Financial Innovation · 2024 · 16 citations · 54 references
BusinessEconomicsFinanceForecastingMachine LearningMacroeconomicsEconomic ForecastingFinancial CrisesFinancial ForecastingFinancial RiskRisk ForecastingRisk IdentificationCrisis Probability ModelsMachine Learning Forecasting ModelsMacroeconomic CrisisMacroeconomic ImpactsPrediction ModelsSystemic Crisis ManagementSystemic Risk ForecastingTransparency
Abstract This study proposed a cutting-edge, multistep workflow and upgraded it by addressing its flaw of not considering how to determine the index system objectively. It then used the updated workflow to identify the probability of China’s systemic financial crisis and analyzed the impact of macroeconomic indicators on the crisis. The final workflow comprises four steps: selecting rational indicators, modeling using supervised learning, decomposing the model’s internal function, and conducting the non-linear, non-parametric statistical inference, with advantages of objective index selection, accurate prediction, and high model transparency. In addition, since China’s international influence is progressively increasing, and the report of the 19th National Congress of the Communist Party of China has demonstrated that China is facing severe risk control challenges and stressed that the government should ensure that no systemic risks would emerge, this study selected China’s systemic financial crisis as an example. Specifically, one global trade factor and 11 country-level macroeconomic indicators were selected to conduct the machine learning models. The prediction models captured six risk-rising periods in China’s financial system from 1990 to 2020, which is consistent with reality. The interpretation techniques show the non-linearities of risk drivers, expressed as threshold and interval effects. Furthermore, Shapley regression validates the alignment of the indicators. The final workflow is suitable for categorical and regression analyses in several areas. These methods can also be used independently or in combination, depending on the research requirements. Researchers can switch to other suitable shallow machine learning models or deep neural networks for modeling. The results regarding crises could provide specific references for bank regulators and policymakers to develop critical measures to maintain macroeconomic and financial stability.
54
Greedy function approximation: A gradient boosting machine.
Jerome H. Friedman · The Annals of Statistics · 2001 · 29K citations · Full text
Leo Breiman · Machine Learning · 1996 · 17K citations · Full text
Marco Túlio Ribeiro, Sameer Singh, Carlos Guestrin · 2016 · 15.2K citations
Gene Selection for Cancer Classification using Support Vector Machines
Isabelle Guyon, Jason Weston, S. Barnhill et al. · Machine Learning · 2002 · 9.8K citations · Full text
From local explanations to global understanding with explainable AI for trees
Scott Lundberg, Gabriel Erion, Hugh Chen et al. · Nature Machine Intelligence · 2020 · 8.9K citations · Full text