基于交互作用MIDAS-AR模型的通货膨胀预测效应研究  

Research on Inflation Expectation Effect Based on MIDAS-AR Model of Interaction

作　　者：汪玲 王沁[1] 张红梅 何婷 WANG Ling;WANG Qin;ZHANG Hongmei;HE Ting(School of Mathematics, Southwest JiaoTong University, Chengdu 611756, China)

机构地区：[1]西南交通大学数学学院,成都611756

出　　处：《重庆理工大学学报（自然科学）》2021年第9期239-247,共9页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(71671145);教育部人文社会科学研究项目(17YJC790119)。

摘　　要：基于多元自回归混频数据抽样模型,研究了我国货币供应量、国房景气指数及其交互作用对通货膨胀的传导机制及未来走势。结果表明:通货膨胀具有强烈惯性特征。货币、房价及其交互作用均对通货膨胀有显著影响,但影响存在模式、时滞和力度差异。货币对通货膨胀有短期乘数效应,房价和交互作用对通货膨胀的冲击则更持久,同时通货膨胀受近期房价变动以及货币与房价间交互作用的影响强烈;就变量的贡献程度而言,交互作用在三者间对通货膨胀的贡献程度最大,其次是货币因素,最后是房价因素。相比传统无交互作用的混频模型,交互作用能显著提高模型的拟合效果、短期预测精度和预测拐点捕捉。Based on the multivariate autoregressive mixed-frequency data sampling model,this paper empirically studies the transmission mechanism and future trends of our country’s money supply,national housing climate index and their interaction on inflation.The results show that:inflation has strong inertia characteristics.Currency,housing prices and their interactions all have a significant impact on inflation,but there are differences in the mode,time lag,and intensity of the impact.Currency has a short-term multiplier effect on inflation,and the impact of housing prices and interaction on inflation is more lasting.At the same time,inflation is strongly affected by recent housing price changes and the interaction between currency and housing prices.In terms of the degree of contribution of variables,the interaction among the three contributes the most to inflation,followed by monetary factors,and finally housing prices.Compared with the traditional mixing model without interaction,interaction can significantly improve the model’s fitting effect,short-term prediction accuracy and prediction inflection point capture.

关 键 词：CPI M2 房价 交互作用 MIDAS-AR 

分 类 号：F820.5[经济管理—财政学]