Event Details

Title: A joint scoring model for peer-to-peer and traditional lending: a bivariate model with copula dependence

We analyse the dependence between defaults in peer-to-peer (P2P) lending and credit bureaus. To achieve this aim, we propose a new flexible bivariate regression model suitable for binary imbalanced samples. We use different copula functions to model the dependence structure between defaults in the two credit markets. We implement the model in the R package BivGEV and we analyse its main characteristics through a Monte Carlo study. The application of this proposal to a comprehensive dataset provided by Lending Club shows a significant level of dependence between the defaults in P2P and credit bureaus. Finally, we find that our model outperforms the bivariate probit and univariate logit in predicting P2P default, in estimating the Value at Risk and the Expected Shortfall.