R. Peláez Suárez, R. CAO ABAD, J. VILAR FERNANDEZ
For a fixed time, t, and a horizon time, b, the probability of default (PD) measures the probability that an obligor, who has paid their credit until time t, will run into arrears no later than time t+b. This probability is one of the most crucial elements that influence the credit risk. Previous works have proposed nonparametric estimators for the PD derived from Beran's estimator and a doubly smoothed Beran's estimator of the conditional survival function for censored data. Asymptotic theory has been developed for them, but no practical method for choosing the smoothing parameters involved has been provided. In this work, bootstrap procedures are proposed to approximate the bandwidths of Beran's and the smoothed Beran's estimators of the PD. Bootstrap algorithms for calculating confidence intervals of the probability of default are also proposed. Extensive simulation studies and the application to a real data set prove the good performance of the techniques presented.
Keywords: Bootstrap, Censored data, Credit risk, Kernel method, Survival analysis
Scheduled
Ramiro Melendreras Award IV
November 8, 2023 10:10 AM
CC4: Room 2