All impulse response functions in one graph

[aloprofundo]

Hello everyone, I have a question about how to put all impulse response functions in "just one graph". Since the impulse response functions are not series, I cannot simply save them as group and show them at the same time in a graph. Please note that in my case, I have rescaled the end points of horizontal axis (y), and the periods of response are set to be equal. I will appreciate any kind of comment, thanks in advance

[EViews Gareth]

If the "Combined Graphs" option doesn't give you what you want, all you can do is output the tables, then copy the table data into a matrix or group object and make the graph manually.

[aloprofundo]

Hello Gareth, first of all, thanks for answering. Sorry to bother again, but could you please tell me where to find the "Combined Graphs" option? If this is the same as selecting two graphs (highlight them) and opening them at the same time, then it simply shows two separated graphs in a larger graph, what I need is to have the graphs overlapping each other. I am afraid that the second method does not work in my case, because impulse response cannot be "outputted". For regular series, I will be able to get the data after clicking "proc" and "display output data", but it seems to me that the latter option (display output data) does not appear for the impulse response graph. Please let me know if there are other ways to do it, maybe via the help of some codes? Thanks again!

[EViews Gareth]

On the impulse response menu.

[aloprofundo]

Hey Gareth, thanks again for your help, but your solution has partly solved my problem. Imagine I hava run a VAR model of two series, then I want to look at the impulse response of the variables. If I had chosen the "combined function", I would end up getting two graphs, each containing the impulse response of the two variables due to a shock in one of them. I think this is what you suggested me to do, however, I would like to have ALL the impulse responses (in this case 4 impulse responses) in one graph. Doing so sounds a bit strange, but it does help interpret my result. Please let me know if this could be done!!

[EViews Gareth]

If the "Combined Graphs" option doesn't give you what you want, all you can do is output the tables, then copy the table data into a matrix or group object and make the graph manually.

[aloprofundo]

EViews Gareth wrote:If the "Combined Graphs" option doesn't give you what you want, all you can do is output the tables, then copy the table data into a matrix or group object and make the graph manually.

I do not know how to "output the table" and am not sure if this could possible be done for impulse response function (I cannot find the options), could you please be more specific? by the way, I am using eviews 7.2

[EViews Gareth]

On the impulse response dialog select "Table".

[aloprofundo]

Thanks a million!!!

[wardy94]

Hi Gareth,

I wondered if you knew how to determine whether the impulse responses from a unrestricted VAR are statisically significant in E views? I have produced my table with the impulse responses over 24 periods with the reported standard errors. I'm assuming that you can't simply divide the IRF coefficients by the standard errors, like you would do in standard regression analysis, to get the t-stats of the coefficient estimates. Any help would be greatly appreciated!

Best,

James