Title: An intelligent multi-level optimisation model for retail loan portfolio

Authors: Srilatha Kappagantula; Vikas Srivastava

Addresses: Indian Institute of Management Lucknow, Prabandh Nagar, IIM Road, Lucknow-226013, India ' Indian Institute of Management Lucknow, Prabandh Nagar, IIM Road, Lucknow-226013, India

Abstract: The paper discusses the multi-level portfolio selection problem, which combines hierarchical optimisation of credit portfolio, incorporating regulatory and capital constraints, in the context of emerging retail-banking loans. The proposed model allows for twin objectives of risk minimisation, simultaneously providing scope for maximising returns. The present paper analyses the portfolio optimisation problem, as a holistic 2-level optimisation problem: 1) at loan level, to reduce the default risk; 2) at bank level, to decide the right capital allocation between loan classes. The current study develops a model for multi-level optimisation of loans, and solves the model using multi objective algorithm for allocation of loan data across four retail asset classes, namely small business loans, credit card loans, home loans and auto loans, using a dataset of 229,000 loan records. The multi-level optimised portfolio is compared against the original portfolio for potential gains.

Keywords: banking; portfolio optimisation; portfolio allocation; portfolio selection; retail banking; machine learning; artificial intelligence; AI.

DOI: 10.1504/IJICBM.2024.139165

International Journal of Indian Culture and Business Management, 2024 Vol.32 No.2, pp.164 - 186

Received: 19 Sep 2022
Accepted: 26 Dec 2022
Published online: 24 Jun 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article