Dynamic prediction of loss given default in Norwegian retail credit card debt: advanced machine learning versus statistical models
Ho, Cathy, Sandersen, Isabel A. C., Selliken, Espen, Risstad, Morten, Lange, Petter E., Westgaard, Sjur Digital Finance
· Springer Nature
· 2026-10-23
· DOI: 10.1007/s42521-026-00230-2
Abstract
The digitalization of financial services has generated large volumes of borrower data, enabling the use of advanced analytics in credit risk modeling. This study evaluates the performance of statistical and machine learning models for predicting and updating final Loss Given Default (LGD) estimates during recovery, using proprietary data from Norwegian retail credit card portfolios. Traditional regression models (Ordinary Least Squares, Ridge, and Lasso) are compared with ensemble-based machine learning techniques, including Random Forest and Extreme Gradient Boosting (XGBoost). The results show that machine learning models generally achieve lower prediction error, with XGBoost delivering the strongest predictive performance. However, linear models retain greater interpretability and higher explained variance. The findings highlight trade-offs between predictive accuracy and transparency and provide new empirical evidence on dynamic LGD modeling in unsecured retail credit markets.
Ho, Cathy, Sandersen, Isabel A. C., Selliken, Espen et al. Dynamic prediction of loss given default in Norwegian retail credit card debt: advanced machine learning versus statistical models. // Digital Finance. – 2026. – DOI: 10.1007/s42521-026-00230-2. – URL: http://link.springer.com/openurl/fulltext?id=doi:10.1007/s42521-026-00230-2
Ho, Cathy, Sandersen, Isabel A. C, Selliken, Espen, Risstad, Morten, Lange, Petter E, Westgaard, & Sjur (2026). Dynamic prediction of loss given default in Norwegian retail credit card debt: advanced machine learning versus statistical models. Digital Finance. https://doi.org/10.1007/s42521-026-00230-2
Ho, et al. “Dynamic prediction of loss given default in Norwegian retail credit card debt: advanced machine learning versus statistical models.” Digital Finance, 2026. doi:10.1007/s42521-026-00230-2.
@article{Ho2026,
title = {Dynamic prediction of loss given default in Norwegian retail credit card debt: advanced machine learning versus statistical models},
author = {Ho and Cathy and Sandersen and C., Isabel A. and Selliken and Espen and Risstad and Morten and Lange and E., Petter and Westgaard and Sjur},
journal = {Digital Finance},
year = {2026},
doi = {10.1007/s42521-026-00230-2},
url = {http://link.springer.com/openurl/fulltext?id=doi:10.1007/s42521-026-00230-2}
}
Granger causality remains a central framework for studying predictive causal relations in economics and finance, but its methodological scope now extends far beyond classical linear vector autoregression. Existing surveys often examine...
This study investigates the feature importance of return and volatility connectedness between Islamic sectoral indices and meme assets across quantiles. Using daily data from August 2020 to March 2024, we adopt a multistage approach...
Unless utility functions are logarithmic or posed problems are single-period, it is very difficult to explicitly solve constrained optimal portfolio problems. This paper proposes a new numerical method for solving the multiple-period...
We propose novel realized dynamic conditional correlation (Realized DCC) models with measurement errors to forecast the covariance of asset returns. This study aims to forecast the conditional covariance matrix of asset returns and to...
For small values of correlation a method is given to de-correlate the instantaneous volatility from the price process in stochastic volatility models. The result of the de-correlation is that the implied volatility skew is rotated into...