VISUALIZING DYNAMIC BALANCE-SHEET RISK IN NON-BANKING FINANCIAL COMPANIES THROUGH AI-ENABLED CASH-FLOW NETWORKS

Rupali Patil; Manisha Raghuvanshi; Somnath Wategaonkar; Savita Patil; Milind Eknath Rane; Medha V Wyawahare

doi:10.29121/shodhkosh.v7.i7s.2026.7930

Authors

Rupali Patil Department of Electronics and Telecommunication Engineering, Bharati Vidyapeeth College of Engineering, Navi Mumbai, India
Manisha Raghuvanshi Engineering Science Department, AISSMS Institute of Information Technology, Pune, India
Somnath Wategaonkar Department of Electronics and Telecommunication Engineering Bharati Vidyapeeth College of Engineering, Navi Mumbai, India
Savita Patil Department of Electronics and Telecommunication Engineering Bharati Vidyapeeth College of Engineering, Navi Mumbai, India
Milind Eknath Rane Department of Electronics and Telecommunication Vishwakarma Institute of Technology, Pune, India
Medha V. Wyawahare Department of Electronics and Telecommunication Vishwakarma Institute of Technology, Pune, India

DOI:

https://doi.org/10.29121/shodhkosh.v7.i7s.2026.7930

Keywords:

Non-Banking Financial Companies (NBFCs), Liquidity Risk Modeling, Temporal Graph Neural Networks (TGNN), Dynamic Balance Sheet Analysis, AI-Driven Risk Assessment

Abstract [English]

Non-Banking Financial Companies (NBFCs) play a crucial role in the Indian credit landscape. However, their inherent vulnerabilities to liquidity crises position them as highly susceptible to significant shocks. Current approaches to risk assessment for these companies are inadequate for understanding the dynamics of liquidity risk. An AI-enabled approach to modeling the dynamic balance sheets of NBFCs using high-frequency data on their cash flows can help improve risk assessment for these companies. A network of NBFCs can be represented as a graph, with each NBFC as a node in the graph and the transactions between these companies forming the edges of the graph. The AI model that can be used to analyze this graph is a Temporal Graph Neural Network (TGNN). TGNNs are deep learning models specifically designed to learn from network graphs and capture the relationships between the entities represented in the graph nodes. On top of the TGNN, a dynamic risk index for NBFCs can be created by calculating various metrics for the graph created by the NBFCs. These metrics will represent the various aspects of the risk for these companies at any given time. Furthermore, the model can be evaluated using high-frequency datasets of NBFCs that have been simulated to contain realistic liquidity risk dynamics. These results can be compared with those of other risk models currently in use for NBFCs. The proposed model will have superior performance indicators to other risk models for NBFCs. Not only will the model be able to accurately forecast the risk of individual NBFCs, but also the liquidity risk that exists within the entire NBFC industry as a whole. This AI-enabled model will allow for the early identification of the liquidity risks that individual NBFCs and the industry as a whole face. Consequently, the regulators and the companies themselves will be able to take steps to mitigate these risks. Using such a model will improve the monitoring of the NBFC industry by effectively integrating artificial intelligence into the risk assessment of its constituent companies.

References

F. Allen and D. Gale, “Financial contagion,” Journal of Political Economy, vol. 108, no. 1, pp. 1–33, 2000. doi: 10.1086/262109

L. Eisenberg and T. H. Noe, “Systemic risk in financial systems,” Management Science, vol. 47, no. 2, pp. 236–249, 2001. doi: 10.1287/mnsc.47.2.236.9835

S. Battiston, M. Puliga, R. Kaushik, P. Tasca, and G. Caldarelli, “DebtRank: Too central to fail? Financial networks, the FED and systemic risk,” Scientific Reports, vol. 2, p. 541, 2012. doi: 10.1038/srep00541

D. Acemoglu, A. Ozdaglar, and A. Tahbaz-Salehi, “Systemic risk and network formation,” American Economic Review, vol. 105, no. 2, pp. 564–608, 2015. doi: 10.1257/aer.20130456

L. C. Freeman, “Centrality in social networks: Conceptual clarification,” Social Networks, vol. 1, no. 3, pp. 215–239, 1978. doi: 10.1016/0378-8733(78)90021-7

