CMStatistics 2022: Start Registration
View Submission - CFE
Title: Loss given default in shipping finance: A machine learning approach Authors:  Rita DEcclesia - Sapienza University of Rome (Italy) [presenting]
Aida Salko - sapienza University of Rome (Italy)
Abstract: Different parametric and non-parametric modeling methods for estimating the Loss Given Default (LGD) of bank loans for shipping companies are analyzed. The shipping industry is subject to several other risks, which create the need to accurately measure the possible losses to estimate the LGDs for the banking industry. We use a unique database of defaulted loans in European banks involved in shipping finance. The aim is twofold: to compare the performance of alternative LGD modeling methodologies in shipping finance and to provide some insights into what drives LGD in the shipping industry. We find that non-parametric methods, predominantly random forest, lead to a remarkable increase in prediction accuracy and outperform the traditional statistical models in terms of both in-sample and out-of-sample results. To investigate the risk drivers in the shipping business, we use a variable importance measure built on the idea of permutation importance. We find the energy index to be paramount and the most important risk factor in estimating shipping finance LGD. We find that crude oil prices play a big role and may affect the financial health of shipping firms and then the LGDs of shipping loans.