Hi Safa Aouinti

Thanks for your suggestion. I have looked for it, unfortunately "margins" function is not working. "Margins" in R only works for time series or cross section data. Not for Panel data. Some forums suggests that there is a function "spml.impacts". But I didn't get any function like this at least in the recent version of R and Rstudio. I have also tried using codes suggested by the following article, unfortunately it didn't work either. http://rri.wvu.edu/wp-content/uploads/2012/11/Piras_ImpactEstimatesForStaticSpatial2013-05.pdf

If I've got it right, this should work:

library(spdep)

impacts(yourmodelsplm)

Hi Elizaveta Gaivoronskaia ,

No it will not work. For spatial econometric we use Spdep package. But for spatial panel econometric model we cannot use spdep. However, we can use splm package. But impact statement in splm package is not working.

Hello Sajid Al Noor ,

have you gotten an answer to your question?

I am also estimating a dynamic spatial panel using R and am facing a similar problem..

Sorry I can't help answer your question on R,

but from what I understand, we can estimate the marginal effects for dynamic (or static) panel in Stata by using the xsmle command.

Hi Tifani Husna Siregar ,

Currently there is no package in R that can estimate impact for spatial panel model. In addition, R cannot perform your model, if your data is unbalanced panel. I don't think STATA can also do that, because I searched many sources. But some researcher has used MATLAB. So if you know MATLAB, you can do it.

Therefore, I have changed my model to SDM and SARAR (Static).

Hello Sajid Al Noor ,

thank you for your reply.

can R do it if the dataset is balanced spatial panel (dynamic)?

Thank you

Tifani Husna Siregar No R cannot estimate marginal effects for balanced spatial panel. However, you can get summary result. So if you are using Spatial error model then your summary results are your marginal effects. So you can think for spatial error model for panel data.

Sajid Al Noor, thank you for your reply.

I see. given the theoretical background, the dataset i'm using should be in a dynamic model (instead of static) and the results from the diagnostic testing which I have done also supports SDM (instead of SAR or SEM).

Anyway, thank you so much for your comments. Just fyi, maybe it can be of help to you, Stata's XSMLE command can be used for static and dynamic spatial panel data models. we can also obtain the marginal effects. However, the calculation is done by maximum likelihood estimation (instead of GMM, whereas if i'm not mistaken, R's splm package can calculate by both ML and GM), so it might be troublesome if you have many dummy variables and not so many observations nor variations in the dependent variables (which is the problem that i'm facing, some of my equations failed to converge).

Tifani Husna Siregar Thank you

Dear Sajid Al Noor , I think the link below can help you!

Link: https://rdrr.io/cran/splm/man/spml.html

Dear Gabrielito Menezes Thank you for your link. It is a nice link. However, while I performed the same codes, I got an error. Thus, I used the cross section model. May be r now updated its packages. I will try this again. Hope for the best. Thanks once again.

Sajid Al Noor I am also using the same panel data can you explain how to create the weight matrix using panel data

you can get the impact from R splm by installing the old version of spatialreg

install.packages("http://cran.r-project.org/src/contrib/Archive/spatialreg/spatialreg_1.1-3.tar.gz", repo=NULL, type="source")

https://github.com/rsbivand/ECS530_h19/blob/master/ECS530_VII.Rmd

https://rpubs.com/quarcs-lab/tutorial-spatial-regression

https://ignaciomsarmiento.github.io/2017/02/07/An-Introduction-to-Spatial-Econometrics-in-R.html

https://spatial.burkeyacademy.com/

Mustafa Kocoglu