数据抽样频率对Granger因果关系的影响

The Effects of Sampling Frequency on Granger Causality

研究目标:分析数据抽样频率对Granger因果关系的影响。研究方法:依据基础变量之间是否存在Granger因果关系,本文分三种情形研究系统抽样和时期归并对变量之间Granger因果关系的影响。进而,依据季度数据,利用靳庭良(2013)提出的ECM-DM检验程序,对我国改革开放以来货币供给量与价格水平、产出之间的Granger因果性重新进行了检验。研究发现:数据抽样频率是否影响变量之间的Granger因果性与基础变量之间的Granger因果性、模型的滞后阶数及其预测误差之间的同期相关性有关。研究创新:本文给出了Granger因果关系是否随着数据抽样频率变化的一般判定条件。研究价值:为解释依据不同抽样频率数据所得Granger因果关系检验结果存在的差异,提供了基本理论支撑。

Research Objectives： The effects of sampling frequency on Granger causality. Research Methods： On the basis of the causality between basic variables, this paper studies systematically the effects of systematic sampling and temporal aggregation on Granger causality in three eases. Furthermore, applying ECM-DM test （Jin Tingliang, 2013）, this paper retests the Granger causality among the Chinese money supply, price level and output by utilizing quarterly data. Research Findings： Whether sampling frequency effects on the Granger causality is related to the causality between basic variables, the contemporaneous correlation between their prediction errors, and the order of the basic model. Research Innovations： This paper gives the determinant conditions about whether the Granger causality changes with sampling frequency. Research Value： The results provide the theoretical support for explaining the differences in the results of Granger causality tests based on different sampling frequency data.