中文摘要：

Granger因果检验是计量经济学的重要组成部分，也是现代经济、金融学分析的重要工具，近年来Granger因果检验在非线性检验方向有了较大进展。本文在线性Granger因果检验的基础上，阐述了Granger因果检验的非线性进展，重点总结了针对一阶矩的基于回归模型、非参函数和信息理论的三大类非线性方法以及针对二阶矩的基于残差交叉相关系数和多元条件方差模型下的两大类非线性方法，讨论了不同非线性Granger方法中数据要求、核心模型、建模关键以及模型优缺点，提出了Granger因果检验“线性一非线性”的整体框架和研究范式。通过模型分析和比较，本文可为因果检验的非线性理论和模型研究提供参考，并对因果检验在经济和金融领域的更广泛应用提供支持．

英文摘要：

Granger causality test is a very important part of econometrics and it＇s also an important instrument of modern economic and finance analysis. Granger causality test has great development in nonlinear fields in recent years. Based on linear Granger test, this paper elaborates the developments of nonlinear Granger causality; summarizes three nonlinear methods for the first moment which are based on regression model, non-parametric function model and information theory and two nonlinear methods for the second moment which are based on residual cross-correlation function and multivariate conditional variance model; discusses data requirement, core model, key of modeling and advantages and disadvan- tages of different nonlinear methods; proposes ＂linear-nonlinear＂ whole framework and research paradigm of Granger causality. Through analyzing and comparing the models, this paper could provide reference for nonlinear theory and models of Granger causality and provide support for extensive application of Granger causality on the economic and finance field.