中文摘要：

深入剖析混频数据模型（MIDAS）的内部结构,推导出MIDAS模型与传统同频率EQW模型的区别与联系,并且证明EQW模型参数估计的偏误,在此基础上,进一步结合多种不同形式的权重函数,推导出MIDAS类模型非线性最小二乘估计方法的具体机理及理论演绎过程,并且以资产价格对中国经济增长效应的样本内预测为例,对MIDAS模型的精度进行了实证检验。实证结果表明：多元EQW模型的拟合效果及样本内预测精度均低于最优滞后阶数下Exp Almon-AR-M-MIDAS模型,Exp Almon-AR-M-MIDAS模型能够提取更多高频解释变量的有效信息。

英文摘要：

In this paper,the internal structure of the Mixing Data Sampling model（MIDAS）was thoroughly analyzed,and the difference and connection between the MIDAS model and the traditional EQW model were deduced,and the error of parameter estimation of EQW model was proved.On the basis of the above analysis,five different forms of weight function were constructed,and the concrete mechanism and deduction process of the nonlinear least squares estimation method of MIDAS model were obtained.Furthermore,the intra-sample forecast of the effect of asset price on China＇s economic growth was taken as an example to test the accuracy of MIDAS model.The empirical results show that the fitting effect and the intra-sample prediction accuracy of the multivariate EQW model are lower than those of the Exp Almon-AR-M-MIDAS model under the optimal lag order;the Exp AlmonAR-M-MIDAS model can extract more high-frequency interpretations.