should have been accepted. The proposed method has an advantage of predicting the time to delayed repayment for an applicant with associated characteristics so that the proper loan duration can be set.