关键词: 非线性Granger因果检验, Monte Carlo模拟, 动态非线性滚动分析

Abstract: Since traditional Granger causality tests do not take account of the nonlinear character of economic variables, conclusions based on traditional tests may be biased. To overcome these limitations, nonlinear Granger causality tests are proposed in the literature and become important analysis tools in economic research. However, so far in the in the literature there is lack of systemic and comprehensive investigation on finite-sample properties of nonlinear Granger causality tests within different nonlinear models. In light of this, this paper employs Monte Carlo simulations to systemically and comprehensively examine the power and over-rejection properties, and analyze the sensitivity of test results to the changes of common lag lengths and bandwidth parameters. Based on the perspective of nonlinear dynamic analysis of rolling to do further discussion on the finite sample properties, and this paper puts forward some suggestions has practical value of application of nonlinear on future research.

Key words: Nonlinear Granger Causality Test, Monte Carl simulations, Dynamic Nonlinear Rolling Analysis Nonlinear Granger Causality Test, Monte Carl simulations, Dynamic Nonlinear Rolling Analysis