New model accurately estimates bank loan losses, outperforming popular methods.
A new model using generalized additive models was developed to estimate Loss Given Default (LGD) for bank loans. This model can capture the high concentration of LGDs at the boundaries and uncover nonlinear effects of different factors on LGDs. By including macroeconomic variables, the model can also estimate LGDs in downturn conditions. The model outperformed popular models like Tobit, decision tree, and linear regression models in predicting LGDs for different time horizons based on a survey of Italian banks' loan recovery processes.