基于混频数据模型的税收收入预测研究——以河南省为例  

Study on Tax Revenue Forecast Based on MIDAS Model:The Case of Henan Province

作　　者：邵明振[1] 赵蕊 谷艳艳 屈小芳 SHAO Ming-zhen;ZHAO Rui;GU Yan-yan;QU Xiao-fang(School of Economics,Henan University,Kaifeng 475004,China)

机构地区：[1]河南大学经济学院,河南开封475004

出　　处：《数学的实践与认识》2023年第7期102-109,共8页Mathematics in Practice and Theory

基　　金：国家税务总局河南省税务局项目:税收经济指标模型建设(SKH2020172);河南大学教育教学改革研究与实践项目:基于学科竞赛的创新型人才培养模式研究(HDXJJG2022-107)。

摘　　要：税收收入在税收工作中具有重要意义.利用河南省税收收入数据,首先将采购经理人指数(PMI)等反映经济景气度和经济活动的经济指标引入到河南省税收收入的混频数据预测模型中.然后选择预测效果最好的PMI指数作为高频解释变量,构造指数Almon多项式权重的MIDAS(3,3,2)-AR(4)混频数据模型对河南省税收收入进行预测.最后与经典的时间序列预测方法对比,构造的税收收入混频数据模型具有明显的预测优势.Tax revenue forecasting is of great significance in taxation work.Using the tax revenue data of Henan Province,firstly,the purchasing manager index(PMI)and other economic indicators reflecting economic prosperity and economic activities are introduced into the mixed data prediction model of tax revenue of Henan Province.Then the PMI index with the best prediction result is selected as the high-frequency explanatory variable,and the MIDAS(3,3,2)-AR(4)mixed frequency data model with the index Almon polynomial weight is used to predict the tax revenue of Henan Province.Finally,compared with the classic time series forecasting methods,the tax revenue mixed data model has obvious forecasting advantages.

关 键 词：税收收入预测 混频数据 采购经理人指数 多项式权重 

分 类 号：F812.42[经济管理—财政学]