Empirical Research On The Influencing Factors Of Soybean Futures Price Volatility Based On GARCH-MIDAS-Skewed T Model

In recent years,China has experienced a very high market demand for soybeans and an increased dependence on foreign soybean imports,which have triggered dramatic fluctuations in soybean futures prices.Risk management in the soybean futures market has become more important than ever for stakeholders to reduce losses.Since the behavioral characteristics of long and short trades in the futures market are very different,leading to asymmetry in the distribution of futures returns and the presence of typical fat-tail characteristics,Therefore,determining the main determinants of soybean futures price volatility has become a key issue for decision makers to formulate measures and avoid market risks.This paper theoretically analyzes the influencing factors of soybean futures price volatility,specifically,the influence mechanism of 18 variables on soybean futures price volatility in terms of commodity market environment,supply and demand,and macroeconomic factors.On this basis,the GARCH-MIDAS model and GARCHMIDAS-Skewed T model are applied to analyze the relationship between each factor and soybean futures price volatility in terms of both the level effect and the volatility effect,respectively analyze the relationship between each factor and soybean futures price volatility.This paper constructs a multi-factor GARCH-MIDAS-Skewed T model to identify the main influencing factors of soybean futures price volatility and observe the joint effect of multiple factor combinations with different frequencies on soybean futures price volatility.This paper applies Va R backtesting method to test the Va R forecasting effect of the multi-factor GARCH-MIDAS-Skewed T model.The empirical results show that the single-factor GARCH-MIDAS-Skewed T model fits better than the single-factor GARCH-MIDAS model.Firstly,the estimation results of the single-factor GARCH-MIDAS-Skewed T model show that Brent crude oil price(level),China soybean futures position(level),China soybean output volume(level),China consumer price index(volatility),China soybean futures trading volume(volatility),and freight rate(level)are significantly and negatively related to soybean futures price volatility,and the increase in the corresponding values of these variables would slow down soybean futures price volatility.In contrast,China corn futures price(volatility),China soybean oil futures price(level),China soybean meal futures price(level),exchange rate(level),international soybean futures price(level),China soybean spot price(level),China money supply(volatility),China soybean import volume(level),and China soybean consumption(level)are significantly and positively correlated with soybean futures price volatility,and the corresponding values of these variables will lead to increased volatility in soybean futures price.Besides,China soybean crush(level),China agricultural production price index(level),and China soybean stock(level)has no significant effect on soybean futures price volatility.Among them,the variance ratio based on the Brent crude oil price level effect model is as high as 50.29%,which has the greatest impact on soybean futures price volatility among all variables.Secondly,the constructed multi-factor GARCH-MIDAS-Skewed T model includes 10 variables: Brent crude oil price(level),China corn futures price(volatility),China soybean oil futures price(level),China soybean meal futures price(level),China consumer price index(volatility),China money supply(volatility),China soybean import volume(level),China soybean futures trading volume(volatility),China soybean consumption(level)and freight rate(level).With a variance ratio of80.79% for the multi-factor GARCH-MIDAS-Skewed T model,these variables overall exert a stronger influence on soybean futures price volatility and are important factors influencing soybean futures price volatility.Finally,the Va R backtesting method is applied to test the Va R forecasting ability of GARCH(1,1),GARCH-MIDAS model,and GARCH-MIDAS-Skewed T model.By calculating the mean PI,MAVa R,LR and DQ test statistics of the three models,the comparison reveals that the GARCH-MIDASSkewed T model has better Va R prediction.Based on the results of the empirical study,the following policy recommendations are proposed for the current situation of China soybean futures market: formulate reasonable policies on soybean farmers' subsidies and soybean imports,strengthen the management of the soybean spot market,and improve the soybean futures market regulatory system and information disclosure system.The paper consists of 6 figures,14 tables and 104 references.