Hi Elahe,

Granger causality may be analysed in different ways. As general suggestion, you can try to run a multivariate time-series regression of each dependent variable on lags of itself and on lags of all the other dependent variables (see VAR command in Stata, for reference). Then, you can perform a set of Granger causality tests by regressing y on its own lagged values and on lagged values of x and test the null hypothesis (the estimated coefficients on the lagged values of x are jointly zero). Be aware of reverse causality possibility.

A further possibility of investigating the dynamic relationship between two or more variables is provided by the Impulse Response Function (IRF). This kind of analysis requires that the dynamic relationship between  variables have already been estimated through a Vector Auto-Regressive model, in order to assess the effect of lagged values on current values of our dependent variable. So,in case of panel data, you can try to run a Panel Vector Autoregressive Regression (PVAR),  for each  variabile of interest, including one lag of each variable in the estimated models. Once the PVAR has been performed, estimates of the error components are available. These errors terms are used to implement the IRF

Hope it can help

Hello,

Attached PDF may be useful for you.

Regards

Thanks to Witold Orlik. But in the PDF that you attached like so many other scientific papers, they just use this method. These kinds of papers can not enlighten me about steps. 

But isn't Granger Causality rather a concept than an analysis method? Predicting/explainig time series x with another time series y, where y explains variance over and above x's own past? Therefore, it seems as if there were several methods possible to accomplish this rather broad concept.

I used Granger causality for my EEG research and we used multivariate autoregressive modeling and partial directed coherence. I can send you some articles, but I don't know if this would help in your research field.

Hi Elahe,

Granger causality may be analysed in different ways. As general suggestion, you can try to run a multivariate time-series regression of each dependent variable on lags of itself and on lags of all the other dependent variables (see VAR command in Stata, for reference). Then, you can perform a set of Granger causality tests by regressing y on its own lagged values and on lagged values of x and test the null hypothesis (the estimated coefficients on the lagged values of x are jointly zero). Be aware of reverse causality possibility.

A further possibility of investigating the dynamic relationship between two or more variables is provided by the Impulse Response Function (IRF). This kind of analysis requires that the dynamic relationship between  variables have already been estimated through a Vector Auto-Regressive model, in order to assess the effect of lagged values on current values of our dependent variable. So,in case of panel data, you can try to run a Panel Vector Autoregressive Regression (PVAR),  for each  variabile of interest, including one lag of each variable in the estimated models. Once the PVAR has been performed, estimates of the error components are available. These errors terms are used to implement the IRF

Hope it can help

Dear Duesing

Granger Test is  a  statistical hypothesis test which means it create a hypothesis that is testable on the basis of observing a process that is modeled via a set of random variables. 

What I want to show is that the grow on variables X, have effect or benefits variables Y.

So if you can suggest any other methods, it can be great.

Dear Balestri

When I use Eview to perform granger test, it produces a T-test and F-test result table. 

why should these two tests perform for having a granger test? 

I want to know what is the algorithm behind Eview's Granger test.

the granger causality test computes and reports Wald tests that the coefficients on all the lags of an endogenous variable are jointly zero. the null hypothesis is that each of the endogenous variables does not Granger-cause the dependent variable in the equation

I'm not familiar with Eviews,you may have a look at this page:

http://www.eviews.com/help/helpintro.html#page/EViews%209%20Help/groups.018.17.html

It should help you

Hi Elahe,

the Granger causality is first of all a concept. According to Granger (1969) a time series variable (x) is causal for another time series variable (y) if the former helps to improve the forecast of the latter. 

This kind of interdependence is of great importance.

See the original paper!

best regards

As a great book reference I can only recommend  the 3rd chapter in:

Applied Time Series Econometrics by Lütkepohl and Krätzig (2004)

ISBN 978-0-521-54787-1

Simple & illustrative !!!

Best regards