Incorporating Asset Depreciation in SME Leasing Default Prediction: An XGBoost Approach

Wensheng Wu

doi:10.54691/09wcpw24

Authors

Wensheng Wu

DOI:

https://doi.org/10.54691/09wcpw24

Keywords:

SME financing, leasing default prediction, asset depreciation, XGBoost, SHAP.

Abstract

This study develops an XGBoost approach to forecast default risk in small and medium-sized enterprise (SME) equipment leasing, with a particular focus on asset depreciation patterns. The model was trained and tested on a dataset of 6,218 contracts from a Chinese leasing company. Beyond conventional predictors, it incorporates key depreciation metrics: residual value ratio, asset age, and remaining useful life. The proposed XGBoost model achieved an AUC of 0.813, outperforming all benchmark models. SHAP analysis identified the residual value ratio as the second most important predictor, underscoring the critical role of asset-specific factors in leasing defaults. Crucially, the SHAP dependence plot revealed a pronounced negative correlation between the residual value ratio and default probability. This finding aligns with the economic intuition underpinning leasing: equipment with a higher projected residual value strengthens the lessor’s collateral position and reduces the lessee’s incentive to default, as the asset retains significant recoverable value.

Downloads

Download data is not yet available.

References

[1] C. Zhang and X. Zhou, “Forecasting Credit Risk of SMEs in Supply Chain Finance Using Bayesian Optimization and XGBoost,” Mathematical Problems in Engineering, vol. 2023, no. 1, p. 5609996, 2023/01/01 2023, doi: 10.1155/2023/5609996.

[2] A. Kozina et al., “The default of leasing contracts prediction using machine learning,” Procedia Computer Science, vol. 225, pp. 424-433, 2023/01/01/ 2023, doi: 10.1016/j.procs.2023.10.027.

[3] L. Barbaglia, S. Manzan, and E. Tosetti, “Forecasting Loan Default in Europe with Machine Learning,” Journal of Financial Econometrics, vol. 21, no. 2, pp. 569-596, 2023, doi: 10.1093/jjfinec/nbab010.

[4] A. Bitetto, P. Cerchiello, S. Filomeni, A. Tanda, and B. Tarantino, “Can we trust machine learning to predict the credit risk of small businesses?,” Review of Quantitative Finance and Accounting, vol. 63, no. 3, pp. 925-954, 2024/10/01 2024, doi: 10.1007/s11156-024-01278-0.

[5] H. Cheraghali and P. Molnár, “SME default prediction: A systematic methodology-focused review,” Journal of Small Business Management, vol. 62, no. 6, pp. 2847-2905, 2024/11/01 2024, doi: 10.1080/00472778.2023.2277426.

[6] M. Hernes, J. Adaszyński, and P. Tutak, “Credit risk modeling using interpreted XGBoost,” European Management Studies, vol. 21, no. 3, pp. 46-70, 2023, doi: 10.7172/2956-7602.101.3.

[7] R. Hlongwane, K. Ramabao, and W. Mongwe, “A novel framework for enhancing transparency in credit scoring: Leveraging Shapley values for interpretable credit scorecards,” PLOS ONE, vol. 19, no. 8, p. e0308718, 2024, doi: 10.1371/journal.pone.0308718.

[8] S. Yang, Z. Huang, W. Xiao, and X. Shen, “Interpretable Credit Default Prediction with Ensemble Learning and SHAP,” arXiv preprint, vol. arXiv:2505.20815, 2025, doi: 10.48550/arXiv.2505.20815.

[9] H. Cheraghali and P. Molnár, “SME default prediction: A systematic methods evaluation,” Journal of Small Business Management, vol. 63, no. 4, pp. 1466-1517, 2025/07/04 2025, doi: 10.1080/00472778.2024.2390500.

[10] L. Lin and Y. Wang, “SHAP Stability in Credit Risk Management: A Case Study in Credit Card Default Model,” Risks, vol. 13, no. 12, p. 238, 2025, doi: 10.3390/risks13120238.