A Comparative Study of LSTM and BP Models in Supply Chain Finance Risk Assessment of A-stock Listed Steel Companies

Abstract

The iron and steel industry faces the high-risk management challenges in the supply chain finance due to its capital-intensive characteristics and complex supply chain structure. Based on the financial data of 26 iron and steel companies in A-stock, this study uses Long Short-Term Memory (LSTM) neural network and the Back-Propagation (BP) models to assess and compare supply chain finance risks. A multidimensional assessment system covering corporate creditworthiness, supply chain capability, and risk level is established by analyzing the financial data from 2020-2023. The results show that although the LSTM model is good at dealing with long-time series data, the BP model outperforms the LSTM and better fits the actual data when the sample size is small. Therefore, the BP model is preferred when the amount of data is limited, while the LSTM model may perform better when the amount of data increases significantly to improve prediction accuracy and decision support.