L. Breiman, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. doi: 10.1023/A:1010933404324

C. Cortes and V. Vapnik, “Support-vector networks,” Machine Learning, vol. 20, no. 3, pp. 273–297, 1995. doi: 10.1007/BF00994018

J. H. Friedman, “Greedy function approximation: A gradient boosting machine,” Annals of Statistics, vol. 29, no. 5, pp. 1189–1232, 2001. doi: 10.1214/aos/1013203451

S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997. doi: 10.1162/neco.1997.9.8.1735

T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” arXiv preprint, 2017. doi: 10.48550/arXiv.1609.02907

Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and P. S. Yu, “A comprehensive survey on graph neural networks,” IEEE Transactions on Neural Networks and Learning Systems, vol. 32, no. 1, pp. 4–24, 2021. doi: 10.1109/TNNLS.2020.2978386

J. Zhou, G. Cui, S. Hu, Z. Zhang, C. Yang, Z. Liu, L. Wang, C. Li, and M. Sun, “Graph neural networks: A review of methods and applications,” AI Open, vol. 1, pp. 57–81, 2020. doi: 10.1016/j.aiopen.2021.01.001

P. Veličković, G. Cucurull, A. Casanova, A. Romero, P. Liò, and Y. Bengio, “Graph attention networks,” arXiv preprint, 2018. doi: 10.48550/arXiv.1710.10903

E. Rossi, B. Chamberlain, F. Frasca, D. Eynard, F. Monti, and M. M. Bronstein, “Temporal graph networks for deep learning on dynamic graphs,” arXiv preprint, 2020. doi: 10.48550/arXiv.2006.10637

S. M. Kazemi, R. Goel, K. Jain, I. Kobyzev, A. Sethi, P. Forsyth, and P. Poupart, “Representation learning for dynamic graphs: A survey,” Journal of Machine Learning Research, vol. 21, no. 70, pp. 1–73, 2020. doi: 10.48550/arXiv.2004.00453

D. J. Watts, “A simple model of global cascades on random networks,” Proceedings of the National Academy of Sciences, vol. 99, no. 9, pp. 5766–5771, 2002. doi: 10.1073/pnas.082090499

P. Gai and S. Kapadia, “Contagion in financial networks,” Journal of Financial Stability, vol. 6, no. 3, pp. 174–181, 2010. doi: 10.1016/j.jfs.2010.01.002

P. Glasserman and H. P. Young, “Contagion in financial networks,” Journal of Economic Literature, vol. 54, no. 3, pp. 779–831, 2016. doi: 10.1257/jel.20151228

C. Brownlees and R. F. Engle, “SRISK: A conditional capital shortfall measure of systemic risk,” Review of Financial Studies, vol. 30, no. 1, pp. 48–79, 2017. doi: 10.1093/rfs/hhw060

T. Adrian and M. K. Brunnermeier, “CoVaR,” American Economic Review, vol. 106, no. 7, pp. 1705–1741, 2016. doi: 10.1257/aer.20120555

D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint, 2015. doi: 10.48550/arXiv.1412.6980

K. Xu, W. Hu, J. Leskovec, and S. Jegelka, “How powerful are graph neural networks?” arXiv preprint, 2019. doi: 10.48550/arXiv.1810.00826

Y. Li, R. Yu, C. Shahabi, and Y. Liu, “Diffusion convolutional recurrent neural network,” arXiv preprint, 2018. doi: 10.48550/arXiv.1707.01926

M. Defferrard, X. Bresson, and P. Vandergheynst, “Convolutional neural networks on graphs,” Advances in Neural Information Processing Systems, 2016. doi: 10.48550/arXiv.1606.09375

M. E. J. Newman, Networks: An Introduction. Oxford University Press, 2010. doi: 10.1093/acprof:oso/9780199206650.001.0001

I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016. doi: 10.7551/mitpress/10243.001.0001

M. T. Ribeiro, S. Singh, and C. Guestrin, “Why should I trust you?” KDD, 2016. doi: 10.1145/2939672.2939778

S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” NeurIPS, 2017. doi: 10.48550/arXiv.1705.07874

W. L. Hamilton, Graph Representation Learning. Morgan & Claypool, 2020. doi: 10.2200/S01018ED1V01Y202003AIM046

D. J. Watts and S. H. Strogatz, “Collective dynamics of ‘small-world’ networks,”