Kang Li
Research Subjects
Kang Li has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Han Wu
School of Electronics and Control Engineering, Changn University, Xin, China
1 shared publication - Jan Nokkala1 shared publication
Jia-Ni Heng
1 shared publicationXiao-Zhi Gao
1 shared publication
Latest Publications
- 25 June 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 325
An improved brain-motivated network for forecasting day-ahead stock prices of electricity companies
- 25 June 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 325
An improved brain-motivated network for forecasting day-ahead stock prices of electricity companies
- Open Access
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Credit Concentration Risk Model in Portfolios Based on Artificial Neural Network
The aim of this study is to propose a novel credit concentration risk estimation tool for banks and other credit institutions. The topic of credit concentration risk in portfolios has attracted more and more attention since the increasing adoption of internal and external credit ratings made by Bank regulation. Accurate assessment of concentration risk in portfolios becomes of great importance to banks nowadays because most of banking crises were caused by concentration risk. The recent sophisticated artificial intelligence and machine learning technologies with efficient, powerful, and automatic data analysis capability offer a new tool and possibility to increase the accuracy and implementation of credit concentration risk evaluation to the next level. We propose a credit concentration risk model based on one of machine learning algorithms called artificial neural network. To test the feasibility and effectiveness of the proposed model, we use the data of 282 portfolios from 30 financial institutions in the year 2019. Our main results suggest that it is feasible and reliable to measure concentration risk based on our proposed model, and its prediction accurate rate can exceed 98% in training sample and 92% in testing sample. The novel model broadens the areas for application of AI in risk management of banks and banking supervisors as it produces measurement possibilities to a risk which banks are also obliged to reserve capital if risk is material.
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Open AccessCredit Concentration Risk Model in Portfolios Based on Artificial Neural Network
The aim of this study is to propose a novel credit concentration risk estimation tool for banks and other credit institutions. The topic of credit concentration risk in portfolios has attracted more and more attention since the increasing adoption of internal and external credit ratings made by Bank regulation. Accurate assessment of concentration risk in portfolios becomes of great importance to banks nowadays because most of banking crises were caused by concentration risk. The recent sophisticated artificial intelligence and machine learning technologies with efficient, powerful, and automatic data analysis capability offer a new tool and possibility to increase the accuracy and implementation of credit concentration risk evaluation to the next level. We propose a credit concentration risk model based on one of machine learning algorithms called artificial neural network. To test the feasibility and effectiveness of the proposed model, we use the data of 282 portfolios from 30 financial institutions in the year 2019. Our main results suggest that it is feasible and reliable to measure concentration risk based on our proposed model, and its prediction accurate rate can exceed 98% in training sample and 92% in testing sample. The novel model broadens the areas for application of AI in risk management of banks and banking supervisors as it produces measurement possibilities to a risk which banks are also obliged to reserve capital if risk is material.