基于ARMA与GARCH-M模型对我国豆粕期货价格波动的分析预测

Based on ARMA and GARCH-M Model to Analyze and Predict the Future Price Fluctuation of Soybean Meal in China

为探索豆粕期货价格与收益率的内在波动规律,构建了自回归滑动平均ARMA模型和多变量广义自回归条件异方差GARCH-M模型。研究发现:短期豆粕期价存在自回归条件异方差ARCH效应,对数波动率方差每增加1百分点豆粕期价收益率下降0.0024百分点,且最终收敛于0.00017~0.00018的无条件方差,表明豆粕期货市场以风险回避型投资者为主;中期看对数豆粕期价每增加1百分点导致10期后的对数期价下降0.433百分点,随机误.差每提升1百分点导致两期后的豆粕对数期价下降0.888百分点;最终提出建立豆粕期价预警机制、发挥套期保值功能、调整我国豆粕进口结构以及提升豆粕提取技术水平的建议。

In order to explore the internal variation rule of soybean meal futures prices and yields,build the autoregressive moving average ARMA model and multivariable generalized autoregressive conditional heteroscedasticity GARCH-M model,the study found:short-term soybean meal futures exist autoregressive conditional heteroscedasticity ARCH effect,volatility of logarithmic variance one percent increase return on soybean meal futures fell 0.0024%and eventually converges to 0.00017—0.00018 unconditional variance,indicating that soybean meal futures market risk avoider investors primarily.In the medium term,the logarithmic soybean meal price decreased by 0.433 percentage points if the logarithmic soybean meal price increased by one percentage point after 10 stages,and the logarithmic soybean meal price decreased by 0.888 percentage points if the random error increased by one percentage point after two stages.Finally,some suggestions were put forward to establish soybean meal price early warning mechanism,give play to the hedging function,adjust the import structure of soybean meal and improve the extraction technology level of soybean meal.