ABSTRACT

We employ time series data to empirically determine the causal relationship between economic policy uncertainty and the GDP growth rates of seven emerging market economies while controlling for the effect of oil price, interest rates, and the CPI. Due to differences in sampling frequencies between the GDP series and other variables, a multi-horizon mixed frequency VAR model is specified. This model fully exploits the recently developed mixed frequency Granger causality test in order to circumvent the distorting effects of temporal aggregation. The empirical results show a strong statistical evidence for causality flowing from EPU to GDP in Brazil, Chile, and India in the mixed frequency case while weak statistical evidence is found for Colombia, Mexico, and Russia. For comparative analysis, the low-frequency Granger causality test is also employed and strong statistical evidence of causality flowing from EPU to GDP in Brazil, Chile, India, Mexico is uncovered. Analyzing the causal patterns uncovered in both specifications show that the low-frequency Granger causality results are less intuitively appealing than those that are obtained from the mixed frequency Granger causality test specifications. The results have empirical as well as policy implications which are discussed.

摘要

我们使用时间序列数据，在控制油价、利率和 CPI 影响的同时，实证确定经济政策不确定性与七个新兴市场经济体 GDP 增长率之间的因果关系。由于 GDP 序列和其他变量之间的采样频率存在差异，因此指定了多水平混合频率 VAR 模型。该模型充分利用了最近开发的混合频率格兰杰因果检验，以规避时间聚集的扭曲效应。实证结果表明，在混合频率情况下，巴西、智利和印度从 EPU 到 GDP 的因果关系具有强有力的统计证据，而哥伦比亚、墨西哥和俄罗斯的统计证据则较弱。为了比较分析，还采用了低频格兰杰因果关系检验，发现了巴西、智利、印度、墨西哥从 EPU 到 GDP 的因果关系的有力统计证据。分析两个规范中发现的因果模式表明，低频格兰杰因果关系结果的直观吸引力不如从混合频率格兰杰因果关系检验规范中获得的结果。研究结果具有实证意义和政策意义，我们对此进行了讨论。