Value-at-Risk with quantile regression neural network: New evidence from internet finance firms
Document Type
Article
Publication Date
11-1-2023
Abstract
Traditional risk measurements have proven inadequate in capturing tail risk and nonlinear correlation. This study proposes a novel approach to measure financial risk in the Internet finance industry: a new Value-at-Risk (VaR) measurement based on quantile regression neural network (QRNN). Sparrow Search Algorithm (SSA) is utilized to optimize the QRNN model, which improves the model's performance in predicting internet finance risk. By comparing the TGARCH-VaR and QR-VaR approaches, our study demonstrates the effectiveness of the QRNN-VaR approach and its potential to improve the accuracy of risk prediction in the Internet finance industry. This study further examines and compares the risks between the traditional and internet finance industries. It also considers the unique impact of COVID-19 on industry risk based on statistical testing for differences and machine learning models. Our results indicate that the level of risk in the Internet finance industry is higher than in the traditional finance industry. Moreover, COVID-19 has contributed to increased risk within the Internet finance industry. These findings have significant implications for investors and policymakers seeking to better understand and manage risks within the Internet finance industry, particularly in the ongoing COVID-19 pandemic.
Keywords
financial risk, internet finance, quantile regression neural network, value-at-risk
Divisions
Faculty_of_Business_and_Accountancy
Publication Title
Applied Stochastic Models in Business and Industry
Volume
39
Issue
6
Publisher
Wiley
Publisher Location
111 RIVER ST, HOBOKEN 07030-5774, NJ USA