Influence Analysis And Early Warning Of Digital Financial Risk In China

With the continuous innovation of China’s digital financial industry,mobile payment,big technology credit,online investment and China’s digital currency have gradually become an important part of people’s lifestyle.Digital finance,as a product of the deep combination of finance and technology,provides technological means to manage and prevent financial risk,further promoting the solid development of China’s financial industry.This paper provides data support for risk prevention of digital finance from the perspectives of risk measurement,influencing factors and risk warning,which is of great significance for the healthy development of digital finance.First of all,this paper constructs an indicator system of digital financial risk for23 provinces and cities in China from five aspects,including operational risk,credit risk,market risk,liquidity risk and policy risk,and uses the Lagrange multiplier method to obtain the optimal comprehensive weight of AHP and entropy weight.On this basis,the digital financial risk indexes of the above regions in China from 2013 to2021 are measured and analyzed.Afterward,the areal heterogeneity of digital financial risk is measured by using the Theil index.Secondly,based on the skew-normal and normal panel data models,we explore the influencing factors of digital financial risk in the eastern,central and western areas of China,and then establish the early warning variables.Furthermore,seven machine learning models such as BPNN,XGBoost and SVM are used to construct a risk early warning model for the eastern,central and western areas.Last but not least,the importance of early warning variables is identified based on the optimal early warning model of eastern,central and western areas,and the impact of important variables on the prediction effect is analyzed.This paper draws the following conclusions through the exploration of digital financial risk measurement,influencing factors and risk warning in China.Firstly,the concentration distribution of digital financial risk in China has a trend of transfer from west to east,and the difference mainly comes from within the area and is concentrated in the east.Secondly,the outbreak of COVID-19 temporarily delays the expansion of digital financial risk in the eastern,central and western areas.Meanwhile,the expansion of population size intensifies the digital financial risk in the eastern and central areas,but slows down in the western area.Moreover,the rising inflation level increases digital financial risk in the eastern and western areas,but moderates that in the central area.Finally,for the early warning of digital financial risk,the random forest model is the optimal model in the eastern area,while the XGBoost model is the optimal model in the central and western areas.For the eastern,central and western areas,the scale of government fiscal expenditure,population size,and national economic development level variables are the most important early warning variables respectively.What’s more,the addition of COVID-19 can improve the risk early warning ability of the optimal model in each area.