Small Sample Properties of the Granger Spurious Causality Test
Liu Tian-xiang · Tongji yu xinxi luntan · 2011
This paper studies the small sample properties of Granger spurious causality test based on Monte-Carlo simulation methods,the results show that the probability of finding Granger causal relation increases with the persistence of data process,but the probability decreases with sample size.The Granger causality test modified by the Newey-West methods has no significant advantage over the traditional Granger test owing to the specification of Granger test equation.The paper studies that the persistence of explanatory and explained variables has effects on the Granger spurious causality,and it shows that the autocorrelation or heteroskedasticity of random error term is the most important factor resulting in spurious causality through comparing with spurious regression based on OLS method,which provides a unified research framework on spurious causality and spurious regression.