TY - JOUR
T1 - Machine-learning models for bankruptcy prediction: do industrial variables matter?
AU - Bragoli, Daniela
AU - Ferretti, Camilla
AU - Ganugi, Piero
AU - Marseguerra, Giovanni
AU - Mezzogori, Davide
AU - Zammori, Francesco
PY - 2021
Y1 - 2021
N2 - We provide a predictive model specifically designed for the Italian economy that classifies solvent and insolvent firms one year in advance using the AIDA Bureau van Dijk data set for the period 2007–15. We apply a full battery of bankruptcy forecasting models, including both traditional and more sophisticated machine-learning techniques, and add to the financial ratios used in the literature a set of industrial/regional variables. We find that XGBoost is the best performer, and that industrial/regional variables are important. Moreover, belonging to a district, having a high mark-up and a greater market share diminish bankruptcy probability.
AB - We provide a predictive model specifically designed for the Italian economy that classifies solvent and insolvent firms one year in advance using the AIDA Bureau van Dijk data set for the period 2007–15. We apply a full battery of bankruptcy forecasting models, including both traditional and more sophisticated machine-learning techniques, and add to the financial ratios used in the literature a set of industrial/regional variables. We find that XGBoost is the best performer, and that industrial/regional variables are important. Moreover, belonging to a district, having a high mark-up and a greater market share diminish bankruptcy probability.
KW - firm distress analysis
KW - industrial variables
KW - logistic regression
KW - machine learning
KW - firm distress analysis
KW - industrial variables
KW - logistic regression
KW - machine learning
UR - http://hdl.handle.net/10807/189731
U2 - 10.1080/17421772.2021.1977377
DO - 10.1080/17421772.2021.1977377
M3 - Article
SN - 1742-1772
VL - October 2021
SP - 1
EP - 22
JO - Spatial Economic Analysis
JF - Spatial Economic Analysis
ER -