The inclusion of your time-different covariates into survival analysis has led to better predictions of enough time to default in behavioural credit score scoring types. However, when these time-various covariates are endogenous, There are 2 main troubles: estimation bias with the survival model